CANINE HEARTRATE MEASUREMENT - Sensor technologies and data analysis
Koistinen, Jouni Pekka (2021)
Lataukset:
Koistinen, Jouni Pekka
2021
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-202104084385
https://urn.fi/URN:NBN:fi:amk-202104084385
Tiivistelmä
Animal welfare and sports market is fast-growing market. However, existing canine heart rate measurement solutions on the market do not deliver on their promises; they are difficult to use, and results are unreliable. The current technologies are not found suitable for canine sports. Therefore, Iotas Oy was interested in understanding whether existing sensor technologies are potentially applicable into canine heart rate measurements, to build reliable solution. The other objectives were understanding of measurement problems and physical phenomena related to.
The study theory focuses on potential sensor technology for canine heart rate measurement. The purpose is, not to develop a new product, but instead to collect and create input data for further product development of more reliable solution. The greatest challenge found was canine fur, which creates great resistance for electromagnetic signals penetration.
Auscultation method was selected for study experiments due it is well known method for canine heart rate analysis in veterinary medicine. Even tough not being the most preferrable technology for a new product development due its size, sensitivity to noise, weight of stethoscopes and patents related. However, auscultation was found to be good choice to build a better understanding of the physical phenomena and measurement problems involved in.
The analysis part of the study demonstrates basic signal analysis, frequency analysis, noise filtering in the canine heart rate measurement case. Theory of basic discrete signal analysis is applied in practice with MATLAB code. Final experiment tries to apply anomaly detection machine learning algorithms in the use cases. The code samples use Microsoft Anomaly Detection API. All the source codes are available at GitHub and are free to use for any learning purposes.
The study theory focuses on potential sensor technology for canine heart rate measurement. The purpose is, not to develop a new product, but instead to collect and create input data for further product development of more reliable solution. The greatest challenge found was canine fur, which creates great resistance for electromagnetic signals penetration.
Auscultation method was selected for study experiments due it is well known method for canine heart rate analysis in veterinary medicine. Even tough not being the most preferrable technology for a new product development due its size, sensitivity to noise, weight of stethoscopes and patents related. However, auscultation was found to be good choice to build a better understanding of the physical phenomena and measurement problems involved in.
The analysis part of the study demonstrates basic signal analysis, frequency analysis, noise filtering in the canine heart rate measurement case. Theory of basic discrete signal analysis is applied in practice with MATLAB code. Final experiment tries to apply anomaly detection machine learning algorithms in the use cases. The code samples use Microsoft Anomaly Detection API. All the source codes are available at GitHub and are free to use for any learning purposes.