Procedure for controlling 1C engines
This invention relates to the provision of a means of providing linear control of a non-linear system. In particular it relates to the provision of a system and method for removing the non-linear nature of the torque control problem in spark ignition internal combustion (gasoline) engines and for allowing the control of engine efficiency as well as torque. More specifically there is described a method of using a set of signal manipulation techniques at the (signal) input and output of the gasoline engine that allows the use of standard linear control techniques to accurately control both engine torque and efficiency (or, equivalently, torque reserve) .
Technical Background: Traditionally, the torque output of a spark ignition gasoline engine has been controlled directly by the driver moving the throttle in the air intake path, allowing more or less air to be inducted into the
cylinder with each induction stroke. The function of the carburettor or fue . injection system being to add fuel to the air such that che in-cylinder combination has a fixed mass ratio. The spark timing would be controlled (mechanically or electronically) with the goal of obtaining the maximum torque possible from each combustion event.
More recently, it has become commonplace to move the spark timing away from that which gives the optimal torque production in certain engine operating modes. Most particularly the idle mode, a more complete discussion of this application is given further in this document. In order for the operating efficiency to be reduced whilst still producing the same overall net torque from the engine, the airflow into the engine must be increased. For the engine controller to do this there must either be an electronically controlled parallel airpath into the engine or an electronically controlled main airpath into the engine. Prior art techniques, whilst explaining the various advantages of this reduction in efficiency, have not explained explicitly how to achieve this seamlessly in practice. Practical realisation of the reduction in efficiency is straightforward to achieve if the desired change in efficiency is slowly changing and the engine is at one operating point, however if the desired change is quickly varying or if the operating point of the engine is rapidly changing then prior art methods are very difficult to implement because of the nonlinear nature of the engine system. This is the case for the three examples given in this document, the techniques described herein enabling greatly improved control in these cases. The following section expands on the particular
requirements and difficulties of the idle control problem.
The control of the speed of gasoline engines when operating in the idle condition whilst keeping the rate of fuel usage of the engine as low as possible presents a number of difficulties. The speed of a gasoline engine at idle is affected by many forms of disturbance; cyclic variability m torque production, varying friction characteristics of the engine and mechanical loads from accessories such as the alternator, power steering pump and air conditioning pump. The magnitude of the accessory disturbances m particular has increased as the level of standard equipment in vehicles increases . Thus on a cold morning a driver may enter his/her car, start the engine, and with the engine m idle condition, switch on the front and rear electrical demisters, move the electric chair to a more suitable condition, turn on the electric chair heater and then make several lock-to-lock steering movements to get out of the parking spot. The power demand of the steering is at its highest when the vehicle is stationary. All of the energy drains must be countered by an increase m torque from the engine but without perceivable changes m idle speed. To reject the effect on speed of these disturbances, modern engines use electronic engine control units and feedback control techniques to keep the speed constant. Whilst m idle mode, the airflow into the engine is controlled by the engine management system using a secondary throttle, (also known as an idle speed control valve (ISC) or air bypass valve (ABV) ) m parallel with the primary, driver controlled throttle, or an electronically controlled main throttle. Because the speed of response of this actuator is slow (due to the time for the effect of the change m
secondary throttle, or electronic main throttle, position to propagate through the manifold and affect the air inducted into the cylinder) , the spark timing is also used. It being retarded from the point of optimal torque production to a point that allows some torque reserve which can be available for almost instant rejection of torque disturbances. The further away from the timing for optimal torque production that the engine is operated at, the more torque reserve the engine/controller has but the more fuel is used to keep the engine at the desired speed. Methods for controlling the speed use a measurement of speed and control the secondary throttle or electronic main throttle, and spark timing. The effectiveness of the idle control mechanism will dictate the idle speed set point of the engine. A poor controller will have to have a high idle set point so that low side excursions do not cause the engine to stall. A figure used by the industry is that every lOOrpm decrease in idle speed setpoint is approximately equivalent to one mile per gallon improvement in fuel economy. Note again that a poor idle controller will have to operate far from optimal spark timing to achieve the torque reserve necessary for stable idle. This alone can increase the fuel used during idle by a factor of 2 or 3. Clearly an effective controller which can operate close to optimal spark timing and at a lower speed will make significant fuel economy improvements.
Prior art techniques have traditionally relied on simple linear controllers (typically of the proportional + integral type) to control the spark and bypass valve in response to a speed variation. These controllers have to be calibrated at many different operating points so that although the controller structure is simple, the
parameters that the controller uses will be based on large look-up tables and the control action will have many special case responses which again are the result of many man months of calibration effort. The reason for the inadequacy of this linear technique is that the system itself is very non-linear. If the speed of the engine drops as a result of an unanticipated load, then the effect of the next spark will be different because we are now at a different operating point, less torque will be generated and thus the size of the excursion will increase. Many researchers have tried more complicated linear control techniques which have the advantage that they may overcome some of the effects of the nonlmeaπty and the disadvantage that they are generally significantly harder to calibrate for the areas where a variation in controller parameter is necessary. Conversely, some non-linear techniques are available which will enable the control engineer to take account of some or all of the non-linear characteristics of the engine when designing the controller. However these non- linear techniques are very cumbersome m the design stage so that designing a controller is possible but designing a good controller is very difficult.
In a more general form the technique for making linear the control of torque and operating efficiency (equivalently, torque reserve) could be used m other engine operating modes. Two examples are:
1) m the transition from homogeneous to stratified charge mode m direct injection gasoline engines it will be desirable to increase the manifold pressure before the switch without increasing the torque. This can be done by reducing the efficiency of the
combustion which is easily effected using this technique
2) When preparing tor the deactivation of one or more cylinders in more general engine types it is desirable to increase the torque reserve so that when the one or more cylinders is/are deactivated, the remaining active cylinders can rapidly increase their torque to keep the overall torque output of the engine at the desired level. This functionality is easily achievable using the technique presented here.
In both of these examples the required change in efficiency is rapid, existing techniques being very difficult to implement successfully in these cases.
It is an object of the present invention to provide a control system and method that allows the use of standard linear control techniques to control an essentially non- linear system such as a gasoline engine.
It is a further object of the present invention to improve a gasoline engine's response to load disturbances in the idle condition.
It is a still further object of the invention to provide a means of varying the operating efficiency of a gasoline engine in order to increase the thermal energy output of the engine in specified circumstances.
It is a still further object of the present invention to provide a control system and method which requires significantly less calibration time in the development stage than prior techniques.
According to the present invention there is provided a control assembly for a non linear system comprising non- linear map interface means connecting input and output points of the non linear system to a linear control arrangement .
The non-linear system may, for example, be a spark ignition gasoline engine.
The control assembly may be the idle speed control system or engine management system of the gasoline engine.
Controlled inputs to the engine are preferably spark timing and secondary throttle, or electronic main throttle, settings and measured engine outputs include engine speed and manifold pressure.
Further according to the present invention there is provided an internal combustion engine management system which provides control of engine input parameters m accordance with an anticipated load disturbance resulting from the operation of engine auxiliary components m advance of the application of the said load.
Preferably, but not necessarily, the engine management system controls both the engine input parameters and the load disturbing components such that engine output torque may be increased m accordance with increased load application.
Still further according to the present invention there is provided an internal combustion engine management system arranged for allowing the operation of the engine at
close to optimal efficiency m the idle condition, the system being further arranged so as to decrease the operating efficiency of the engine and increase the thermal output of the engine m predetermined conditions.
Increased thermal output may be required m a cold start situation where it is desirable to achieve normal operating temperature as quickly as possible.
Alternatively, increased thermal output may be required m a cold start situation to allow the catalyst to reach operating temperature as quickly as possible.
Alternatively increased thermal output may be required m order to maximise the heat output of an associated heating system.
Embodiments of the present invention will now be described by way of example with reference to the accompanying drawings :
Figure 0 is a general representation of a method for enclosing a non-linear engine system using the method described herein to produce a resulting system which is linear or almost linear from input to output; Figure la is a diagram illustrating the effect of spark timing or torque production for a given engine speed and manifold pressure (or, equivalently, air flow) ; Figure lb is a diagram illustrating the map for spark advance required to give desired torque production for a given engine speed and manifold pressure (or, equivalently, air flow) ;
Figure 2a is a representation of the effect of the secondary throttle position on the manifold air pressure and thence the distance from peak torque (at any operating point) ; and Figure 2b is a representation illustrating the use of the inverse map for representing the non-linear part of the manifold dynamics, such that the second input-output pair also becomes a linear system (at any operating point) .
Disclosure of the invention: The present invention overcomes the shortcomings described above of prior art methods by utilising a control scheme that includes signal manipulation techniques to make a more easily controllable system. The method, as illustrated in Figure 0, creates an overall system which is not only linear, or almost linear, (depending on implementation) from (new) inputs to outputs, but also creates another syntnetic output which allows the direct on-line control of the efficiency, or equivalently torque reserve, of the engine. Thus by 'wrapping' the engine using the method described, the engine will appear linear, or almost linear, and intuitive at all operating points likely to be met in the idle regime. This has the advantage of allowing advanced linear control techniques to be used, which can extract more performance from the controller for less calibration time. The nature of the interfacing method is such that the parameters which characterise it can be identified from a small number of test operations taking, compared to prior art techniques, a vastly reduced amount of testing time. The second output (MBT-T) being a measure of efficiency of the engine allows the engineer to design controllers which will allow very efficient operation
when few disturbarces are anticipated, and quickly move to a less efficient point when load disturbances are imminent. Alternatively it enables smooth transitions between engine ope.ι ating modes by matching the torque production either side of the transition.
More specifically, the torque produced by the combustion event of a S.I. engine (assuming constant air-fuel ratio) is a static function of the amount of charge in the cylinder, the engine speed (spd) and the spark timing, or equivalently, the manifold pressure (MAP) at the end of the induction stroke, the engine speed and the spark timing. This is illustrated in figure la. If the spark is not allowed to go beyond the point of optimal torque then the torque map can be inverted as suggested in figure lb. This inversion allows the engine management system to control the torque production from the combustion event directly, the relationship between torque and speed being a linear one so the control design problem is significantly easier.
More specifically, the spark timing, in conjunction with a measurement of the manifold pressure at the end (or some other portion) of the induction stroke of the cylinder for which the next combustion event occurs, and an estimate of the engine speed when the combustion event occurs, can be set based on an inversion of the static map, to give any desired combustion torque (Tcom) over the next combustion event within the limits of the map.
In this new system, the use of the second input (ssMAP) is effectively to control the amount of torque reserve of the engine, the first input can immediately demand a torque increase and it will achieve it (within the
limitations imposed by the static map) , however the second input must always be operating such that the engine has the required amount of spare torque capacity available. The distance, MBT-T (m units of torque) that the engine is operating from its peak torque production is again a static function of the engine speed, the spark timing and the manifold pressure, or equivalently the engine speed, the manifold pressure and the combustion torque. I.e. MBT-T = f (speed, TLumt, MAP). This static relationship may be linear m the variables or if not will be smooth enough to allow linearisation for design purposes. Note that since this path is not critical to the cycle by cycle behaviour of the system, the effect of any linearisation errors will be small. The manifold pressure itself behaves in a non-linear dynamic way affected by the secondary throttle, or electronic mam throttle, and the engine speed. This non-linear dynamic behaviour can be separated into a linear dynamic portion (when measured m the discrete, event based domain) and a non-linear static map, the relationship between the secondary throttle position and the distance from peak torque being illustrated figure 2a. The non-linear static map can again be inverted and when the inversion is used, the second output is now a linear function of the second input (termed ss_MAP here), and the first input and output. This is illustrated in figure 2b.
Detailed description of the preferred embodiments.
Since the combustion process is itself an inherently event based process, the natural framework for a high performance idle speed controller is m the (combustion) event based discrete time domain. The following describes
the engine model formed as a result of the use of the method described. 1) Since the combustion torque is a function of the spark (which is set at least one control event prior to the combustion event) and the manifold air pressure at the end of the induction stroke, delays must be applied to each input. For a four-stroke, four cylinder engine, the manifold pressure delay is one unit. 2) The manifold pressure dynamics can be assumed to be first order with fixed coefficient (although this method would also include higher order models) 3) The engine speed then follows from simple Newtonian mechanics. 4) The second output is a linear combination of the delayed (desired) combustion torque, the delayed manifold pressure and the speed.
The model for a four-stroke, four cylinder engine will thus typically have four states, the delayed Tcomb, the delayed ss_MAP, the manifold pressure and the engine speed. The technology described here can also be used to control engines with more or less cylinders, and with faster or slower sampling rates by changing the model in a way which will be clear to experienced practitioners. The model description given above does not preclude the use of the method described in a continuous time based design.
When the controller so designed using the modified engine system as described above is implemented, it would be usual, although not necessary, to substitute more accurate estimations of the torque map and actual readings of the manifold pressure to calculate the second output .
The values in the non-linear maps can be obtained by operating the engine in steady state at various operating points and measuring the torque produced at that point. For the secondary throttle, or electronic main throttle, to manifold pressure relationship, the secondary throttle, or electronic main throttle, can be opened in step increments and the behaviour of the manifold pressure recorded. Regression curves, look-up-tables or other mapping techniques can be used to perform the non- linear mappings.
The use of this method facilitates the use of several other techniques with an ease that was not available with prior art techniques. These include: 1) The use of feedforward torque estimates for known load disturbances. Since the input to the augmented system is combustion torque, if the load torque of a disturbance can be estimated, it can be cancelled by simply adding this estimate to the first system input. This eliminates the response time inherent in feedback systems and allows a potentially significant improvement in controller performance.
2) If the load disturbance can be delayed by a small amount (e.g. the air-conditioning is under the control of the engine management system and can easily be delayed by say 0.5 seconds before the pump is engaged) then the torque reserve of the engine can be quickly increased by changing the reference on the second input. Then when the load is engaged there is a lot of reserve torque available to reject the disturbance. Soon (e.g. 0.5seconds or less) after the load has been engaged the operating point can be moved back to a more
fuel efficient one. Items 1) and 2) can of course be utilised togeth ?r to get. the benefits of both.
3) Sometimes it is desirable to spend time with the engine efficiency decreased significantly from optimal, by the nature of the method described above, this is trivially easy to achieve. When the efficiency is reduced, the 'wasted' energy becomes thermal energy this may be desired for other purposes. Two examples of this are i) After the engine has been started from cold it is desirable to heat the catalytic converter up to its normal operating temperature as quickly as possible. By reducing the efficiency of the engine, there is more thermal energy in the exhaust gas that is transferred in part to the catalytic converter. ii) When idling in cold climates, the energy being transferred to the engine block may not be sufficient to keep the heater matrix hot enough to keep the passenger compartment comfortable. Reducing the efficiency of the engine increases the amount of thermal energy transferred to the engine block and hence the heater matrix. iii) When starting from cold, it may be desirable to transfer thermal energy rapidly to the engine block in order to quickly reach normal operating temperatures and to enable rapid heating of the heater matrix, facilitating fast warm-up of the passenger compartment and defrosting/demisting system.
It will be appreciated that the ability of the system to compensate for a wide range of potential load disturbances allows the running of the engine far closer
to optimal conditions for more of the time than with previous techniques. At the same time sub-optimal operation is also possible where there is a requirement to do so .
Modifications and improvements may be made without departing from the scope of the present invention.