GB2442751A - Engine Parameter Signal Estimation - Google Patents

Abstract

An engine management system computes an estimation of a first signal representative of a engine parameter (e.g., engine cylinder pressure) based on a second signal dependent upon a measured engine operation parameter (e.g., engine speed) and applies a bias compensation to the estimation of the first signal to determine the first signal representative of the first engine parameter. The estimation of the first signal may be computed using frequency analysis. The fourier transform of the second signal may be calculated, the result divided by a transfer function, and then the inverse fourier transform of the result of the division taken in order to compute the estimation of the first signal.

Description

ENG[NE PARAMETER SIGNAL ESTIMATION

BACKGROUND

The present invention relates to estimation of a signal representative of an engine parameter.

In recent years, increasing requirements for improved engine diagnostics and control have led to the development of sensing and signal processing technologies to provide for more efficient engine management.

It is known to provide for engine management under the control of a so-called torque-on-demand algorithm, whereby the engine is regulated so that it meets a driver's operational intention in terms of amount of torque required. According to the instantaneous engine torque and many other parameters, the algorithm deploys the most appropriate control to optimize the engine conditions. The instantaneous engine torque can be calculated by volumetric efficiency (calculated by engine speed) and an amount of intake air then compensated by ignition advance and air-fuel ratio.

However the accuracy of such an approach can be limited if it does not take account of atmosphenc conditions, engine wearing and abnormal combustion etc. In a so-called closed loop control algorithm, an engine's actual instantaneous torque can be calculated based on an Indicated Mean Effective Pressure (IMEP) obtained directly from in-cylinder pressure sensing, by which the accuracy is fully ensured. A cylinder pressure signal for a closed loop algorithm can be measured directly by a cylinder pressure sensor. A cylinder pressure sensor has to operate in a harsh operation environment and is exposed to high temperature gas and flame, which are caused by mixture combustion. This extremely high temperature condition usually leads to some error on the sensor output signal. For research and instrumental purpose, the water-cooling or non-water-cooling high-accuracy sensor can be used.

But the high cost for this type of sensor makes it very difficult to be used for I, production engine. A cost-effective cylinder pressure sensor on the other hand cannot meet the accuracy requirement for engine control purpose.

Accordingly, an embodiment of the invention seeks to provide for improved performance and emission efficiency of a production engine in a cost-effective manner. p

SUMMARY

Aspects of the invention are defined in the accompanying claims.

An example embodiment of the invention can provide an engine management system that computes an estimation of a first signal representative of a first engine parameter using a second signal representative of a measured engine operation parameter and applies a bias compensation to the estimation of the first signal to determine the first signal representative of the first engine parameter. l0

The bias compensation can be calculated from the ideal adiabatic process for the compression stroke. A value of bias compensation can varied until the standard deviation of a constant is minimised, which means that the group of constants derived based on a bias compensation is closest to the constant term. The engine management system can include storage for storing the determined bias compensation.

An example embodiment of the invention can facilitate improved performance and emission efficiency of a production engine in a cost-effective manner without needing the production engine to be provided with sensors for directly sensing the combustion process inside an engine cylinder.

In an example embodiment the first signal can be representative of an engine cylinder pressure and the second signal can be an engine speed signal. The instantaneous engine torque can then be computed from the engine cylinder pressure of respective engine cylinders.

In an example embodiment, the estimation of the first signal can be calculated using frequency analysis. For example, the estimation of the first signal can be computed by calculating a Fourier transform of the second signal, dividing the result by a transfer function, and taking the inverse Fourier transform of the result of the division.

The transfer function can be predetermined as a Fourier transform of a reference first signal generated by measuring the first engine parameter for a reference engine, divided by the Fourier transform of a reference second signal generated by measuring the engine operation parameter for the reference engine. The engine management system can include storage for storing the predetermined transfer function. The engine management system can use the predetermined transfer function to compute the estimation of a first signal.

The engine management system can include processing logic for implementing the estimation and/or compensation phases and/or a software-controlled processor and program instructions for implementing the first and/or second means.

An example embodiment of the invention can also provide a method of engine management in which an engine management system computes an estimation of a first signal representative of a first engine parameter using a second signal representative of a measured engine operation parameter and applies a bias compensation to the estimation of the first signal to determine the first signal representative of the first engine parameter.

An example embodiment of the invention can provide a computer program product comprising program instructions operable to control a processor of an engine management system to perform such a method.

Although various aspects of the invention are set out in the accompanying independent claims, other aspects of the invention include any combination of features from the described embodiments and/or the accompanying dependent claims with the features of the independent claims, and not solely the combinations explicitly set out in the accompanying claims.

BRIEF DESCRIPTION OF THE FIGURES

Specific embodiments of the present invention will now be described by way of example only with reference to the accompanying Figures in which: Figure 1 is a schematic representation of a vehicle drivetrain; Figure 2 is a schematic representation of an engine management system for an internal combustion engine of the engine drivetrain of Figure 1; Figure 3 is a flow diagram giving an overview of a method of estimating a signal representing an engine parameter; Figure 4 is a flow diagram of an example of a calibration phase of the method of Figure 3; Figure 5 is a schematic representation of the calibration phase; Figure 6 is a flow diagram of an example of an on-line estimation phase of the method ofFigure3; Figure 7 is a schematic representation of the on-line estimation phase; Figure 8 is a schematic representation of the basis for bias compensation as part of the on-line estimation; Figure 9 is a schematic representation of the calculation of the bias compensation; and Figure 10 is a schematic representation of an optional further optimisation.

While the invention is susceptible to various modifications and alternative forms, specific embodiments are shown by way of example in the drawings and are herein described in detail. It should be understood, however, that drawings and detailed description thereto are not intended to limit the invention to the particular form disclosed, but on the contrary, the invention is to cover all modifications, equivalents and alternatives falling within the spirit and scope of the present invention as defined by the appended claims. b

DESCRIPTION OF EMBODIMENTS

Aspects of the invention are defined in the accompanying claims.

In the following an example embodiment of the invention will be described in which an engine management system uses a frequency analysis based estimation process to estimate a first signal representative of a first engine parameter using a second signal representative of a measured engine operation parameter and applies a bias compensation to the estimation of the first signal to determine the first signal representative of the first engine parameter.

In the described example embodiment, the first signal is representative of an engine cylinder pressure and the second signal is an engine speed signal. The instantaneous engine torque can then be computed from the engine cylinder pressure of respective engine cylinders. However, in other embodiments the first and second signals could be representative of other parameters.

The engine management system is responsive to driver pedal input (e.g., user activation of the accelerator pedal) to determine the intentions of the driver to determine a torque value for the engine and to adjust engine parameters in response thereto with the aim of achieving the driver's desired vehicle performance while taking account of emissions requirements. In such a system, the cylinder pressure can be used to determine the engine torque.

Many types of cylinder pressure sensor exist in the marketplace. However, reliable and accurate cylinder pressure sensors are very expensive, especially when considering larger engines with 6 or 8 cylinders. The use of unreliable or inaccurate cylinder pressure sensors is not a viable option as this would have limited benefit and could indeed be detrimental to engine performance and emission efficiency. Further the addition of new sensors to an engine would increase the cost and complexity of that engine, particularly if measures are needed to address potential reliability issues.

The described embodiment of the invention is instead based on cylinder pressure o estimation for a production engine based on an existing engine management system (EMS) signal.

Figure 1 is a schematic representation of a drivetrain 12 for a vehicle 10, where an internal combustion engine 14 produces drive at 16 to a flywheel 18. The drive from the flywheel is connected via a gearbox 20 to a driveshaft 22 and via a differential 24 to the driven wheels 26. By determining the cylinder pressures for the respective cylinders of the internal combustion engine, the torque at the flywheel 18 can be determined, and from this the torque at the driven wheels can be determined.

Figure 2 is a schematic representation of an engine management system (EMS) 30 for the internal combustion engine of Figure 1. Figure 2 illustrates an EMS unit 32 that includes processing logic 34 and storage 36. The processing logic can be formed at least in part by one or more software programmable processors (e.g., microprocessors and/or microcontrollers) and/or at least in part by special purpose logic, for example one or more Application Specific Integrated Circuits (ASICs). The storage 36 can, for example, comprise a combination of volatile and non-volatile memory for the storage of program instructions and data. The EMS unit 32 is responsive to signals 38 representative of, for example, engine parameters sensors, and other sensors, for example for detecting the position of a throttle 39.

An example of an engine parameter that can be sensed by the EMS includes the engine speed, or crankshaft speed. This can be derived, for example from a conventional crankshaft sensor that outputs a signal responsive to the rotation of the crankshaft.

Figure 3 is a flow diagram giving an overview of a two phase method 40 of estimating a signal representing an engine parameter. The method includes an off-line calibration phase 42 and an on-line control phase 70.

Figure 4 represents the first, off-line calibration phase 42 in which a transfer function is determined.

The transfer function (H(s)) can be determined 44 in the first phase 42 by: -determining (46) a Fast Fourier Transform (FF1) (FFT1) of a reference first signal generated by sampling at predetermined intervals a first engine parameter (in the present example a cylinder pressure) for a reference engine; -determining (48) an FF1 (FFT2) of a reference second signal generated by sampling an engine operation parameter (in the present example the engine speed) at the same sampling times as for the first engine parameter for the reference engine (this can be done at the same time as step 46); and -dividing (50) the FFT of the reference first signal by the FFT of the reference second signal (i.e., FFT 1 /FFT2).

In step 52, the determined transfer function can be stored for use in the subsequent on-line control phase.

The steps of the off-line phase of Figure 4 can be carried out using a reference engine that can be provided with accurate and reliable cylinder pressure sensors so that accurate values for the measured cylinder pressures can be generated and the resultant transfer function can be accurately calculated.

Figure 5 is a schematic representation of the off-line calibration phase. H represents the transfer function. As represented in Figure 5, and as described with reference to Figure 4, the FFT of a set of samples of measured pressure signal (P(s)) can be divided by the FFT of a set of samples of measured engine speed (V(s)) signal to compute the transfer function H(s), where H(s) represents the transfer function for the set of samples. In the off-line calibration phase using a reference engine, the cylinder pressure signal P(s) can be determined by sampling the output of cylinder pressure sensors and the engine speed (or crankshaft speed) signal V(s) can be determined by sampling the output of one or more crankshaft sensors simultaneously with the sampling of the cylinder pressure sensors.

Figure 6 is a flow diagram representing the subsequent on-line control phase 70, which is then effected for a production engine that does not then need to be provided with the expensive cylinder pressure sensors. By performing the calibration step on a reference engine, and then using the calibration results on production engines, the cost of each production engine can be kept lower while maintaining the benefits of the accurate calibration performed on the reference engine.

In the present example, prior to the second phase, the transfer function H determined in the first phase is stored 58 in the storage 36 of an EMS of a production engine.

Then in use, the transfer function H is used to provide real time computation of values for the first signal.

The second stage can be considered to be a two stage process. In the second phase 60, the first signal is determined in real time by, in a first stage 62, computing an estimation of the first signal (here the cylinder pressure) using the transfer function and then, in a second stage 70, providing bias compensation of that estimation.

The estimation of the first signal can be computed in the first stage 62 by: -computing 64 an FFT of a second signal (the engine speed of the production engine as sampled in real time); -dividing 66 the FFT computed in step 68 by the transfer function H determined in step 44 and stored in the EMS storage 36; and -taking 68 the inverse fast Fourier transform of the result of step 76.

In the second stage 70, a determination is made at step 72 as to whether a bias compensation factor has been determined.

If a bias compensation factor has not been determined, then in step 74 a bias compensation factor can be computed as described below with reference to Figures 8 and 9 by the EMS unit 32 and in step 76 the bias computation factor is stored in the EMS storage 36.

Where a bias compensation factor has been determined, then in step 78 the bias compensation factor is used to perform bias compensation on the estimation of the first signal computed in step 62. The bias compensation stage is described in more detail in the following with respect to Figures 8 and 9.

The estimation and bias compensation stages 62 and 70 can be performed by the EMS processing logic (e.g., by program code controlling a programmable processor and/or by special purpose processing logic.) Frequency analysis based estimation assumes a SISO (Single-Input-Single-Output) system existing between the instantaneous in-cylinder pressure of an individual cylinder and crankshaft velocity fluctuation, which actually represents the engine dynamics. This is represented in the upper part of Figure 5. The cylinder pressure in the frequency domain can be expressed by equation 1 below: P(S) = H'(S)V'(S) (1) In equation 1, H(S) is the transfer function, or Frequency Response Function (FRF), which is estimated in advance in the off-line calibration phase and stored inside the system. P(S) and V'(S) are the FFT of the pressure and velocity fluctuations respectively.

A table of FRFs can be estimated using a reference engine using a frequency domain estimation according to equation 2 below: H = (P)/(P1,) (2) In equation 2, Pt,,, is the cross spectral density of the velocity fluctuation and cylinder pressure and P is the power spectral density of the velocity function, both of which can be derived from the synchronized measurement of crankshaft velocity fluctuation and cylinder pressure. Non-harmonic and higher unnecessary harmonic frequency components can be rejected from the estimated FRF by setting the amplitude of the unwanted frequency to zero.

In order to eliminate the influence caused by other cylinders and therefore enhance the coherence between the pressure of an investigated cylinder and the engine speed fluctuations, the original sampled data can be applied in a window to preserve only the camshaft angle degree (CAD) range (e.g. less than 180 CAD for 4-cylinder engine) centred round the peak pressure of the investigated cylinder.

The relative cylinder pressure values in the crankshaft degree domain can then be derived by employing the inverse FFT on the previous frequency spectral result from equation I. The lower part of Figure 7 is a schematic representation of an example of the estimation part of the on-line control phase for computing cylinder pressure signals.

As before H(s) represents the transfer function computed in the off-line calibration phase. As represented in Figure 7, and as described with reference to Figure 6, the processing logic of the EMS can be operable to compute the FFT from the samples of a measured engine speed signal V'(S) and to divide the result by the stored transfer function H(s) calculated in calibration phase, and then to take the inverse FFT of the result to find the cylinder pressure signals P(s) in the time domain, P(t).

Figure 8 illustrates the basis for baseline compensation of the estimated pressure. As illustrated in the left hand chart of Figure 8, no-zero order band is identified during the frequency analysis process employed, so that the estimated pressure signal is offset from the actual cylinder pressure signal, as represented at 84 in the right hand chart of Figure 8.

In the described example, in order to generate accurate cylinder pressure values, the bias compensation (step 78, Figure 6) is determined subsequent to an initial on-line estimation.

The calculation of the bias compensation factor (or bias correction) is only required once (in step 72 of Figure 6), and after the bias compensation factor has been determined, it is stored in step 74 in the EMS storage 36 by the EMS unit and is used in the bias compensation step 76, Figure 6. Steps 72 and 74 could be performed as an off-line, or an on-line process.

Figure 9 illustrates the bias compensation factor calculation in more detail.

The aim of the bias compensation is to enable theoretical calibration based on physical phenomena to optimize the baseline in the least-squares sense. Thus, to correct the estimated pressure bias, a theoretical calculation based on the physical model ("PV"=const") is employed. Equation 3 below is used to calculate x(i), the bias error (PbI) as the variable parameter and using the standard deviation to determine the minimum value 92 as represented in the left hand chart of Figure 9.

(P(1)+Pbi) V(i)k = x(i) (3) The value of Pbj corresponding to the minimum standard deviation is then added to the original estimated pressure to realise the bias fitted estimated pressure. In an example represented in the right hand chart of Figure 9 Pbi was calculated at 1 7bar, and the resulting bias fitting pressure can be seen in that chart on the right hand side of Figure 9. In this case, the initial estimated signal 94 is shifted up to the location of the solid line 96 and is similar to the measured signal shown by the dashed line 98.

In one example, the pressure bias or mean value is obtained from preliminary estimation to form the absolute pressure. The bias is determined on the assumption of an ideal adiabatic process for the compression stroke, as described in equation 4 below: (Pm, (a) + tas) * (a)k = const (4) In equation 4, P, is the estimated relative pressure, P is the bias yet to be estimated, Vchambcr is the instantaneous combustion chamber volume, k is the specific heat ratio of the cylinder charge and a is the crankshaft angle.

Different values of P are tried until the standard deviation (STD) of x(n) in equation 5 reaches its minimum, as shown in left part of Figure 9, which means the group of x(n) derived based on such a P is closest to the constant term in equation 4.

(ei('7) + * k'hambe,(11Y' = x(n) n =1..

STD=N'l (_)2) (5) In equation 5, N denotes the number of rei samples selected from compression stroke for the x(n) evaluation. The value of k used in equation 5 is chosen according to the approximate estimation of intake charge composition and temperature, normally around 1.35 for the intake charge. The right part of Figure 7 shows an example of bias correction.

In order to estimate a full 360 CAD range of pressure, the pre-estimated pressure curve needs to be expanded based on the same assumption of an adiabatic process, according to the characteristics of the previously obtained pressure curve, as described in equations 6 and 7 for the compression and expansion strokes respectively.

For the compression stroke, 1c. = x(n) n =1..

COflS(omp = (6) I-(m) = COflSomp / "chambc,(m)' m =1.. M1 For the expansion stroke, s, (chamber(1) = x(n) ..

const = (7) P,,(m) = const / 1'charnber(m) m =1.. M2 In both strokes, the adiabatic const COflStcomp for the compression stroke and COflSte for the expansion stroke need to be evaluated separately by some pre-estimated samples Pesg(fl) and combustion chamber volume VcJmber(fl). Then the expanded range of pressure values Pnew(m) can be estimated.

The optimal value of the parameters used in the equations 6 and 7, e.g. the specific heat ratio kcomp and kexp, can be determined through experiment.

The right hand side of Figure 9 shows an example of the complete estimated pressure curve Pne;v(m) 96 after the bias pressure obtained from the adiabatic equation is included.

During the expansion procedure, a part of the pre-estimated pressure can also be re-estimated in this way, by extending the range parameter m in equations 4 and 5 and Eq.6 into part of pre-estimated CAD range. This has proven effective in reducing oscillations in the pre-estimated pressure caused by interference of the higher harmonic orders.

As a result of the process described with reference to Figures 3 to 9, a good and reliable estimate of the cylinder pressure for each of the cylinders of a production engine can be computed in real time in the EMS of the production without the need for expensive cylinder pressure sensors in the production engine. The torque value for the engine at the flywheel can then be calculated from the cylinder pressure signals for respective cylinders and engine configuration parameters in a known manner.

Optionally, further optimizations of the cylinder pressure signals in the low pressure area of the cycle can additionally be performed. Figure 10 illustrates how possible inaccuracies in the low pressure area(s) 102 (for example as illustrated in the left hand chart of Figure 10), which may be due to a combustion event in another cylinder) can also be corrected using a physical model PVk=const.

In this case, a constant value can be set to equal the average value from one or more windows (e.g. windows 104 and 106 for the intake and exhaust side) of higher pressure where accuracy is better. This constant value can then be applied to the lower pressure area(s), as represented in the right hand chart of Figure 10. This calculation can be performed for both the inlet and exhaust side of the combustion event resulting in the improved estimated pressure signal as shown in the right hand chart of Figure 10.

As described above, the calibration is done off-line using a reference engine. The calibration can be done to provide a single transfer function H(s) that is used for all cylinders. Alternatively, potentially improved accuracy can be obtained is a transfer function H(s) is determined for each cylinder. In this case, each individual cylinder would be calibrated and the respective transfer functions would be stored in the storage 36 of the production engine EMS.

As an alternative to the use of a reference engine, the calibration could be done on a production engine that is provided with low cost cylinder pressure sensors that are know to be accurate when initially installed, but are prone to failure during use. In this case, the calibration could be done on the production engine using the cylinder pressure sensors, but then the estimation process could be used during use of the engine for the estimation of the cylinder pressures rather than relying on the cylinder pressure sensors. As another alternative, cylinder pressure sensors that are integral with special spark plugs could be used, whereby the off-line calibration could be performed on a production engine as part of post production testing, and then the special spark plus could be replaced by conventional spark plugs.

In the above described example, the determination of the bias compensation factor is conducted in a first cycle of the on-line control phase of Figure 6. However, as an alternative, the bias compensation factor determination could be performed as a stage in the off-line calibration phase. This could, for example, be done using a combination of an estimation stage 62 and bias compensation factor determination stage 74 on the reference engine, or a production engine during post production testing, whereby the bias compensation factor could be predetermined and stored prior to the on-line control phase of Figure 6.

In the above described example fast Fourier transform and inverse fast Fourier transform functions are used. However, in other examples, other functions, for example discrete Fourier transform (DTF) and inverse discrete Fourier transform functions could be used.

Accordingly, an example embodiment of the invention can provide a method of engine management in which an engine management system computes an estimation of a first signal representative of a first engine parameter using a second signal representative of a measured engine operation parameter and applies a bias compensation to the estimation of the first signal to determine the first signal representative of the first engine parameter.

A computer program product for implementing the invention can be in the form of a computer program on a carrier medium. The carrier medium could be a storage medium, such as a solid state, magnetic, optical, magneto-optical or other storage medium. The carrier medium could be a transmission medium such as broadcast, telephonic, computer network, wired, wireless, electrical, electromagnetic, optical or indeed any other transmission medium Although the embodiments above have been described in considerable detail, numerous variations and modifications will become apparent to those skilled in the art once the above disclosure is fully appreciated. It is intended that the following claims be interpreted to embrace all such variations and modifications as well as their equivalents.

Claims (23)

1. An engine management system, the engine management system comprising means for computing an estimation of a first signal representative of a first engine parameter based on a second signal dependent upon a measured engine operation parameter, and means for applying a bias compensation to the estimation of the first signal to determine the first signal representative of the first engine parameter.

2. The engine management system of Claim 1, comprising means for determining a bias compensation by varying a bias compensation value until a standard deviation of a constant is minimised, the determined bias compensation then being used to determine the first signal representative of the first engine parameter.

3. The engine management system of Claim 2, comprising means for storing the determined bias compensation.

4. The engine management system of any preceding Claim, wherein first signal is representative of an engine cylinder pressure.

5. The engine management system of any preceding Claim, wherein the second signal is an engine speed signal.

6. The engine management system of any preceding Claim, wherein the means for computing is operable to compute the estimation of the first signal using frequency analysis.

7. The engine management system of any preceding Claim, wherein the means for computing is operable to calculate the Fourier transform of the second signal, to divide the result by a transfer function, and then to take the inverse Fourier transform of the result of the division to compute the estimation of the first signal.

8. The engine management system of Claim 7, wherein the transfer function is predetermined as a Fourier transform of a reference first signal generated by measuring the first engine parameter for a reference engine, divided by the Fourier transform of a reference second signal generated by measuring the engine operation parameter for the reference engine.

9. The engine management system of Claim 8, comprising means for storing the predetermined transfer function.

10. The engine management system of Claim 8 or Claim 9, wherein the means for computing is operable to use the predetermined transfer function to compute the estimation of a first signal.

11. The engine management system of any preceding Claim, comprising processing logic for implementing one or more of the means for computing, the means for applying and the means for determining.

12. The engine management system of any of Claims I to 10, comprising a software-controlled processor and program instructions for implementing one or more of the means for computing, the means for applying and the means for determining.

13. A method of engine management comprising an engine management system computing an estimation of a first signal representative of a first engine parameter based on a second signal dependent upon a measured engine operation parameter and applying a bias compensation to the estimation of the first signal to determine the first signal representative of the first engine parameter.

14. The method of Claim 13, wherein bias compensation is determined by varying a bias compensation value until a standard deviation of a constant is minimised, the determined bias compensation then being used to determine the first signal representative of the first engine parameter.

15. The method of Claim 14, comprising storing the determined bias compensation.

16. The method of any of Claims 13 to 15, wherein first signal is representative of an engine cylinder pressure.

17. The method of any of Claims 13 to 16, wherein the second signal is an engine speed signal.

18. The method of any of Claims 13 to 17, wherein the estimation of the first signal is computed using frequency analysis.

19. The method of any of Claims 13 to 18, wherein the estimation of the first signal is computed by calculating the Fourier transform of the second signal, dividing the result by a transfer function, and then taking the inverse Fourier transform of the result of the division.

20. The method of any of Claims 13 to 19, wherein the transfer function is predetermined as a Fourier transform of a reference first signal generated by measuring the first engine parameter for a reference engine, divided by the Fourier transform of a reference second signal generated by measuring the engine operation parameter for the reference engine.

21. The method of Claim 20, comprising pre-storing the predetermined transfer function.

22. The method of Claim 20 or Claim 21, wherein the estimation of a first signal is computed using the predetermined transfer function.

23. A computer program product comprising program instructions operable to control a processor of an engine management system to performing the method of any of Claims 13 to 22.