Analysis Of Combustion Information Extraction Based On ICE Vibration Signals

The in-cylinder combustion process of the internal combustion engine(ICE)determines the power performance,fuel consumption and pollutant emissions,therefore the real-time monitoring and feedback control of the combustion process is necessary to help improving the power performance and reducing fuel consumption and pollutant emissions of the ICE.The information required by the monitoring and feedback control algorithm of the in-cylinder combustion process is able to be provided by the in-cylinder pressure analysis that is usually measured by the pressure sensor mounted in the cylinder.However,such direct measurement needs to refit the cylinder head to install the sensor,which is usually limited by the high cost of the sensor and the harsh test environment in practical application.The vibration signal from engine surface contains valuable information related to the combustion process in the cylinder and other excitations,besides,the vibration sensor is relatively inexpensive and easy to install.Thus,the measurements of engine surface vibrations have a great potential for non-intrusive monitoring of the operating state of ICEs.Comprehensive knowledge of combustion processrelated information by engine vibration measurement is of great theoretical significance and practical value to realize the real-time monitoring of combustion status of ICEs and the closed-loop control of the combustion process.To analyze the distortion of combustion induced vibration signals from the engine surface,reveal the coupling relationship among the vibration signals measured on the cylinder head,extract the combustion induced vibration signal with a high signal-to-noise ratio,and extract the combustion process parameters,based on a single-cylinder diesel engine,the vibration signal measured on the cylinder head were analyzed by using both numerical simulation and experimental methods.The distortion frequency ranges of the measured vibration signal were found,and the main excitations of the measured vibration signals on the cylinder head surface were determined.The coupling relationship between the vibration response signals measured on the cylinder head surface was therefore revealed.The combustion-induced vibration signals with a high signal-to-noise ratio were extracted,and the combustion process parameters were identified based on the extracted vibration signals.The main work and conclusions of the thesis are as follows:A virtual prototype model and a finite element model of the single-cylinder diesel engine are established.Both of the models were well calibrated by the tests.The frequency response functions of the model were calculated.The distortion frequency bands of the vibration response signal on the cylinder head surface were discussed based on the simulations.The effect of varying engine bracket stiffness on the distortion frequency band was analyzed to identify combustion process.Based on the proposed finite element model,the frequency response function was calculated,and the distortion frequency bands of the vibration response signals of the cylinder head were further discussed.The effect of changing the block elastic modulus on the distortion frequency band as well as identification of combustion process parameters were analyzed.The results show when the engine bracket stiffness changes,the combustion process parameters identified based on vibration signals have the same trends as the parameters identified from cylinder pressure.The fluctuation of phase-related combustion process parameters is relatively small compared to that of magnitude-related combustion process parameters.As to the finite element model,the maximum undistorted frequency band of the vibration signals is about 0 to 275 Hz,and the upper limit of this band increases as the block elastic modulus increases.When the block elastic modulus changes,the combustion process parameters identified based on vibration signals have the same trends as that based on cylinder pressure.The magnitude and lag-angle of combustion process parameters decrease as the block elastic module increases.Based on the finite element model of the single-cylinder diesel engine,the vibration responses of the sweep frequency excitation were calculated at different measuring points.Based on the calculated frequency response function,the distortion of the vibration signals at different measuring points and frequencies was analyzed.The differences of the combustion process parameters identified based on the simulated combustion-induced vibration signals of different measuring points were discussed.A one-dimensional thermodynamic simulation model of the single-cylinder diesel engine is established,which enables the incylinder pressure signals calculated at different engine working conditions and combustion model parameters.The spectral characteristics of distortion of different frequency components of the in-cylinder pressure were explored.The effect of different filters on the identification of combustion parameters was discussed.The results show that the non-distortion frequency range of the vibration response signal is 0 to 250Hz.The vibration displacement response signals are seriously distorted in the range of 250 to 450Hz.The vibration velocity response signals are seriously distorted in the ranges of 250 to 450Hz,and 1250 to 2500Hz.The vibration acceleration response signals are seriously distorted in the range of 1000 to 3000 Hz.The location of the measuring points mainly affects the identification of the magnitude-related combustion process parameters extracted based on the vibration signals.As can be seen from the analysis of Fourier Series of the in-cylinder pressure signals,the reconstructed in-cylinder pressure and pressure rise rate based on the undistorted frequency components can only describe the basic shape of the original in-cylinder pressure and pressure rise rate signals,lack of the detail information such as the combustion process parameters.The combustion-induced vibration signal needs at least the components in the frequency band corresponding to the first 140th order harmonics to ensure its integrity and effectiveness for combustion process parameter evaluation.The maximum deviation of the magnitude-related combustion process parameters identified based on the reconstructed vibration signals of this frequency band is about 1.1%.The maximum deviation of the phaserelated combustion process is 0.4 deg.The combustion-induced vibration signals extracted by using band filters can be used to identify phase-related combustion process parameters.The force analysis of the single-cylinder diesel engine was carried out,the main non-combustion excitation signals were calculated,and their time-frequency characteristics were analyzed.Based on the tests of the engine,in-cylinder pressure signals and vibration signals of the cylinder head were measured at the different engine working conditions,and the time-frequency characteristics of the measured vibration signals were analyzed.Based on the established virtual prototype model,the main excitations of the measured vibration signals were determined.Through the time-frequency analysis of the simulated vibration response signals of each excitation,the coupling relationships among the measured vibration response signals were analyzed.The results show that the measured vibration velocity signal contains a low-frequency component with a large amplitude,and the phase of this low-frequency component varies as the engine speed changes.Simulation results show that this low-frequency component is mainly caused by reciprocating inertial force and friction,and the former plays a major role.The friction-induced vibration signal exhibits different fluctuation characteristics when the piston moves upward and downward.Reciprocating inertial force and friction are regular,and the vibration velocity signals before 300 deg are mainly induced by those two excitations.So,the vibration velocity signals at this crank angle range can be modeled,and then be eliminated from the original vibration signals.The simulation analysis shows that the main excitations of vibration acceleration signals measured on the cylinder head are the in-cylinder pressure and reciprocating inertial force.The frequency bands of their vibration response signal overlap with each other,and the energy of the overlap components depends on the engine conditions.Based on the coupling relationship among the vibration signals measured on the cylinder head,the effectiveness of coherence analysis-based filtering methods on the extraction of combustion-induced vibration signals was discussed.The results of the Empirical Mode Decomposition(EMD)method were analyzed,and the excitations of each Intrinsic Mode Function(IMF)were figured out.To improve the issue of mode mixing between IMFs,combustion-related information contained in IMFs mainly induced by non-combustion excitations were extracted and added to IMFs mainly induced by combustion based on a non-combustion response describing model.Start of combustion(SOC)and CA50 were identified by using the Ensemble Empirical Mode Decomposition method based on vibration velocity signals measured on the cylinder head of a modified four-cylinder diesel engine.The results show that the combustion process and the engine types will affect the frequency distribution of non-combustion and combustion-induced vibration signals,and cut-off frequency selection of coherence analysis based band-pass filtering methods,which will deviate the corresponding relationship between the characteristic points on the extracted vibration acceleration curve and the parameters on the in-cylinder pressure curve.There is a mode mixing issue in the IMFs mainly induced by combustion and the non-combustion excitations.By using the reciprocating force-induced vibration signal description model,combustion-related information contained in IMFs mainly induced by noncombustion excitations were extracted and added to IMFs mainly induced by combustion.The lag angle discreteness of the characteristic point corresponding to the combustion starting point on the combustion-induced vibration signals is reduced from 1.355 to 0.476.The low-frequency components in the measured vibration velocity signal are mainly caused by reciprocating inertial forces.The high-frequency components distributed in the range of-30 to 90 deg are mainly caused by combustion.The EEMD algorithm can be used for the extraction of combustion-induced vibration velocity signals.The characteristic points on the extracted IMF6 have a good correspondence with SOC and CA50 under different engine working conditions.The identified SOC and CA50 based on vibration signal lag behind the parameters obtained based on pressure rise rate and heat release rate.The maximum identification deviations are 1.1 deg and 1.4 deg,respectively.To extract combustion-related information,the Fourier Decomposition Method(FDM)algorithm was used to decompose vibration signals measured on the singlecylinder diesel engine head.Simulated results show that FDM can overcome the shortage of EMD based methods such as mode mixing.Then,FDM was used to decompose the measured vibration signal into several FIBFs and the summation of combustion-related FIBFs was used to extract combustion parameters at different working cases.According to the correspondence between feature points on the averaged curves of the second derivate of pressure and processed vibration,some crank angle-related combustion parameters were identified.To evaluate the effect of cyclic variability on identification results,the proposed methods were tested based on data from 90 continuous cycles.The cycle-by-cycle analysis of the results show that in the discussed cases(800 rpm 0 N·m,1000 rpm 10 N·m,1200 rpm 10 to 50 N·m,and 1400 rpm 10 to 50 N·m),the maximum deviation of SOC,LOMPRR,and LOPP are no more than 0.9?A,0.7?A,and 0.9?A,respectively.The parameters identified based on extracted vibration and in-cylinder pressurerelated signals have similar trends,and it suggests that the proposed FDM-based methods can be used for identifying the combustion parameter.The results show that the wavelet coherence method can reveal the coherence of in-cylinder pressure signals and the measured vibration signals in the time-frequency domain and can give more information compared with Fourier Transform based coherence analysis.Based on the wavelet coherence analysis results,in the first region where the crank angle range is 180 to 540 deg and the frequency range is 60 to 250 Hz,and the second region where the crank angle range is 270 to 400 deg and the frequency range is 250 to 3000 Hz,the wavelet coherence lever is bigger than 0.75,which indicate that in these regions the vibration signal is highly related to combustion.A method is proposed to characterize the peak pressure by using the sum of the amplitudes of wavelet coefficients of vibration acceleration signals in these regions,and the maximum identification deviation of the peak pressure of about 96%of the cycle is 0.4MPa.