Adaptable slope estimation module and its application in a coolant monitoring system for predictive observation

Abstract

In this thesis, a slope estimation module for predictive maintenance of manufacturing based variables is developed. Based on the output of a state estimation filter, the slope of the data is calculated and used for predictive maintenance of a system. This helps to indicate possible failure of a system or a component of a system ahead of time. For achieving this goal, two common sensors are examined in particular: an ultrasonic distance sensor and a pH meter. The output of these two sensors is used to determine the requirements for such a slope estimation module. The thesis puts its emphasis on 4 different main points. In the beginning, the applied sensors are investigated on their output behavior regarding noise, resolution, hysteresis and so forth. This step is done to define the requirements for the slope estimation module. In the second step, a set of example signals is designed based on the output of the sensor testing. These example signals contain all the characteristics which are required to be detected by the slope estimation module. Afterwards, four different state estimation filter algorithms are developed, tuned and tested on the example signals. Based on the output of the final state estimation filter, the slope is computed and a warning is shown if the variable shows a trend and is approaching a minimum or maximum value. At the end, the developed algorithm is validated with the help of an original signal from both sensors. Based on the developed slope estimation algorithm, a warning can be given to the operator if a system variable is approaching a critical range close to its maximum or minimum. The algorithm is very robust against high frequency noise, interruptions in sampling data as well as wrong measurements. However, low frequency noise with a high amplitude can lead to false warnings.