A model-based methodology is presented, which allows the estimation of the characteristic phases of diesel combustion using a semi-physical model approach combined with state and parameter estimation through extended Kalman filtering. The physical relation between the fuel injection and the characteristic diesel combustion phases, such as premixed, diffusive combustion and burn-out, are modeled separately by linear dynamic transfer functions formulated in crank angle frequency domain and transformed into state space representation. The resulting state variables are the released burning energy and its derivatives of each combustion phase. Associated crank angle constants determine the dynamics of the combustion phases and represent the rate parameters to be estimated. By incorporating further physical assumptions regarding the fuel path and air–fuel-mixing dynamics, the combustion phase parameters are estimated online for each working cycle. Cylinder pressure signals and online combustion analysis are used to determine the burn rate of the diesel engine at the test bench. Investigations have shown that the estimated rate parameters depend on the current engine operation point. They are estimated during measurements and stored in lookup tables through an online-learning method based on a fast recursive least squares estimation algorithm.

中文翻译：

---

用于实时应用的柴油燃烧阶段的半物理状态和参数估计

提出了一种基于模型的方法，该方法允许使用半物理模型方法通过扩展卡尔曼滤波结合状态和参数估计来估计柴油燃烧的特征阶段。燃料喷射和特征柴油燃烧阶段（如预混、扩散燃烧和燃尽）之间的物理关系通过在曲柄角频域中制定的线性动态传递函数分别建模并转换为状态空间表示。由此产生的状态变量是释放的燃烧能量及其每个燃烧阶段的衍生物。相关的曲柄角常数决定了燃烧阶段的动态特性并代表了要估计的速率参数。通过结合有关燃料路径和空气燃料混合动力学的进一步物理假设，可以在线估计每个工作循环的燃烧阶段参数。气缸压力信号和在线燃烧分析用于在试验台上确定柴油机的燃烧率。调查表明，估计的速率参数取决于当前的发动机运行点。它们在测量期间被估计并通过基于快速递归最小二乘估计算法的在线学习方法存储在查找表中。调查表明，估计的速率参数取决于当前的发动机运行点。它们在测量期间被估计并通过基于快速递归最小二乘估计算法的在线学习方法存储在查找表中。调查表明，估计的速率参数取决于当前的发动机运行点。它们在测量期间被估计并通过基于快速递归最小二乘估计算法的在线学习方法存储在查找表中。