Fallprediktion på träd

Sammanfattning: It is nothing new that trees will at times fall and cause harm to infrastructure, people, or property. When this occurs, it is often the prelude to insurance cases in an attempt to obtain redemption to injury or capital to rebuild damaged property or infrastructure. The main objective of this thesis, as intended by the patron of the project, was to research the possibility to predict if there is a risk of a tree failure beforehand, preventing unnecessary harm. This is feasible, making use of multiple sensors and gather values on multiple parameters over time. Using the gathered data to register anomalies in a trees behaviour to predict heightened risk of tree failure, and thus give the user a chance to take preventive action. The conclusion made from the measurements is that it is possible to analyse data from the prototype. The accelerometer gave us good data on how great the angle of the tree was, and it gave us a difference in the angle during more windy sessions. The angle of the tree can then be of use during a longer time of studying trees to get a more accurate reading on how the angle can change during the trees life and what it means. The earth humidity sensor, the air humidity sensor and the air temperature sensor all functioned as planned and gave a good summary of the condition of which the tree was under. These data points are a good way to study the overall health impacts of the surroundings of the tree over time. To make the prototype more complete we would need to add a sap-flow sensor and a weather station. The sap-flow sensor is a crucial instrument in our opinion to get a more complete reading of the health state of the tree. The sap-flow sensor is although a very expensive product to buy and did not make it into the project for just this reason. The weather station is a good addition to the overall data collection. The data that has been collected so far is just a start to a bigger project, due to the complexity of a tree’s anatomy and health. The data that would be needed to determine a fall prediction of a tree, is data that is collected during a long period of time and on several different trees and tree-sorts. The goal to predict a tree fall has not been achieved due to the complexity of the prediction. To predict a tree fall we would need to study trees during a much longer time frame and with multiple sensors to even come close to predicting a tree fall. Although the project itself failed to produce a prototype to answer the question of fall prediction, this is a first real step to achieving that goal.