Research On Knock Combustion Characteristics And Identification Algorithm Of Natural Gas Engine

With the increasingly prominent problems of energy shortage and environmental pollution,the development and application process of alternative fuel for automotive engines has been accelerated.Natural gas engine has become the most widely used gas fuel engine in the market due to its advantages of low emission pollution and sufficient energy reserves.With the putting into practice of China VI emission standards,the emission technical scheme of natural gas engine is adjusted from lean combustion to theoretical air-fuel ratio,which increases the possibility of engine knock under equivalence ratio combustion.Knock detection and identification under equivalent combustion is an urgent problem for internal combustion engines.Starting from the poor identification accuracy of engine knock processing chip in the condition of slight knock,this paper takes a natural gas engine that meets the China VI emission standards as the research object to carry out experimental research on knock combustion characteristics and knock identification algorithm.The research work in this paper has important engineering application value and academic significance for improving the knock control and thermal efficiency of natural gas engines and deepening the theoretical research on engine knock.According to the test requirements,the test bench,knock test system,gas supply system,cooling water circulation system and monitoring and control system of natural gas engine are built in this paper.Based on the test bench system of natural gas engine,the knock combustion characteristics of natural gas engine and the influence of different ignition advance Angle and EGR rate on the knock combustion characteristics were studied.The experimental results show that the cylinder pressure signal can effectively reflect the characteristics of knock,while the knock sensor signal is difficult to identify the occurrence of knock in some slight knock states.The experimental results also show that increasing ignition advance Angle will increase knock tendency and knock intensity of engine,while increasing EGR rate can effectively suppress the occurrence of knock.The knock recognition algorithm is studied.Burg algorithm based on AR model is used to estimate the power spectral density of knock sensor signal and cylinder pressure signal,and the characteristic frequency of the engine in this study is determined to be8 k Hz-25 k Hz.Through analyzing the merit and demerit of each recognition algorithm and characteristics of knock sensor signal,wavelet transform and VMD algorithm are selected to extract knock characteristics of knock sensor signal.The wavelet transform algorithm is used to denoise the knock sensor signal,and remove the interference noise outside the specific frequency range,and get the knock characteristic sub-band signal.Based on the center frequency,the optimal decomposition layers and penalty factors of knock sensor signals decomposed by VMD algorithm were determined.The knock characteristic sub-band signals obtained after denoising were decomposed by VMD algorithm to obtain knock characteristic components,and then the feature extraction of knock sensor signals was completed.The determination of knock intensity of knock sensor signal is studied.Taking MAPO as the characterization parameter of knock intensity of cylinder pressure signal,knock intensity grade of cylinder pressure signal was divided.And taking the knock intensity classification result of cylinder pressure signal as reference,analysis the performance of different time domain parameters to characterize detonation intensity,and put forward to the RMS of knock characteristic component of the knock sensor signal as the knock intensity characterization parameters.The knock intensity is graded to determine the knock intensity of the knock sensor signal.Finally,the determination results of knock sensor signal are verified.The results show that the determination method of knock intensity described in this paper can effectively identify the occurrence of knock and determine the knock intensity.