Application of a Neural Network for the Optimization of Tire Design

1999 ◽  
Vol 27 (2) ◽  
pp. 62-83 ◽  
Author(s):  
Y. Nakajima ◽  
H. Kadowaki ◽  
T. Kamegawa ◽  
K. Ueno
Keyword(s):  
Neural Network ◽  
Mathematical Programming ◽  
Design Of Experiments ◽  
Design Space ◽  
Global Optimum ◽  
Large Computer ◽  
Multiple Peak ◽  
Design Variables ◽  
The Neural Network ◽  
Computer Resources

Abstract When there are multiple peaks in the design space, it is difficult to obtain the global optimum by mathematical programming. The procedure to obtain the pseudo-global optimum in the multiple-peak problem is to select the best solution from many local optimums starting from different initial values. However, this approach requires large computer resources and has difficulty in obtaining the pseudo-optimum in a problem with many peaks, due to the complicated design space. We propose more efficient and robust optimization procedures for the multiple-peak problem. In this approach, a neural network is applied for the approximation of the design space by learning the data when design variables are systematically changed, by employing a design of experiments. Since the neural network can approximate the nonlinear space, it can be applied for the approximation of the complicated design space. Moreover, since the number of analyses is determined by the design of experiments, it is much smaller than that of the mathematical programming. Hence the CPU time for optimization can be decreased by the proposed method. The proposed method is applied to the optimization of tire contact performance and is verified to be effective to improve the handling and wear of a tire.

Neural Networks for Tubular Joint Optimization in Offshore Jacket Structures

2016 ◽  
Author(s):  
Niels Hørbye Christiansen ◽  
Benny Korsholm Tang
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Design Space ◽  
Joint Optimization ◽  
Joint Models ◽  
Tubular Joints ◽  
Design Variables ◽  
The Neural Network ◽  
Jacket Structures ◽  
Tubular Joint

The use of jacket structures to support offshore installations has for a long time been a popular choice in places with appropriate water depths. In recent years the use of jacket structures as offshore wind turbine foundations has also attracted increasing attention and is becoming a feasible alternative to traditional monopile foundations. One of the key challenges in jacket design is optimizing tubular joints in terms of fatigue resistance. As it is not practically possible to include detailed FEM joint models in global jacket models designers are forced to look for alternative methods to obtain realistic joint representations. This is done by calculating influence factors (INF) and stress concentration factors (SCF) to be applied to simplified models of relevant tubular joints in global models in order to achieve a realistic force flow in the structure. One simple and widely used method is to apply parametric formulas, e.g. those suggested by Efthymiou. However, these approximating formulas have a fairly limited validity range. Therefore, on complex joint the most reliable way to determine INF’s is by setting up refined FEM models of relevant joints and then subsequently using the calculated factors in the global model. This strategy is computationally demanding and hence, very time consuming, as a new detailed FEM analysis of the tubular joint must be conducted for each step in the optimization process. The present paper demonstrates how this time consuming procedure can be avoided by use of artificial neural networks (ANN) trained to estimate INF’s on tubular joints. The neural network is trained on a pre-generated library of detailed FEM joint models and is then able to predict INF’s on joints that are not part of the library — and thereby providing a significant reduction in calculation time during the jacket/joint optimization process. The analysis is conducted on a typical joint on a three legged jacket structure. The joint is located on a jacket leg and has two incoming braces. Such a joint has a finite number of free design variables, e.g. chord diameter/thickness, brace diameter/thickness, brace angle, gap etc. Each of these free variables can be considered as a dimension in the joint design space. Having a sufficient number of FEM joint models in the library the neural network can be trained to recognize and predict underlying patterns in this design space. The method is demonstrated on a limited number of design variables but should easily be extended to cover all variables as the joint library is expanded to include all dimensions.

scholarly journals Optimization of machining parameters during Drilling by Taguchi based Design of Experiments and Validation by Neural Network

2018 ◽  
Vol 15 (2) ◽  
pp. 294-301
Author(s):  
Reddy Sreenivasulu ◽  
Chalamalasetti SrinivasaRao
Keyword(s):  
Neural Network ◽  
Design Of Experiments ◽  
Computational Cost ◽  
Machining Parameters ◽  
Burr Size ◽  
Taguchi Design Of Experiments ◽  
The Neural Network ◽  
Input Parameters ◽  
Hole Making ◽  
The Cost

Drilling is a hole making process on machine components at the time of assembly work, which are identify everywhere. In precise applications, quality and accuracy play a wide role. Nowadays’ industries suffer due to the cost incurred during deburring, especially in precise assemblies such as aerospace/aircraft body structures, marine works and automobile industries. Burrs produced during drilling causes dimensional errors, jamming of parts and misalignment. Therefore, deburring operation after drilling is often required. Now, reducing burr size is a serious topic. In this study experiments are conducted by choosing various input parameters selected from previous researchers. The effect of alteration of drill geometry on thrust force and burr size of drilled hole was investigated by the Taguchi design of experiments and found an optimum combination of the most significant input parameters from ANOVA to get optimum reduction in terms of burr size by design expert software. Drill thrust influences more on burr size. The clearance angle of the drill bit causes variation in thrust. The burr height is observed in this study.  These output results are compared with the neural network software @easy NN plus. Finally, it is concluded that by increasing the number of nodes the computational cost increases and the error in nueral network decreases. Good agreement was shown between the predictive model results and the experimental responses.  

scholarly journals Design Space Exploration for YOLO Neural Network Accelerator

Electronics ◽  
2020 ◽  
Vol 9 (11) ◽  
pp. 1921
Author(s):  
Hongmin Huang ◽  
Zihao Liu ◽  
Taosheng Chen ◽  
Xianghong Hu ◽  
Qiming Zhang ◽  
...  
Keyword(s):  
Neural Network ◽  
Design Space Exploration ◽  
Design Space ◽  
Space Exploration ◽  
Computation Time ◽  
Memory Bandwidth ◽  
Block Transmission ◽  
Transmission Strategy ◽  
The Neural Network ◽  
The Given

The You Only Look Once (YOLO) neural network has great advantages and extensive applications in computer vision. The convolutional layers are the most important part of the neural network and take up most of the computation time. Improving the efficiency of the convolution operations can greatly increase the speed of the neural network. Field programmable gate arrays (FPGAs) have been widely used in accelerators for convolutional neural networks (CNNs) thanks to their configurability and parallel computing. This paper proposes a design space exploration for the YOLO neural network based on FPGA. A data block transmission strategy is proposed and a multiply and accumulate (MAC) design, which consists of two 14 × 14 processing element (PE) matrices, is designed. The PE matrices are configurable for different CNNs according to the given required functions. In order to take full advantage of the limited logical resources and the memory bandwidth on the given FPGA device and to simultaneously achieve the best performance, an improved roofline model is used to evaluate the hardware design to balance the computing throughput and the memory bandwidth requirement. The accelerator achieves 41.99 giga operations per second (GOPS) and consumes 7.50 W running at the frequency of 100 MHz on the Xilinx ZC706 board.

scholarly journals Prediction of engine NOx for virtual sensor using deep neural network and genetic algorithm

2021 ◽  
Vol 76 ◽  
pp. 72
Author(s):  
Jongmyung Kim ◽  
Jihwan Park ◽  
Seunghyup Shin ◽  
Yongjoo Lee ◽  
Kyoungdoug Min ◽  
...  
Keyword(s):  
Neural Network ◽  
Genetic Algorithm ◽  
Deep Neural Network ◽  
Global Optimum ◽  
Operating Conditions ◽  
Nox Emissions ◽  
Hyperparameter Optimization ◽  
Sensor Model ◽  
Design Variables ◽  
Nox Sensor

The Nitrogen Oxides (NOx) from engines aggravate natural environment and human health. Institutional regulations have attempted to protect the human body from them, while car manufacturers have tried to make NOx free vehicles. The formation of NOx emissions is highly dependent on the engine operating conditions and being able to predict NOx emissions would significantly help in enabling their reduction. This study investigates advanced method of predicting vehicle NOx emissions in pursuit of the sensorless engine. Sensors inside the engine are required to measure the operating condition. However, they can be removed or reduced if the sensing object such as the engine NOx emissions can be accurately predicted with a virtual model. This would result in cost reductions and overcome the sensor durability problem. To achieve such a goal, researchers have studied numerical analysis for the relationship between emissions and engine operating conditions. Also, a Deep Neural Network (DNN) is applied recently as a solution. However, the prediction accuracies were often not satisfactory where hyperparameter optimization was either overlooked or conducted manually. Therefore, this study proposes a virtual NOx sensor model based on the hyperparameter optimization. A Genetic Algorithm (GA) was adopted to establish a global optimum with DNN. Epoch size and learning rate are employed as the design variables, and R-squared based user defined function is adopted as the object function of GA. As a result, a more accurate and reliable virtual NOx sensor with the possibility of a sensorless engine could be developed and verified.

An Optimization Procedure in a Sirocco Fan Using Experimental Data

1997 ◽  
Author(s):  
H. S. Min ◽  
H. S. Rew ◽  
C. O. Ahn
Keyword(s):  
Neural Network ◽  
Taguchi Method ◽  
Flow Rate ◽  
Noise Level ◽  
Optimization Procedure ◽  
Design Variables ◽  
The Neural Network ◽  
Sirocco Fan ◽  
Important Design ◽  
Good Agreement

Abstract The flow rate and the specific noise level of 18 sirocco fans were measured and these data were analyzed by the Taguchi method and the neural network. The optimal design value obtained by the neural network generally showed good agreement with that by the Taguchi method. The effects of eight important design variables on the flow rate and the specific noise level were discussed.

Numerical Control Interpolation Algorithm of Aspheric Surface Based on the Genetic Algorithms and Neural Network

2007 ◽  
Vol 364-366 ◽  
pp. 25-29
Author(s):  
Fei Hu Zhang ◽  
D.J. Chen ◽  
L.J. Li
Keyword(s):  
Neural Network ◽  
Genetic Algorithms ◽  
Global Optimum ◽  
Numerical Control ◽  
Local Optimum ◽  
Interpolation Algorithm ◽  
Aspheric Surfaces ◽  
The Neural Network ◽  
Circular Curves ◽  
Circular Curve

When the Neural Network model is used to interpolate the non-circular curves, there are shortcomings of converging slowly and getting into the local optimum easily. A novel numerical control interpolation algorithm based on the GA (Genetic Algorithms) and NN (Neural Network) was introduced for the ultra-precision machining of aspheric surfaces. The algorithm integrated the global searching of GA with the parallel processing of NN, enhanceed the convergence speed and found the global optimum. At the end, the quintic non-circular curve was taken as an example to do the emulation and experiment. The results prove that this algorithm can fit the non-circular curve accurately, improve the precision of numerical control interpolation and reduce the number of calculating and interpolation cycles.

The Most Probable Optimal Design Method for Global Optimization

1999 ◽  
Author(s):  
Qinzhong Shi ◽  
Ichiro Hagiwara ◽  
Futoshi Takashima
Keyword(s):  
Neural Network ◽  
Optimal Design ◽  
Design Method ◽  
Global Optimum ◽  
Local Optimum ◽  
Space Design ◽  
Design Variables ◽  
Discrete Design Variables ◽  
Global Optimal ◽  
Optimal Design Method

Abstract In this study, to find the global optimum efficiently, holographic neural network is introduced to be an activate function of response surface methodology. Since the accuracy of approximation function near the global optimal design is merely important, techniques to search the region of containing the global optimal design using conditional random seeds, and techniques for finding more, accurate approximation near the global optimal design, using holographic neural network are exploited. In the study, the proposed approach called the most probable optimal design (MPOD) method to pick up one local optimum design which has the biggest probability in the design space. Design example of crash worthiness for the passenger injury with continuous and discrete design variables are shown the validity of the method.

Optimization for Motorcycle Tire Using Explicit FEM

2001 ◽  
Vol 29 (4) ◽  
pp. 230-243 ◽  
Author(s):  
M. Koide ◽  
H. Heguri ◽  
T. Kamegawa ◽  
Y. Nakajima ◽  
H. Ogawa
Keyword(s):  
Neural Network ◽  
Finite Element ◽  
Design Space ◽  
Objective Functions ◽  
Explicit Finite Element ◽  
Handling Performance ◽  
The Neural Network ◽  
Practical Development ◽  
Contour Design ◽  
Good Agreement

Abstract A new procedure of the crown contour design for the motorcycle tire is proposed in this paper. The explicit finite element method (FEM) combined with the neural network has been utilized to optimize the cornering property of the motorcycle tire. For the motorcycle tire, the camber thrust is one of the most important cornering characteristics. The explicit FEM has been utilized for the camber thrust prediction to avoid poor numerical stability that will be caused by the implicit FEM. The prediction of the camber thrust that the results of finite element analyses (FEA) were in good agreement with the experimental results has been verified. For the approximation of the design space in the optimization, the neural network has been utilized to circumvent the multipeak and huge CPU time problems. The objective functions of the optimization were both the linearity of camber thrust and the uniformity of pressure distribution in the contact area. The design variable was the crown contour expressed by three variables, and the number of variables was defined in consideration of decreasing the number of FEA. The procedure has been applied to the practical development of a motorcycle tire and verified to be an effective method to improve the handling performance at the proving ground.

Neural Network for Automatic Analysis of Motility Data

1994 ◽  
Vol 33 (01) ◽  
pp. 157-160 ◽  
Author(s):  
S. Kruse-Andersen ◽  
J. Kolberg ◽  
E. Jakobsen
Keyword(s):  
Neural Network ◽  
Muscular Activity ◽  
Recording System ◽  
Automatic Analysis ◽  
Biologically Relevant ◽  
System A ◽  
The Neural Network ◽  
Valuable Method ◽  
Motor Abnormalities ◽  
Trained Network

Abstract:Continuous recording of intraluminal pressures for extended periods of time is currently regarded as a valuable method for detection of esophageal motor abnormalities. A subsequent automatic analysis of the resulting motility data relies on strict mathematical criteria for recognition of pressure events. Due to great variation in events, this method often fails to detect biologically relevant pressure variations. We have tried to develop a new concept for recognition of pressure events based on a neural network. Pressures were recorded for over 23 hours in 29 normal volunteers by means of a portable data recording system. A number of pressure events and non-events were selected from 9 recordings and used for training the network. The performance of the trained network was then verified on recordings from the remaining 20 volunteers. The accuracy and sensitivity of the two systems were comparable. However, the neural network recognized pressure peaks clearly generated by muscular activity that had escaped detection by the conventional program. In conclusion, we believe that neu-rocomputing has potential advantages for automatic analysis of gastrointestinal motility data.

Sign in / Sign up

Export Citation Format

Share Document