Patent Application
20090018746
The invention relates to a method and apparatus for obtaining a corrected engine operational parameter value, in particular using cylinder pressure information.
Known engine management systems (EMS) monitor and control the running of an engine in order to meet certain pre-set or design criteria based on engine performance data. Typically the design criteria are good driveability coupled with high fuel efficiency and low emissions, and the engine performance data includes engine cylinder pressure. An internal combustion engine is controlled by an engine control unit which receives sensor signals from a sensor group including a pressure sensor and issues control signals to an actuator group and including, for example valve, spark or fuel injection actuators. The engine control unit also receives external control inputs from external inputs such as throttle controlor gear sensors.
Sensor signals can be obtained from a dynamic cylinder pressure measurement system consisting of a sensing element and a transducer element which provide a signal to signal amplification and normalisation electronics. The signal normalisation electronics are required for signal conditioning to compensate for variabilities and nonlinearities in the raw signal from the pressure sensing element in the cylinder. Such variabilities arise from the temperature dependence of the pressure measurement given by the ideal gas law; PV∂CT and also from hysteresis and sensor drift effects. A typical prior art sensing approach 10 is shown in
A problem with known systems is that the normalisation electronics require calibration for the complete range of temperatures experienced in the cylinder that depend on engine build, fuel type, combustion control strategy and engine conditions. Also the inherent tolerances of the electronic components in question must be considered. Typical electrical and electronic sensing components can have tolerances as high as 10 percent. Furthermore, normalisation electronics of this type are both costly in terms of production and calibration. In order to be effective each sensing system must be adapted specifically for a particular vehicle specification. Furthermore it is necessary in some cases to embed a temperature sensor into the pressure sensor to compensate for variations in engine temperature. Yet further, such an approach does not compensate for changes in performance arising from in service engine wear nor does it compensate for sensor hysteresis, or real-time sensor drift.
The invention is set out in the claims.
Advantages of the present invention compared to that of the prior art are that the present invention provides an estimation of the cylinder temperature and/or pressure or other engine effects stored in hardware to provide an accurate correction strategy rather than a signal normalisation/conditioning regimen. In addition the transient effects of hysteresis can be overcome by using previously estimated cylinder temperatures and/or pressure, or other engine parameters.
Embodiments of the invention will now be described by way of example with reference to the drawings, of which:
a shows an approach to correcting for time lag in sensed pressure; and
b shows an approach to correcting for hysteresis.
In overview, the invention makes use of a pressure sensor model to compensate for inaccuracies in the raw signal from the real pressure sensor, rather than providing dedicated and application specific compensation hardware. In one embodiment a real pressure sensor signal output is compared for discrepancies with an estimated output from an engine pressure model including an engine model and a sensor model. The engine pressure model is corrected dependent on the discrepancy and the modelled engine pressure is then taken as the output value. Because the sensor model can include a “pure” pressure sensor model and a “pure” temperature sensor model, these values, which will both influence the output of the real pressure sensor, can be decoupled in the model.
In particular, cylinder pressure measurements are carried out whereby inaccuracies inherent with such measurements are taken into account by comparing data provided within a model with actual data and updating the model based on this comparison. Accordingly, effects such as null shift hysteresis and spark timings during the ignition stroke of each cylinder, or where variations in air/fuel ratio are compensated. In particular the model data converges with the actual data to provide an accurate picture of the engine performance for use in an engine management control unit. Furthermore, linearization of the measured parameter takes place after signal amplification thereby allowing manipulation of the sensor signal in accordance with model data.
The sensed cylinder pressure is a function of other operational parameters such as cylinder temperature. Both of these are modelled and the model updated from comparisons of the sensor pressure measurement and the modelled sensor output until convergence is reached. The decoupled value of pressure (or other operational parameters) can then be extracted from the values used by the converted model.
In particular, real-time in-cycle cylinder temperatures are estimated based on current engine cycle observations and used to condition, correct and improve cylinder pressure measurements. As a result temperature sensors embedded in the pressure sensor are not required. The improved pressure measurements are more representative of the real cylinder pressure. Furthermore, the measurements can be used to compensate for dynamic effects such as hysteresis, and can be stored in hardware for correcting future estimations by deciding where on the sensor hysteresis curve the sensor is operating and correcting the sensed pressure measurement accordingly.
Various approaches can be adopted for correcting for hysteresis which can be better understood with reference to
Referring now to
The arrangement shown in
Furthermore as a direct result of using real-time in-cycle cylinder temperature estimations based on the current engine observations it is possible to compensate for sensor drift for example by estimating the polytropic index n variations between engine cylinders and cycles thereby overcoming sensor drift, as discussed in more detail in UK patent application number 0112338.9, incorporated herein by reference.
In another embodiment the real pressure sensor signal output is received as an input by an inverted sensor model which also receives an estimated temperature from an engine model. The output from the inverted model is then the modelled pressure in the engine. It is also contemplated that a real temperature sensor signal is received as an input by the inverted sensor model which can also receive an estimated pressure from an engine model. The output from the inverted model is then the modelled temperature in the engine.
As discussed in more detail below the real time digitised signal data from the analogue to digital converter (A/D) 35 is compared with that of the engine model data and the results fed back to modify the engine model 38. Following each iteration, of measure, compare and feedback the engine model data is updated such that pressure data which has undergone normalisation gradually converges with the actual pressure in the cylinder.
The engine has control inputs such as throttle or gear selection and additional outputs such as engine speed and power output. The pressure sensor output is received as an input by pressure value comparator 43. The additional sensor outputs are received as inputs by one or more respective additional value comparators 44.
The engine model 42a models engine pressure {circumflex over (P)}cyl and temperature {circumflex over (T)}cyl based on common control inputs 46 with real engine 41a. The modelled pressure and temperature are received as inputs by a sensor model 42b which outputs a modelled sensor output 47 to the pressure value comparator 43. The engine model also provides additional sensor outputs as inputs to the additional value comparator 44. The engine model 42a contains information related to the engine in question such as fuel type, engine capacity and efficiency. The sensor model 42b contains a prediction of how the real sensor 41b will respond to a given pressure in the modelled engine. The various components individually are well known to the skilled reader e.g. The Ricardo Network Vehicle Controller (see Channon, S & Miller, P, An Advanced Network Vehicle Controller (NVC) To Support Future Technology Applications, Berlin, Germany; Springer-Verlag, 2003, Advanced Microsystems For Automotive Applications 2003, based on VDI/VDE Conference May 2003) which contains the engine model implemented on a Motorola MPC8260 microprocessor.
In operation the observer based approach is implemented as follows. A measurement of the engine cylinder pressure is taken using the real cylinder pressure sensor 41b and dependent on the actual in-cylinder conditions Pcyl and Tcyl. This measurement is compared to the modelled output at comparator 43. The difference is then fed back into the model 42 in order to update the model data. The predicted pressure and temperature from the model 42 are then fed to an engine control strategy 45 where they are interpreted and used to control the real engine 41a and further update engine model 42a. Following several iterations, in real time, of comparison and feedback, the model data values will approach the real data values until the two values converge. When the two values converge, the model data can be considered to be an accurate prediction of real time engine performance, in terms of cylinder pressure or other modelled parameters.
In order to improve the model yet further, as well as cylinder pressure monitoring, there is also provided yet a further level of monitoring, where additional sensor output data from the engine is compared at comparator 44 with that generated by the model. As with the pressure measurements described above, the real engine data is compared with the model data at the additional value comparator and the difference of this comparison is fed back to update the engine model further.
The steps of reading the pressure and updating the model to control the real engine are carried out in real time thereby constantly updating the model for all engine conditions. Such conditions can include idle, accelerating, decelerating or running at a constant speed. In this way the performance of the engine under all conditions is known and it is thereby possible to provide an engine that approaches maximum efficiency. Furthermore, such a system can also be used to provide feedback of cylinder conditions where the engine is operating under exceptional circumstances, for example where one of the cylinders has failed or where there has been a gradual change in cylinder geometry through engine wear and deterioration, including the hysteresis of the sensor.
The converted pressure value can be used for any appropriate purpose, for example modifying the engine timing for improved performance as described in UK application 0112338.9, incorporated herein by reference.
Although the above discussion contemplates correcting the model throughout engine operation, it will be appreciated that model correction can be carried out only during certain parts of engine operation, for example only during certain parts of the cylinder cycle. For example during the exhaust stroke of the cylinder the sensed real pressure values can be disregarded and during this phase the model can remain uncorrected. It will be appreciated that during this phase accurate prediction of the pressure is less significant. Furthermore modelling of the pressure in this phase is more complex, and correction correspondingly so such that computational resources can be saved in this manner. However, during the compression stroke, when air in the cylinder comprises a trapped mass, modelling is simplified and can be derived from the ideal gas law PV=MRT, given that the variation of volume with time is known from the cylinder dimensions and crank shaft position. As a result the modelled values are only updated during significant phases of operation and are frozen during other phases of operation. It will be seen that this approach is particularly applicable where individual phases of operation of the engine are modelled independently such that correction to the models during one phase will not affect the model during other phases.
By virtue of the features described above it will be seen that the invention provides a stable signal output related to cylinder pressure which can be modelled empirically in real time and provides a fast response time beneficial for knock detection and overall stability whilst being able to compensate for slow sensor response times. The invention is robust to cylinder temperature variations and hysteresis effects and can compensate for drift. In addition the (need for conditioning electronics is removed or reduced, reducing the cost as well as the effects of production tolerances. The discussion above is principally directed to obtaining an accurate value for pressure. However, it will be appreciated that it can be expanded to embrace a range of parameters. Indeed where two or more parameters in an engine cylinder affect a sensor output then the model may include modelled “pure” sensors for one or more of the decoupled parameters. In this case the model will be corrected by the real sensor output (which can even be in relation to a parameter unrelated to the modelled parameters) and the modelled sensed value for any of the individual parameters can be obtained. Accordingly the invention extends to a method of obtaining a corrected engine operational parameter value comprising obtaining a measurement of an operational parameter value from said engine using a sensor, independently estimating an operational parameter value measurement updating the estimate based on the measured operational parameter value and providing a corrected engine operational parameter value based on said update estimate.
Although the above description is based on cylinder pressure, it will be appreciated that other sensor data such as engine data or cylinder temperature values can be compared in a similar manner. Instead of an engine or sensor model it could be based on calibrated look up tables or other method of similarly obtaining data on equivalent parameters to establish other known measures of engine performance.
| Number | Date | Country | Kind |
|---|---|---|---|
| 0410135.8 | May 2004 | GB | national |
| Filing Document | Filing Date | Country | Kind | 371c Date |
|---|---|---|---|---|
| PCT/GB2005/01631 | 4/28/2005 | WO | 00 | 6/6/2007 |
