BRIEF DESCRIPTION OF THE DRAWINGS
The advantages described herein will be more fully understood by reading an example of an embodiment in which the invention is used to advantage, referred to herein as the Detailed Description, with reference to the drawings wherein:
FIG. 1 shows cylinder pressure, raw vibration sensor signal, and windowing of the vibration sensor signal plotted as a function of crank angle degree;
FIG. 2 is a schematic view of an internal combustion engine;
FIG. 3 is a schematic view the inventive method according to an aspect of the invention;
FIG. 4 shows data comparisons between prior art in-cylinder pressure transducer sensing calculations of combustion phasing compared the inventive method according to an aspect of the present invention; and
FIG. 5 is a plot of engine control using the inventive method when cetane number of the fuel is changed.
DETAILED DESCRIPTION
A 4-cylinder internal combustion engine 10 is shown, by way of example, in FIG. 2. Engine 10 is supplied air through intake manifold 12 and discharges spent gases through exhaust manifold 14. An intake duct upstream of the intake manifold 12 contains a throttle valve 32 which, when actuated, controls the amount of airflow to engine 10. Sensors 34 and 36 installed in intake manifold 12 measure air temperature and mass air flow (MAF), respectively. Sensor 31, located in intake manifold 14 downstream of throttle valve 32, is a manifold absolute pressure (MAP) sensor. A partially closed throttle valve 32 causes a pressure depression in intake manifold 12 compared to the pressure on the upstream side of throttle valve 32. When a pressure depression exists in intake manifold 12, exhaust gases are caused to flow through exhaust gas recirculation (EGR) duct 19, which connects exhaust manifold 14 to intake manifold 12. Within EGR duct 19 is EGR valve 18, which is actuated to control EGR flow. Fuel is supplied to engine 10 by fuel injectors 30, injecting directly into cylinders 16. For embodiments in which the engine is either a spark ignition engine or a spark-assisted, homogeneous-charge, compression-ignition engine, each cylinder 16 of engine 10 contains a spark plug 28. The engine may have port injectors 26. Such port injectors are often used with spark ignition engines to provide a mostly premixed fuel-air mixture to the engine. HCCI engines, in which a homogeneous charge is desired may also use a port injector. Diesel engine, which rely on compression ignition, sometimes supply a portion of the fuel in a premixed mode, sometimes called fumigation. This premixed portion may be supplied by port injectors or a carburetor (not shown). The crankshaft (not shown) of engine 10 is coupled to a toothed wheel 20. Sensor 22, placed proximately to toothed wheel 20, detects engine 10 rotation. Other methods for detecting crankshaft position may alternatively be employed. Sensor 24 is a vibration sensor, which in a preferred embodiment is an accelerometer mounted to the engine structure. Alternatively, sensor 24 is a microphone or other device capable of detecting vibration of the engine structure. Accelerometers provide a charge signal, which is known to be converted to a voltage signal prior to data collection. The vibrations of the engine structure which are caused by the combustion are sensed by vibration sensor 24. The signal of vibration sensor 24 is preferably directly lowpass-filtered to avoid alias effects (C. Vigild, A. Chevalier, and E. Hendricks: “Avoiding signal aliasing in event-based engine control”, SAE Paper: 2000-01-0268, 2000).
Engine 10 of FIG. 2 is preferably pressure charged by a supercharger or turbocharger. The exhaust turbine 56 of a turbocharger is shown which is a variable geometry type. A turbocharger is coupled to an intake compressor by a shaft (not shown). By adjusting the variable geometry turbocharger, the amount of intake boost can be adjusted. The intake compressor 58 of FIG. 2 is either the turbocharger variety, which is shaft driven by the exhaust turbine or a supercharger, which is driven via the engine's crankshaft (connection not shown). The supercharger can be waste gate controlled to control intake boost.
Continuing to refer to FIG. 2, electronic control unit (ECU) 40 is provided to control engine 10. ECU 40 has a microprocessor 46, called a central processing unit (CPU), in communication with memory management unit (MMU) 48. MMU 48 controls the movement of data among the various computer readable storage media and communicates data to and from CPU 46. The computer readable storage media preferably include volatile and nonvolatile storage in read-only memory (ROM) 50, random-access memory (RAM) 54, and keep-alive memory (KAM) 52, for example. KAM 52 may be used to store various operating variables while CPU 46 is powered down. The computer-readable storage media may be implemented using any of a number of known memory devices such as PROMs (programmable read-only memory), EPROMs (electrically PROM), EEPROMs (electrically erasable PROM), flash memory, or any other electric, magnetic, optical, or combination memory devices capable of storing data, some of which represent executable instructions, used by CPU 46 in controlling the engine or vehicle into which the engine is mounted. The computer-readable storage media may also include floppy disks, CD-ROMs, hard disks, and the like. CPU 46 communicates with various sensors and actuators via an input/output (I/O) interface 44. Examples of items that are actuated under control by CPU 46, through I/O interface 44, are fuel injection timing, fuel injection rate, fuel injection duration, throttle valve 32 position, spark plug 28 timing, EGR valve 18, boost pressure, etc. Various other sensors 42 and specific sensors (engine speed sensor 22, vibration sensor 24, engine coolant sensor 38, manifold absolute pressure sensor 31, air temperature sensor 34, and mass airflow sensor 36) communicate input through I/O interface 44 and may indicate engine rotational speed, vehicle speed, coolant temperature, manifold pressure, pedal position, engine vibration, throttle valve position, air temperature, and air flow. Some ECU 40 architectures do not contain MMU 48. If no MMU 48 is employed, CPU 46 manages data and connects directly to ROM 50, RAM 54, and KAM 52. Of course, the present invention could utilize more than one CPU 46 to provide engine control and ECU 40 may contain multiple ROM 50, RAM 54, and KAM 52 coupled to MMU 48 or CPU 46 depending upon the particular application.
Referring to FIG. 3, the raw vibration sensor data are collected over a crank angle window in step 90. An example of a raw vibration sensor signal is shown as curve 60 in FIG. 1. In step 92, the data are adjusted so that the mean over the windowing interval is zero. Thus, the data set are either raised or lowered, as appropriate, to cause the data set to have a mean of zero. Curve 60 in FIG. 1 has an average voltage below 0, which is in contrast to plot 106 showing raw data, which has an average voltage at 0 (visually appears to be centered around zero). In step 94, the data are bandpass filtered removing frequency components below CL and above CH. CL and CH are calibratable quantities which can vary depending on the engine architecture, mounting of the vibration sensor, engine speed, and any engine operating conditions. The filter can be run either forward or forward-backward to cancel the phase delay due to the filter. CL is expected to be of the order 1 kHz and CH of the order of 10 kHz. The filtered data appear as shown in graph 104 of FIG. 3. The data are squared in step 96, K2(θ) per the equation above, when a=2. Alternatively, other powers can be used. The data are rectified (absolute value) and lowpass filtered in step 98 in which all frequency components above CX are removed, as shown in graph 108. CX is calibratable and may vary based on similar parameters as with CL and CH. The data are summed in step 100,
forming a vector of the E values at each crank angle in the window from θo to θn, where θo is the starting crank angle and θn is the ending crank angle of the window of interest:
{right arrow over (E)}
θ
=[E
θ0
,E
θ1
, . . . E
θn].
Plotting the values of the E vector with respect to crank angle results in the curves shown in graph 110. Because the data of graph 104 are positive, E, of graph 110, is a monotonically increasing function with its peak value occurring at On, the end of the sampling window. Thus, the maximum E value is Eθn or E(θn). The 10%, 50%, and 90% combustion intensity times are shown in graph 110. These are computed as:
E
θp
=p·E
θn
where p is a percentage. Thus, if the crank angle of 50% combustion energy is sought, θp is the crank angle which satisfies the above equation, i.e., the crank angle at which Eθp most nearly equals 50% of Eθn. Any percentage between 0 and 100% can be similarly determined.
The order of the operations shown in FIG. 3 can be altered without departing from the scope of the present invention. For example, the bandpass filtering of step 94 can be accomplished after computing the power of the signal, step 96. Another example is that the lowpass filtering of step 98 can occur after the summing operation of step 100.
The choice of the sampling window is important, but not terribly crucial because, as can be seen in curve 110, toward the end of the sampling window, the E curve becomes rather flat indicating that combustion is proceeding very slowly, if at all. Thus, very little increase in Eθn would occur should the window be lengthened. Similarly, the slope of the curve near the initiation of rapid combustion is quite flat. The areas of rapid combustion, for the particular data collected in graph 110 happen between 10% and 50% combustion energy and right before 90% combustion energy.
It is known to one skilled in the art that certain combustion intervals are useful for characterizing combustion. It is common to compute the crank angle degrees of 5%, 10%, 20%, 50%, 80%, 90%, and 95% mass fraction burned as well as 5%-95% interval, 10%-90% interval, and 20%-80% interval. According to the present method, the crank angles associated with any percentage of the energy factor can also be computed. The combustion intensity relates to the rate of major fraction of the combustion interval; any of 5%-95%, 10%-90%, 20%-80%, or other intervals can be used.
Referring to FIG. 4, the data are compared between the prior art method, in which combustion phasing is determined via a pressure transducer signal, and the present method using a vibration sensor. In graph 120, the crank angle degree at which 5% of the fuel mass fraction is burned (MFB5 of the x-axis) is plotted against the 10% combustion energy, as computed by the present method, for an operating condition of 1500 rpm and IMEP (indicated mean effective pressure) of 262 kPa. The thick line in graph 120 represents hundreds of individual data points collected over which many engine parameters were varied: beginning of activation (BOA) of the fuel injector, EGR rate, and fuel cetane number. The two quantities correlate exceedingly well, with an offset of about 8°, in which the 10% E value, according to the present method, occurs about 8° after the MFB5 crank angle. This delay has several components: a delay in transmission of the pressure signals within the cylinder to the engine structure, a delay in transmission through the engine structure to the knock sensor location, and delays introduced by the filtering operations. The filters can be of lower or higher order. The higher the order, the more computationally intensive and the greater the delay introduced. However, at certain operating conditions, such a higher order may improve accuracy of the resultant energy factors. Another factor to consider relates to mechanical delays in engine vibrations traveling to the vibration sensor and other delays in the system. The correlation between MFB5 and 10% E value, in graph 120 of FIG. 4, is exceedingly linear with only a few outlier values. Furthermore, the delay of 8° is constant across the range of graph 120 regardless of the type of engine parameter that was varied. Similarly excellent correlations are shown in graphs 122, 124, and 126 of FIG. 4, which correspond to engine rpm conditions of 1200, 1500, and 2000 rpm, all at 500 kPa IMEP. All show linear correlations with nearly constant offsets; although the offset appears to be related to rpm; offsets are about 8°, 9°, and 11° for the 1200, 1500, and 2000 rpm conditions, respectively.
In FIG. 5, an engine was operated with 3 fuels of differing cetane number at an operating condition of 850 rpm and 262 kPA BMEP (brake mean effective pressure). Cetane number relates to the ability of the fuel to autoignite. The higher the cetane number, the more rapidly the fuel autoignites. Recall that in regard to a spark ignition engine autoignition is to be avoided. However, in regard to a compression ignition engine, autoignition is the phenomenon by which ignition occurs. It is desirable to have a fuel which rapidly autoignites because a delay in autoignition allows more fuel to mix with air and the resulting ignition event is harsher, leading to noise and undesirably high pressures in the cylinder. Prior to about cycle 10 in FIG. 5, the engine is operating open loop, i.e., no feedback based on the vibration sensor is being employed. The y-axis of FIG. 5 is a measure of the difference between BOA (when the fuel injector is activated) and the crank angle when the energy factor reaches 10% of its maximum value. As expected, the 35 cetane fuel takes longer, to reach the 10% point because it is a slower igniting fuel than the 55 cetane, which is a faster igniting fuel using (about 14° compared to about 12°, respectively). Feedback control of the injection timing is switched on at cycle 10. Although not shown in FIG. 5, when injection timing is feedback controlled to maintain a particular combustion phasing, the control system causes the 35 cetane injection timing to be advanced and the 55 cetane injection timing to be retarded such that the 10% energy factor point coincides for all three fuels after cycle 10 thereby providing the desired combustion phasing regardless of fuel cetane quality.
The signals can be sensed either in the time domain or in the crankshaft angle domain. When sampling in the time domain, there is a fixed time interval between the sampling points, and when sampling in the crankshaft domain there is a fixed crankshaft angle between the sampling points. Of course, the sampling can also be carried out according to other schemes and the sampling rate may, for example, vary (in the angle domain or in the time domain). In the last mentioned case, it is possible, in particular, to achieve a high signal resolution in specific signal ranges of interest.
For the desired characterization of the combustion behavior by the vibration sensor signal, correct synchronization with the crankshaft angle, θ, is highly significant. However, as a rule θ is sensed by a toothed disk on the flywheel, with angular resolutions of typically 3°, 5°, 6° or 10° are obtained, due to the distance between the teeth. In contrast, here higher resolutions up to 0.1° or less are desired. θ is therefore determined in engine control unit 40 with the necessary fineness by interpolation or extrapolation from the raw data. Interpolation can be applied when θ is not desired immediately, and can therefore be calculated as an intermediate value between two successive sampling points. Otherwise, if θ is to be used immediately, it is extrapolated from the preceding sampling points.
The lowpass and bandpass filtering may either be of the forward type or of the forward/reverse type. Filters of the forward type filter a signal only in the forward direction, that is to say the angle θ increases incrementally in such a filter. For this reason, forward filters are less computationally intensive and are suited for online calculations, i.e., for calculating current events. However, due to the nature of these filters they bring about a phase shift in the input signal. In contrast, filters of the forward/reverse type filter a signal both in the forward direction and in the reverse direction so that they can compensate phase shifts. However, they typically require a higher degree of expenditure on computation than corresponding forward filters and can only be used offline, for example in calculations between combustion events.
While several modes for carrying out the invention have been described in detail, those familiar with the art to which this invention relates will recognize alternative designs and embodiments for practicing the invention. The above-described embodiments are intended to be illustrative of the invention, which may be modified within the scope of the following claims.
Patent Application
20080053405
