Vibration based diagnostics for analysis of combustion properties and noise emissions of IC engines

Sammanfattning: The major challenge in engine diagnostics, monitoring and optimisation is to maintain competitive fuel economy, refinement and specific power while at the same time keeping noise and exhaust emissions at acceptable levels. This requires development and integration of fuel delivery, combustion and after-treatment systems. The main focus of this thesis concerns vibration-based diagnostics and monitoring of the combustion process. Described is the reconstruction from vibration measurements on of the most important parameters, the cylinder pressure, the use of data collected for diagnostic, monitoring and minimazation purposes. The cylinder pressure signal is an optimum parameter for use in the detection of any anomalies present during the combustion process. The basic concept is to use vibration measurements from the engine surface and the engine structure function to perform a reconstruction of the cylinder pressure waveform. While measures of the acceleration signal contain all the necessary information for the reconstruction there is also contaminated by other sources such as impacting valves, piston slaps and combustion of adjacent cylinders. It has therefore been necessary to use advanced signal processing techniques to carry out an accurate reconstruction of the combustion pressure. Since cylinder pressure is an important parameter of engine operation, researchers have been seeking more efficient ways to measure its level. Sensors can measure the cylinder pressure directly or indirectly. The standard method used to collect the pressure signal is the direct method, the so called "intrusive method". This approach entails drilling a hole in the cylinder head and directly monitoring the cylinder pressure signal. This method has many disadvantages, such as, being inappropriate for on-board conditions, expensive to perform, and having a limited lifetime due to the harsh environment inside the combustion chamber. Therefore, the indirect, non-intrusive method that uses the vibration signal has a significant potential for application in on-board diesel engines. The reconstructed cylinder pressure was also used in this thesis research as input data to calculate the heat release rate and to define the start of the combustion process. The start of the combustion has a great importance for optimisation of the fuel injection rate. The heat release measured through the method described here can also be used to determine the efficiency of the engine. Multivariate data analysis (MVDA) was applied to predict noise and exhaust emissions and to calculate optimisation of the combustion process with respect to injection timing, speed, load and fuel quality. Partial least squares (PLS) and principal component analysis (PCA) were used for investigation of the interrelationship between engine parameters and exhaust emissions. PCA is a multivariate projection method that is designed to extract and highlight the systematic variation in a multivariate data matrix. This means that the primary objectives of PCA are to evaluate the underlying dimensional complexity of the data and to obtain an overview of the dominant patterns and major trends in the data. The PLS is a regression extension of PCA, which is used when it is of interest to connect the information associated with the variables to each other. The multivariate data models built were statistically evaluated and the predictive powers were compared with experimental values. The investigation described in this thesis demonstrated that vibration based monitoring can be used to perform the reconstruction of the pressure signal with relatively high accuracy. The reconstructed pressure signal was used to perform the calculation of the heat release rate in the combustion chamber and to predict the noise and exhaust emissions. These findings suggest that vibration based monitoring has the potential to become a versatile tool for diagnostics, engine performance optimisation, and predictions of the combustion process.