DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0044] An estimating method of a NOx occlusion amount according to one embodiment of the present invention will be explained below with reference to the drawings. FIG. 1 is an illustrative block diagram showing the whole constitution of an exhaust system of a vehicle to which the present invention is applied.

[0045] As illustrated, a NOx occlusion catalyst 1 is interposed in an exhaust passage 2, and a NOx sensor 3 which detects NOx concentration at a catalyst outlet is provided downstream of the NOx occlusion catalyst 1.

[0046] Further, though not illustrated, an O₂ sensor, a temperature sensor, an air flow sensor (AFS), and the like are connected to the exhaust passage 2, and oxygen concentration, exhaust gas temperature and exhaust gas flow rate in the exhaust passage 2 are respectively detected by these sensors.

[0047] Furthermore, each sensor 3 is connected to an ECU 4 serving as a control unit. Here, the ECU 4 is provided with an input/output unit, a storage unit (a ROM, a RAM, a non-volatile RAM or the like), a processing unit (CPU), a timer counter and the like, and estimation of a NOx occlusion amount in the NOx occlusion catalyst 1 is performed by the ECU 4.

[0048] Further, various maps are provided in the ECU 4, where respective concentrations of NOx, CO and HC at the catalyst inlet (hereinafter, referred to as catalyst inlet [NOx], catalyst inlet [CO] and catalyst inlet [HC], respectively) are readout from the maps by utilizing, for example, the acceleration opening degree or the number of rotation of the engine as parameters. In this embodiment, these values are obtained from the maps, but these values may be directly detected from the various sensors provided in the exhaust passage 2.

[0049] Now, the above-described second conventional art is constituted such that the characteristics themselves of the NOx occlusion catalyst are described with a mathematical expression to obtain a NOx occlusion amount, meanwhile the present invention is constituted such that a NOx occlusion amount is calculated by using a quartic linear polynomial.

[0050] That is, the polynomial as shown below in the equation (1) is stored in the ECU 4, and a NOx purification rate and a NOx occlusion percentage in the NOx occlusion catalyst 1 are calculated on the basis of the polynomial. 1$\begin{array}{cc}r=f\left(x,y,z\right)& \left(1\right)\\ =k_0+k_1x+k_2y+k_3z+k_4\mathrm{xy}+k_5\mathrm{yz}+k_6\mathrm{zx}+k_7x^2y+k_8{\mathrm{xy}}^2+\dots & \end{array}$

[0051] In the equation (1), r is a NOx purification rate, x is a NOx occlusion percentage, y is an exhaust gas temperature, z is a SV value (or an exhaust gas flow velocity), and k₀, k₁, k₂, . . . are coefficients.

[0052] Here, the NOx occlusion characteristics of the NOx occlusion catalyst 1 can be known by such as an experiment in advance, and it can be approximated with a polynomial. For example, if there are three parameters (cubic), the characteristics of the catalyst can be represented with a curved face as shown in FIG. 2, and it can be expressed by a polynomial. Of course, since the equation (1) is quartic, the characteristics does not form such a curved face as shown in FIG. 2, but the same concept as the cubic case can be applied.

[0053] Further, each coefficient ki (i=1, 2, . . . ) is input with a proper value as an initial value on the basis of the characteristics preliminarily obtained from experiments or the like.

[0054] Then, the latest catalyst condition can always be represented according to the equation (1) by correcting each coefficient ki sequentially on the basis of detected values of the NOx sensor 3 downstream of the catalyst during engine operation.

[0055] Hereinafter NOx estimation means will be explained in further detail. First, the NOx purification rate r in the equation (1) can also be obtained by the following equation (2).

r=NOx concentration at catalyst outlet/NOx concentration at catalyst inlet (2)

[0056] Here, the NOx concentration at catalyst outlet is a value detected at the NOx sensor 3, and the NOx concentration at catalyst inlet can also be obtained from a map. Therefore, the NOx purification rate r obtained by the equation (2) can be said to be a sensor value (an actually measured value), while the NOx purification rate r obtained by the equation (1) can be said to be an estimated value (a calculated value).

[0057] Further, actually measured data or map values are applied to the exhaust gas temperature [y] and SV value [z] in the equation (1). Thereby, the unknown value in the equation (1) is only the NOx occlusion percentage [x] now, while other values except this are known. Accordingly, the NOx occlusion percentage [x] can be obtained by the following equation (3), which is modification of equation (1).

x=[r−(k₀+k₂y+k₃z . . . )]/(k₁+k₄y+ . . . ) (3)

[0058] Then, the NOx occlusion percentage [x] obtained by the equation (3) is returned back in the equation (1) again, and it is used for calculation in the next calculation cycle. That is, in the equation (1), the purification rate r is calculated by using the NOx occlusion percentage [x] obtained the previous time, and newly detected [y] and [z].

[0059] Here, the purification rate r calculated by the equation (1) is an estimated value, but when each coefficient k_i is an accurate value, the estimated value of the purification rate r and the actually measured value obtained by the equation (2) should coincide with each other.

[0060] Therefore, in this invention, the estimated value and the actually measured value of the purification rate r are compared with each other, and when there is a difference between the two, each coefficient k_i is sequentially corrected by using the method of least square so that the estimated value coincides with the actually measured value. Then, the polynomial of the equation (1) can be modified at any time to the equation reflected with the state of the NOx occlusion catalyst 1 accurately by repeating such calculations to correct sequentially each coefficient ki.

[0061] That is, though the characteristics of the NOx occlusion catalyst 1 will vary according to its operating situation, its age deterioration or the like, the equation (1) is kept to serve as an equation representing the state of the catalyst 1 accurately by means of repetitive corrections of the coefficient ki, hence the NOx occlusion amount can be estimated with a high accuracy. Incidentally, the NOx occlusion percentage [x] obtained according to the equation (3) is an instantaneous NOx occlusion amount in the calculation cycle of the initial stage, but an accumulated value of the NOx occluded in the catalyst 1 can be obtained by returning the calculation result to the equation (1) repeatedly to calculate the purification rate r.

[0062] Thus, during lean operation where NOx is occluded, the catalyst outlet [NOx] is captured from the NOx sensor provided downstream of the catalyst and the captured value and the estimated value according to the equation (2) are compared with each other, then an accurate NOx occlusion amount can be obtained by correcting the coefficient ki by the method of least square for each comparison.

[0063] By the way, since NOx is discharged during rich operation, unless an accurate NOx discharge amount can be detected or estimated, an initial value of the NOx occlusion amount at the next lean operation cannot be calculated accurately.

[0064] Therefore, the coefficient ki is held during the rich operation and the NOx discharging amount is calculated according to the following equations (4) and (5), and the NOx occlusion amount during rich operation is calculated by subtracting the NOx discharging amount calculated according to the equations (4) and (5) from the NOx occlusion amount just before the start of the rich operation.

NOx discharging amount=∫ (reducing agent concentration at catalyst inlet×reducing agent utilization rate [r′]−0.5×catalyst inlet [O₂])×exhaust gas flow rate (4)

Reducing agent utilization rate [r′]=f(y, z) (5)

[0065] Here, a reducing agent utilization rate setting map as shown in FIG. 3 is provided in the ECUL. The characteristics of the reducing agent utilization rate based upon the exhaust gas temperature [y] and the SV value [z] are stored in the map, so that the reducing agent utilization rate is set on the basis of the parameters [y] and [z].

[0066] Further, the catalyst inlet reducing agent concentration (catalyst inlet [CO] and catalyst inlet [HC] ) can be obtained from the map utilizing the acceleration opening degree or the number of engine revolution as parameters, and other parameters can also be obtained as map values or detected values.

[0067] Therefore, when switching is preformed from the lean operation to the rich operation, each coefficient ki at the time of lean operation termination is held and the NOx occlusion amount is stored in the ECU4, and when switching is performed from the rich operation to the lean operation again, the remaining amount of NOx is calculated by subtracting the NOx discharged amount at the time of rich operation termination from the NOx occlusion amount.

[0068] Moreover, when a difference between an actually measured value and an estimated value of the NOx purification rate r at the time of lean operation start is equal to or more than a threshold value, the NOx occlusion amount is corrected in the ECU 4. That is, when a difference between the NOx purification rate estimated according to the equation (1) and the NOx purification rate obtained according to the equation (2) immediately after switching has been performed from the rich operation to the lean operation is equal to or more than a predetermined value, judgement is made that the NOx remaining amount calculated at the starting time of lean operation (=the NOx occlusion amount at the time of the latest lean operation termination -the NOx discharged amount at the time of the rich operation termination) is incorrect, and the NOx remaining amount [x] (that is, the NOx occlusion amount at the time of lean operation start) is corrected on the basis of the equation (3).

[0069] In addition, the ECU 4 has such a function that, when the coefficient ki becomes an abnormal value due to an accidental irregularity of the catalyst 1 or the like, the ECU 4 detects such a fact and notifies it to a driver. Specifically, regarding each coefficient ki, moving averages in a predetermined continuous calculation cycle are calculated constantly and when the value of a moving average is deviated from the predetermined range set for each coefficient, an abnormality judgment is made such that deterioration or damage has occurred in the catalyst 1. Incidentally, the predetermined range is established by adding ±α (α is a constant value) to an initial value of each coefficient ki, for example.

[0070] Further, when the ECU4 judges that the catalyst 1 is abnormal, an alarming lamp on the instrument panel, for example, is turned on and the rich operation is inhibited.

[0071] Thus, inhibiting the rich operation at an abnormal occasion of the catalyst 1 can avoid an event like CO discharge in advance properly.

[0072] Since the estimating method of a NOx occlusion amount according to one embodiment of the present invention is constituted as described above, the NOx occlusion amount is estimated in the following manner.

[0073] In a lean operation, first, the actual NOx purification rate r is calculated according to the equation (2) and the exhaust gas temperature [y] and the SV value [z] are obtained from the actually measured data or the map values. Also, in the polynomial of equation (1), a suitable value obtained by experiments or the like is inputted for each coefficient k_i as an initial value. Then, each value is substituted for the equation (3) obtained by modifying the equation (1) to calculate the NOx occlusion percentage [x].

[0074] Next, the NOx occlusion percentage [x] obtained according to the equation (3), the newly obtained exhaust gas temperature [y] and the SV value [z] are substituted for the equation (1) to obtain a NOx purification rater (an estimated value). Then, the estimated NOx purification rate r and the actual NOx purification rate r newly calculated according to the equation (2) are compared with each other, and each coefficient ki is corrected by the method of least square so that the estimated NOx purification rate is coincident with the actual NOx purification rate.

[0075] Then, such calculations are performed repeatedly to update each coefficient k_i at any time, which makes the equation (1) to be an expression representing an accurate state of the catalyst 1, and realizes that the NOx occlusion amount can be estimated with a high accuracy. Further, the accumulated value of the NOx occluded in the catalyst 1 can be obtained by repeated calculation of the NOx occlusion percentage [x] using the equation (3).

[0076] In addition, as NOx is discharged during a rich operation, the calculation of the NOx occlusion amount according to the equation (1) is suspended at this time and each coefficient ki is stored, meanwhile the NOx discharging amount is calculated according to the equations (4) and (5) described above.

[0077] Then, when switching is performed from rich operation to lean operation again, the NOx remaining amount at the time of lean operation start is calculated by subtracting the NOx discharged amount at the time of rich operation termination from the NOx occlusion amount at the time of lean operation termination.

[0078] Further, when a difference between the actually measured value and the estimated value of the NOx purification rate r at the time of lean operation start is equal to or more than a threshold value, the NOx occlusion amount is corrected. Here, the correction means for the NOx occlusion amount will be explained with reference to a flow chart shown in FIG. 4. First, each coefficient k_i of the polynomial (1) is sequentially corrected during the lean operation to identify each coefficient k_i (Step S1).

[0079] Then, when it is judged that a rich operation has started (Step S2), each coefficient ki is held (Step S3). Thereafter, when the rich operation is terminated (Step S4), an actually measured value and an estimated value of the NOx purification rate are compared with each other and judgement is made whether or not a difference between the both is equal to or more than the threshold value (Step S5).

[0080] When the difference between the actually measured value and the estimated value of the NOx purification rate is less than the threshold value, returning back to Step 1 again, and updating each coefficient k_i in the lean operation is continued, while the difference is equal to or more than the threshold value, the NOx occlusion amount is corrected to a value calculated according to the equation (3) (Step S6) and the flow returns.

[0081] Thus, the estimation accuracy of the NOx occlusion amount can further be enhanced by correcting means of the NOx occlusion amount in this manner.

[0082] On the other hand, when deterioration or irregularity of the catalyst 1 is judged, the judgement is notified to the driver by turning on the alarm lamp in the instrument panel and the rich operation is inhibited. This procedure will be explained according to a flowchart shown in FIG. 5. Each coefficient k_i of the polynomial (1) is sequentially corrected and identified (Step S11), and the moving average of each coefficient k_i for a fixed period is then calculated (Step S12).

[0083] Next, judgement is made about whether or not the moving average of each coefficient k_i calculated falls with in a predetermined range (Step S13). When the moving average of any coefficient is within a predetermined range, the flow returns back to Step S2 again. However, when at least one of the moving averages of the coefficients k_i is out of the corresponding predetermined range, judgement is made that abnormality has occurred in the catalyst 1 to give an alarm (Step S14) and inhibit the rich operation (Step S15).

[0084] As described above in detail, according to the estimating method of NOx occlusion amount according to the embodiment of the present invention, since the NOx occlusion amount is estimated by using the polynomial reflected with the NOx occlusion characteristics of the NOx occlusion catalyst 1 and each coefficient k_i of the polynomial is sequentially corrected on the basis of the actually measured NOx purification rates, there is an advantage that the polynomial which is modeled consistently with the latest catalyst state can be obtained and the NOx occlusion amount can be estimated with high accuracy. Further, by estimating the NOx occlusion amount accurately, the optimal rich operation control can be conducted on the basis of the NOx occlusion amount and fuel consumption can be improved.

[0085] Further, even if the NOx occlusion catalyst 1 is changed, since updating to each coefficient ki corresponding to the characteristics of the changed catalyst 1 is conducted, there is such an advantage that modification of the model equation is not required. On the other hand, when a mathematical model equation is applied as the conventional art, there occurs such a problem that works (fitting) for identifying coefficients corresponding to the characteristics of each catalyst is needed, which requires tremendous labor. In the present invention, however, the NOx occlusion amount can be estimated accurately without conducting such fitting works.

[0086] Since the NOx discharging amount is calculated on the basis of the reducing agent utilization rate, the NOx discharged amount can be estimated with a relatively high precision.

[0087] Furthermore, the coefficient k_i of the polynomial is held during rich operation, and when the difference between the NOx purification rate estimated by using the coefficient held at the time of lean operation start and the actually measured NOx purification rate is equal to or more than a threshold value, the NOx occlusion amount is corrected, therefore the estimation precision of the NOx occlusion amount can further be enhanced.

[0088] When the average value of each coefficient k_i for a predetermined period is deviated from a predetermined range, it is judged that the catalyst 1 is abnormal, therefore the abnormality of the catalyst 1 can be judged with a high precision. Further, since an alarm is sent out to the driver at the time of such abnormality judgement, the driver can recognize the abnormality of the catalyst 1 immediately, which promotes the driver to exchange catalysts 1 early. Further, since the rich operation is inhibited in case of such an abnormality of catalyst 1, the event that CO is exhausted as it is can be avoided.

[0089] Incidentally, the estimating method of a NOx occlusion amount of the present invention is not limited to the above-described embodiments, but it may be modified in various ways within the range not deviated from the scope and spirit of the invention. For example, the discharging amount of NOx can be obtained from an approach similar to the estimating method of a NOx occlusion amount.

[0090] Namely, though the reducing agent utilization rate [r′] used for calculating the NOx discharged amount is obtained from the reducing agent utilization rate setting map shown in FIG. 3, it maybe obtained according to a polynomial of catalyst inlet reducing agent concentration [x′], exhaust gas temperature [y] and SV value [z] as shown in the following equation (6). 2$\begin{array}{cc}\begin{array}{c}{r}^{\prime }=f\left(x^{\prime },y,z\right)\\ =m_0+m_1x^{\prime }+m_2y+m_3z+m_4x^{\prime }y+m_5\mathrm{yz}+m_6{\mathrm{zx}}^{\prime }+m_7{x^{\prime }}^2y+m_8x^{\prime }y^2+\dots \end{array}& \left(6\right)\end{array}$

[0091] Here, m_i (i=1, 2, . . . ) are coefficients. Further, the equation (6) is a polynomial reflected with NOx discharging characteristics of the NOx occlusion catalyst, where each coefficient m_i is inputted with a proper initial value obtained from experiments or the like as an initial value similarly to the equation (1).

[0092] In this case, additionally, a sensor (a CO sensor) for detecting a reducing agent (specifically CO) concentration is provided at least in the exhaust passage 2, and the reducing agent utilization rate r′ is obtained by sequentially updating each coefficient m_i from actually measured values of the reducing agent concentration [x′] obtained from the CO sensor and estimated values of the reducing agent concentration [x′] obtained according to the equation (6), and then the NOx discharged amount can be obtained by substituting the value thus obtained for the above-described equation (4).

[0093] The estimation precision of the NOx discharged amount can be improved by obtaining the NOx discharged amount in this manner.