sidaratabey@gmail.tr
2
Gebze Technical University, Kocaeli, Turkey
mustafa.ozcan@gtu.edu.tr
3
Gebze Technical University, Kocaeli, Turkey
f.aliew @gtu.edu.tr
Abstract
. Hall effect sensor is used in automobile industry especially for angle estimation of rotary component. One of the
main challenge in estimating the angle with the Hall effect sensor is temperature. This paper aims to increase the accuracy of the
measurements made with the hall-effect sensor using the artificial neural network algorithm. Hall effect position sensors are used in
automotive on the accelerator pedal to adjust the amount of throttle flowing through the electronic throttle. Therefore, accurate position
measurement with the hall effect sensor has a direct effect on fuel efficiency. Low carbon dioxide emission of spark ignition engine
with high fuel efficiency. Observations at different temperatures are collected with the Hall-effect sensor test setup to eliminate errors
caused by temperature. The observed sensor data was then used to train the artificial neural network. In the test phase, using the data
obtained from the sensor output, the angle is estimated by using both ANN algorithm and classical methods without ANN algorithm.
As a result, it is concluded that the values obtained with the ANN algorithm are more accurate and the temperature-related errors of
the sensor are reduced.
Keywords:
Hall Effect, Position Sensor, accurate position estimation, Artificial Neural Network, Temperature Compensation,
Levenberg-Marquadt
Introduction
Hall effect sensor are the sensors widely used in the industry, especially in the automobile industry[1, 2]. Most
of the applications used for hall effect sensors are angle estimation, voltage regulators, vibration sensors and current
measuring[3-5]. Nowadays, with the development of semiconductor technology, Hall effect sensor is becoming more and
more important especially in automotive industry.
One of the uses of the hall effect sensor in the automobile industry is to estimate the position of the accelerator
pedal. Fuel savings and low emission rates in spark ignition engines are one of the most important issues expected from
vehicle developers. Therefore, for spark ignition engines, precise and accurate positioning of the accelerator pedal, which
controls the speed of vehicles, is one of the factors affecting fuel consumption and low emissions. In conventional
vehicles, the accelerator pedal provides engine control by adjusting the air entering the vehicle by simply moving the gas
throttle. However, today, the control system of modern vehicles adjusts the appropriate engine speed, speed and fuel
amount, taking into account many variables according to the position and movement of the accelerator pedal. Accuracy
and precise gas pedal position estimation ensure stable inputs to the control system. Thus, the control system makes the
right decisions and ensures that it is driven with minimum fuel consumption and emissions. However, the output of the
hall effect sensor is non-linear and affected by temperature. therefore, obtaining stable output from the hall effect sensor
is a challenge. Almost all sensors contain noise, outliers, and are non-linear[6-9]. There are 3 different solutions for such
problems in the literature. Software based solutions, software-hardware based solutions and hardware-based solutions
provide sensor accuracy and precise position estimation. Hardware based accurate and precise position estimation solution
brings extra cost in sensor design[10]. software-based solutions offer more cost-effective solutions to accurate estimation.
Therefore, in many studies with various sensors, software-based algorithms are preferred to obtain high accuracy
predictions using the sensor output [7].
In this paper, an artificial neural network-based software algorithm is proposed to increase the accuracy and
linearity of the hall effect position sensor. In addition, errors due to temperature change have been reduced by using the
ANN algorithm. Results have been verified in both simulation and real-time tests. real-time tests were conducted using a
real-time test setup. Rest of this paper is organized as fallows,
Conclusion
This paper demonstrates a method of temperature compensation based on Artificial neural network to improve
the accuracy and linearity of the Hall effect position measurement sensor. Due to the high cost investment for the hardware
compensation method, this application uses the software method. ANN which is a software method provide cost-effective
solutions to the error caused by temperature. In addition, angle estimation accuracy is increased by using ANN. Also, the
ANN method can be used for accurate and precise measurements for various sensors. This paper represents just one
example of what can be done with ANN. Researchers can use ANN for a variety of different applications.
Do'stlaringiz bilan baham: 
