Measuring and testing – Vehicle chassis – Steering
Reexamination Certificate
2000-03-23
2002-10-08
Fuller, Benjamin R. (Department: 2855)
Measuring and testing
Vehicle chassis
Steering
C701S106000
Reexamination Certificate
active
06460408
ABSTRACT:
BACKGROUND AND SUMMARY OF THE INVENTION
The present invention relates to a method for determining relevant variables representing the internal pressure in the cylinders of an internal combustion engine, the relevant variables being derived from another measured variable.
U.S. Pat. No. 4.761,993 discloses the use of the structure-borne noise of the internal combustion engine as the other measured variable. In this case, the internal cylinder pressure is measured in a reference internal combustion engine and subjected to cross correlation with vibration signals which are recorded at various points of the internal combustion engine. The frequency band in which these signals are evaluated is specified in this case at up to 30,000 Hz. In order in this case to be able to generate defined load conditions, the method described is used to be able to undertake a functional test of the vehicle in a workshop during maintenance of the vehicle.
It is known from the Journal “Revue Automobile, Vol. 85, No. 15, Apr. 5, 1990, page 17, Bern/CH” to undertake a cross correlation between a vibration signal, which is measured on an engine, and the crank angle position, in order to obtain information on the internal cylinder pressure.
The publication “Proceedings of ICASSP '94, IEEE International Conference on Acoustics, Speech and Signal Processing, Vol. 4 99-22 April 1994, pages IV-149 to IV-152, Adelaide/AU” discloses a model used to draw a conclusion on the internal cylinder pressure from measured structure-borne noise signals by calculating the internal cylinder pressure from the measured structure-borne noise signals with the aid of this model. In this case, a constant speed of the internal combustion engine is prescribed as a boundary condition for the model.
The German laid-open patent application number DE 40 06 273 A1 has furthermore disclosed the determination of a transfer function between one of the variables of travel, speed or acceleration, which are detected at the engine block, and the internal cylinder pressure. The corresponding variable travel, speed or acceleration is detected in this case by means of a strain gauge or an arrangement which uses the laser Doppler effect. The transfer function is determined by detecting on a test stand both the appropriate variable and the internal cylinder pressure. The measured signals are subsequently transformed into the frequency domain in which the transfer function is then obtained by forming the autocorrelate and cross-correlate relating to the corresponding signals. After the transfer function relating to a specific engine type has been obtained on the test stand, the transfer function thus obtained is subsequently used together with the corresponding engine type in order to be able to determine the internal cylinder pressure from the measured signal. Furthermore, the crank angle is determined in order to be able to plot the characteristic of the internal cylinder pressure against the crank angle.
Also known is a method in accordance with which pressure sensors are installed in the cylinder head and used to detect the temporal characteristic of the cylinder pressure. The cylinder pressure can be detected with the aid of this method. However, it is disadvantageous in this case that installing the sensor is comparatively complicated. A plurality of sensors have to be installed in an internal combustion engine in accordance with the number of cylinders. Such a method has therefore been used only on the test stand in the case of the prior art mentioned at the beginning, in order to be able to use a simpler measured variable for diagnosis and/or control/regulation in the case of the relatively large number of the motors used.
A further method is disclosed in DE 44 21 950 A1 in which various input variables have been specified for diagnosing and regulating an internal combustion engine. One of these input variables can be, for example, the acoustic sound of the internal combustion engine, which is to be detected by a microphone and evaluated by means of a neural network.
U.S. Pat. 5,093,792 discloses the detection of fault states such as, for example, misfires, knocking (self-ignition of the mixture before the actual ignition point). These fault states are to be detected by means of a neural network to whose input the internal cylinder pressure is to be fed as a measured variable.
By contrast, it is the object of the invention to provide a method for detecting the cylinder pressure in internal combustion engines in a way which is easy to implement during ongoing operation. This method is intended, moreover, to master various operating conditions as flexibly as possible.
This object is achieved according to the invention by a method in which the cylinder pressures of a test engine are determined from the structure-borne noise of the internal combustion engine by using a neural network. The neural network is trained by using measured structure-borne noise signals from other engines together with associated previously measured signals representing the cylinder pressures. The cylinder pressure of a test engine can be subsequently determined by the trained neural network, based on test engine noise-signal values.
The cylinder pressure can be determined with adequate precision on the basis of the variable structure-borne noise, which can be measured with comparative ease. It is possible by using a neural network to draw accurate conclusions concerning the cylinder pressure from the measured signals with high accuracy, because the neural networks are tolerant of noisy input signals.
With regard to the advantageous use of the neural network, reference may be made furthermore to German Patent DE 195 28 708 C2 which discloses determining the compression of an internal combustion engine by operating the internal combustion engine via the starter, there being no ignition of the internal combustion engine, and the evaluation of the compression being performed via an evaluation of the current of the starter. Compression losses have effects on the starter current, something which is to be taken into account by means of various calibration curves, which are obtained by means of neural networks. The method of this prior art is also based on investigating the internal combustion engine during maintenance. There is no continuous monitoring during ongoing operation.
According to the present invention, the structure-borne noise signals are transformed into the time-frequency domain. Consequently, the fact that the frequencies of the structure-borne noise change with time is taken into account. The frequency analysis of the signals is undertaken in each case in time slots, in order to be able to distinguish differences in the frequency spectrum, for example in the case of an ignition which has just been performed, from the frequency spectrum at the end of the power stroke.
In accordance with another aspect, a neural network is used in order to derive the values for all cylinders. Therefore, there is an increase in outlay during the learning phase, so that advantageously there is a need during ongoing operation for the presence of only one neural network.
Furthermore, one neural network can be used in order to derive the relevant variables representing the internal pressure of one cylinder. In this case, there is a reduction in the outlay during the learning phase because each of the neural networks must learn only the conditions at one cylinder. The structure-borne noise signal, which is produced in temporal conjunction with the opening of the exhaust valve or the exhaust valves of the respective cylinder, is used on the impact variable of the neural network. This structure-borne noise signal supplies a signal characteristic of the respective load of the internal combustion engine.
In another variation, the structure-borne noise signal, which is produced during ignition of the mixture of the respective cylinder in the following power stroke of the internal combustion engine, is used as the impact variable of the neural network.
This structure-borne noise is cor
Gimmler Helmut
Gruden Igor
Holdgrewe Kai
Nester Ulrich
Pischinger Stefan
Crowell & Moring LLP
Daimler-Chrysler AG
Fuller Benjamin R.
Stevens Maurice
LandOfFree
Method for determining relevant variables representing the... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for determining relevant variables representing the..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for determining relevant variables representing the... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2934749