Background technique
Generally be widely used on the market and belong to comparatively advanced internal-combustion engine (being engine), mostly be equipped with an engine control unit (Engine Control Unit, ECU), in order to crank angle, rotating speed, the acceleration that reads correlation engine ... or the like basic signal.Wherein the crank angle signal can provide the control of ignition timing and fuel oil timing, and speed of crankshaft and acceleration signal then can be indicated torsion and carry out fault diagnosis in order to predicting engine.And know that ECU is by the synchronous operated flywheel of bent axle side one energy, by the crank angle perceptron that is provided with on the flywheel, to detect above-mentioned crankshaft operation signal.But, cause result of calculation and unreliable because the signal of perceptron is subjected to the interference of extraneous noise easily.
Handling and calculating in the prior art of bent axle signal, including:
A kind of position lex fundamentalis (Position-Based), need learn number of teeth signal more on the flywheel (for example 180 or 360 teeth), just can obtain a more accurate CAP, the also therefore bigger ECU Processing capacity of demand, to avoid too much causing program to take place when the computing to interrupt or the problem of memory space deficiency because of number of teeth signal, but the method equipment needed thereby cost is higher, so favourable puzzlement of using.
Another kind of time basis method (Time-Based), can cooperate less number of teeth signal (for example 1,4,24 or 32 teeth), learn the position of engine crank corner with calculating, so be difficult for taking place the problems referred to above, therefore based on economically consideration, more in the industry this method of employing is to carry out the crankshaft rotating dynamic pre-estimating of engine, to obtain signals such as crank angle, rotating speed and acceleration.But when noise was got involved the bent axle signal, the method easily produced wrong estimation results, how to prevent that therefore noise from disturbing, and just becomes the matter of utmost importance of handling the bent axle signal.
In order to prevent that noise from disturbing the bent axle signal, a kind of low-pass filter (Low Pass Filter that utilizes is arranged in the industry, LPF) filter the method that noise disturbs in the crank angle perceptron, when passing through the back signal of being exported of LPF calculating greater than predefined value with convenient bent axle signal, wave filter can be exported a voltage signal immediately; But this wave filter can produce the effect of delay, and when noise was excessive, wave filter then can be used as actual signal output to this noise, and the ability of predictor error shows deficiency.
In addition, the inventor announces in the I243904 patented technology in Taiwan, also disclose and a kind of Kalman filter of using is arranged to estimate the method for crank angle and rotating speed, comprise and contain the Kalman filter of estimating equation and observing matrix one, be installed in the rotation dynamic pre-estimating system of engine, and utilize the crank angle θ that an above-mentioned circuit signal processing system measured, the engine torque discreet value that an engine torque Prediction System is learnt
And the load torsion discreet value learnt of engine torque load Prediction System
, and then estimate the corner and the rotating speed of bent axle, when disturbing Crankshaft sensor, can reduce influence for engine control with convenient noise.But the method need be used the rotatory inertia and the coefficient of viscosity parameter of engine, and when this parameter had error, what will have influence on wave filter estimated accuracy and system's robustness.
Moreover, relevant pole assignment and Xie Li card are carried (Riccati) equation etc. and are obtained the computational methods of feedback gain value matrix, and the method for differentiating the crankshaft operation stroke, though belong to disclosed previous number reason computing, but Shang Weijian reaches in the estimation of the loop circuit of engine crank corner and rotating speed and is employed on the dynamic sample signal of auxiliary processing bent axle, and gives Chen Ming.
Summary of the invention
The objective of the invention is to estimate the crank position that according to the traditional time basis method that measures, carry out the estimation and the correction of loop circuit, particularly disturb under the bent axle signal situation that perceptron detected, can obtain comparatively accurate crank angle and rotating speed at extraneous noise.And, the present invention need not by air-distributor pressure signal and engine parameters, get final product the estimation of March Shaft angle and rotating speed, so can effectively reduce the distrubance source that extraneous noise may cause, also can reduce simultaneously the load of estimation device processes signal, and then the reduction equipment cost, and can directly be applied on the different engines.
For achieving the above object, technology contents of the present invention comprises use:
One signal processing unit picks up counting from the engine start seasonal timer that turns round, reading the crank teeth news that perceptron detects, and the time difference of reading by timer between each tooth news, and calculating obtains speed of crankshaft ω
kWith acceleration alpha
k, and after above-mentioned timer makes zero also reclocking, according to speed of crankshaft ω
kAnd crankshaft accelerations α
k, calculating obtains dynamic bent axle sampling rotational angle theta
m, rotational speed omega
mAnd acceleration alpha
mAnd
The one estimation unit that is designed by discrete (Discrete) dynamic model comprises by an estimator and cooperates another feedback gain value matrix, carries out loop circuit state estimation, disturb in order to filtering noise information, and estimation obtains crank angle and rotating speed.
And, above-mentioned feedback gain value matrix, can make by following method:
One, use a pole assignment method another coefficient of comparison of arranging in pairs or groups, dynamic to obtain the estimator observation system, and compare pole assignment and the obtained feedback gain value matrix of coefficient of comparisons two methods, to obtain the required feedback gain value matrix value in above-mentioned estimation unit.
Its two, use one to separate the transient solution of Riccati equation formula, wherein comprise and incorporate the renewal rule of estimating covariance, with obtain one can real-time update feedback gain value matrix L
k, and obtain the required adaptive feedback gain value matrix value in above-mentioned estimation unit according to this.
Its three, use above-mentioned Riccati equation formula, and separate its infinite time-domain (Infinite Time Horizon), to obtain the required feedback gain value matrix value in above-mentioned estimation unit.
Except that above-mentioned, the present invention also comprises use one stroke judgement unit, reads the crank teeth news of crank angle perceptron and the speed of crankshaft ω that is learnt by processing unit calculating
m, and (Top Dead Center, TDC) position is with the speed of crankshaft ω of record upper dead center to detect the bent axle upper dead center
TDC (i), and with the preceding speed of crankshaft ω that a time upper dead center write down
TDC (i-1)Make comparisons, and obtain compression stroke upper dead center and intake stroke upper dead center, be calculated to 720 the signals that dynamically make zero when spending to signal processing unit, calculate the dynamic sample signal of above-mentioned bent axle more accurately with auxiliary signal processing unit so that crank angle to be provided.
The present invention is owing to take above design, it has the following advantages: the present invention estimates the crank position that according to the traditional time basis method that measures, carry out the estimation and the correction of loop circuit, particularly disturb under the bent axle signal situation that perceptron detected, can obtain comparatively accurate crank angle and rotating speed at extraneous noise.And, the present invention need not by air-distributor pressure signal and engine parameters, get final product the estimation of March Shaft angle and rotating speed, so can effectively reduce the distrubance source that extraneous noise may cause, also can reduce simultaneously the load of estimation device processes signal, and then the reduction equipment cost, and can directly be applied on the different engines.
Embodiment
First sight is shown in Figure 1, and it discloses the framework of the estimating and measuring method of engine crank corner of the present invention and rotating speed, includes a signal processing unit 2 and another corner and rotating speed estimation unit 3 at least, in order to estimation crank angle 5 and rotating speed 6.In addition, the present invention also can comprise one stroke judgement unit 4, handles the bent axle signal more accurately with auxiliary signal processing unit 2.
Above-mentioned estimating and measuring method, mainly being disposed at one can be in utilizing crank angle perceptron 12 to detect the crank teeth news on the flywheel 11 and having on the engine control unit 10 or its periphery of a timer 13 signals output, and carry out crank angle 5 and the speed of crankshaft 6 of estimating engine, its method is as follows:
As Fig. 2, shown in Figure 3;
(1), use a signal processing unit 2, pick up counting 20 from the engine start seasonal timer that turns round, read 21 (as shown in Figure 2) of crank teeth news that detect by the crank angle perceptron on the flywheel, and detect each tooth news rising edge 22, to write down each crank angle θ
k, θ
K-1Tooth news, and between each tooth news by time difference Δ t that timer read
kAnd Δ t
K-1, and obtain speed of crankshaft ω by equation (1) and (2) calculating
kWith acceleration alpha
k, be expressed as follows:
Equation (1)
Equation (2)
(in the formula: ω
kBe the speed of crankshaft of k tooth news, θ
kBe the crank angle of k tooth news, θ
K-1Be the crank angle of k-1 tooth news, Δ t
kBe that k-1 tooth interrogated the time difference between k the tooth news, Δ t
K-1Be that k-2 tooth interrogated the time difference between k-1 the tooth news, α
kBe the crankshaft accelerations of k tooth news, ω
K-1Be the speed of crankshaft of k-1 tooth news).
Above-mentioned when detecting each tooth news rising edge 22 (as Fig. 2, shown in Figure 3), suppose that the crank teeth news have the n tooth, each tooth news is represented a crank angle θ
kOr θ
K-1, as crank angle θ
kWhen being triggered by the rising edge of k tooth news, above-mentioned processing unit 2 can be interrogated k the time difference Δ t the tooth news from k-1 tooth by record
k, and make timer make zero and reclocking 24, continue to calculate in order to equation (1) and (2), and the record 23 of renewal speed of crankshaft and acceleration.
Above-mentioned timer make zero and reclocking 24 after, signal processing unit 2 can provide other unit of the present invention to carry out the bent axle demand 25 of dynamically taking a sample.In other words, signal processing unit 2 can decided under the supposed situation of acceleration, and calculating obtains the dynamic bent axle sampling rotational angle theta of other arithmetic element demand
m, rotational speed omega
mAnd acceleration alpha
m, represent with following equation (3), (4), (5):
Equation (3)
ω
m=ω
k+ α
kΔ t equation (4)
α
m=α
kEquation (5)
(in the formula: Δ t is k the time difference that tooth is interrogated the sampling demand).
(2) use a Discrete Dynamic model, to design an estimation unit 3, in order to the interference of filtering noise information, and estimation crank angle and rotating speed; This Discrete Dynamic model can utilize following equation of state (6) expression:
x
K+1=Fx
k+ Gu
k+ Г w
kEquation (6)
(in the formula:
x
k=[θ
k?ω
k]
T T is observation system sampling time, u
k=α
m, Г=G is the interference matrix of outer bound pair observation system, w
kInterference for outer bound pair observation system).Output vector is then represented with following equation (7):
y
k=Hx
k+ v
kEquation (7)
(in the formula: by θ
mH=[1 0 as can be known], v
kBe the perceptron error in measurement).
Above-mentioned
estimation unit 3 can receive the bent axle sampling rotational angle theta that
signal processing unit 2 is calculated
mAnd acceleration alpha
m, with the corner and the rotating speed of estimation bent axle.When noise disturbs the crank angle perceptron, can reduce its influence for engine control.This
estimation unit 3 mainly is to cooperate another feedback gain value matrix by an estimator, carry out loop circuit state estimation, and this estimator is essentially an observation system
Itself and feedback gain value matrix can together be represented with following equation (8):
Equation (8)
(in the formula:
Be the observation system state of estimator, L=[L
1L
2]
TBe the feedback gain value matrix).
Before stating estimation unit 3 in design, but need to determine that earlier this observation system is an observation system.When the order (Rank) of observing matrix Ω is full order (Full Rank), but then this observation system is an observation system, and this observing matrix Ω can represent with following equation (9):
Equation (9)
Can obtain equation (10) by equation (9) is expressed as follows:
Rank (Ω)=2 equation (10)
So from the result of equation (10), can learn, but this observation system is an observation system.
The design method of above-mentioned feedback gain value matrix L can be divided into following three kinds:
Method one: being the pole assignment method, mainly is according to the selected observation system free frequency ω of artificer
nAnd dampingratio, and according to standard second order dynamic equation (11), to obtain required desirable observation system dynamic equation (11), be expressed as follows:
z
2-a
1Z+a
2=0 equation (11)
(in the formula: z is the discrete time state,
ω
nBe the parameter of free frequency, ζ is the parameter of damping ratio, and T is this observation system sampling time).
Utilize the method for coefficient of comparisons again, the limit of estimator is put at the pole location that the artificer determined, dynamic to obtain required observation system, wherein the observation system of estimator is dynamic, represents with following equation (12):
z
2+ (L
1-2) z+TL
2+ 1-L
1=0 equation (12)
(in the formula: L
1And L
2Be feedback gain value matrix L value).
Relatively aforesaid equation (11) reaches (12), can solve the value of feedback gain value matrix L, represents with following equation (13):
Equation (13)
Method two: for separating the transient solution of Riccati equation formula, to obtain the feedback gain value matrix L of an energy real-time update
kThis Riccati equation formula (14) is expressed as:
M
K+1=F[M
k-M
kHR
-1H
TM
k] F
T+ Г Q Г
TEquation (14)
(in the formula:
M
kFor estimating covariance P
kThe renewal rule, H=[1 0], R is a perceptron noise amount of variability, Q is the amount of variability of system interference; This two parameter is adjustable parameter).
The above-mentioned covariance P that estimates
k, can represent with following equation (15):
P
k=(M
k -1+ H
TRH)
-1Equation (15)
(in the formula: M
kFor estimating covariance P
kThe renewal rule).
Utilize above-mentioned Riccati equation formula (14) to cooperate and estimate covariance equation (15), can solve adaptability feedback gain value matrix L
k, be shown below:
L
k=P
kH
TR
-1Equation (16)
Method three: separate its infinite time-domain for utilizing the Riccati equation formula,, represent with following equation (17) even the k in the also above-mentioned Riccati equation formula (14) is infinitely great:
M
∞=F[M
∞-M
∞HR
-1H
TM
∞] F
T+ Г Q Г
TEquation (17)
Use and obtain required feedback gain value matrix L
∞, represent with following equation (18):
L
∞=M
∞H
T(HM
∞H
T+ R)
-1Equation (18)
The present invention is also contained under the running of crank teeth news that above-mentioned processing unit 2 can read corner perceptron 12, and additional the use 4 (as shown in Figure 1) of one stroke judgement unit is to read the dynamic bent axle sampling acceleration ω of crank teeth news and signal processing unit 2
m, and according to the bent axle moment dynamic change (as shown in Figure 4) of engine under each stroke, to carry out the stroke judgement of bent axle.Wherein, comprise the mode of utilizing specific tooth news especially, as many teeth or a few tooth, to learn the bent axle upper dead center.The overall operation flow process of this judgement unit 4 (cooperating Fig. 4, shown in Figure 5) includes:
Read the crank teeth news 21 of crank angle perceptron and calculate the speed of crankshaft ω that learns by processing unit 2
m42, and continue to carry out bent axle upper dead center (TDC) position probing 43, when recording upper dead center position 44, immediate record is the speed of crankshaft ω of upper dead center at that time
TDC (i)45, and the speed of crankshaft ω that is noted during more last upper dead center (TDC)
TDC (i-1)46; Work as ω
TDC (i)Less than ω
TDC (i-1)The time, be judged to be compression stroke upper dead center 46a, otherwise then be intake stroke upper dead center 46b; And when air inlet or compression top dead center are detected, can provide crank angle to be calculated to 720 the signals that dynamically make zero when spending, make signal processing unit 2 can calculate dynamic bent axle sampling rotational angle theta more accurately to signal processing unit 2
mAnd acceleration alpha
m, and help promoting the accuracy that crank angles and rotating speed are estimated in estimation unit 3.
In said method of the present invention, also can not select stroke judgement unit 4 for use, this moment, 2 of signal processing units provided the sampling rotational angle theta of crankshaft operation 360 degree
mAnd acceleration alpha
mTo estimating on the unit 3, can allow estimation unit 3 calculating obtain equally than more accurate crank angle of tradition and rotating speed.
Moreover, for confirming the feasibility of said method of the present invention, the flywheel 11 (as shown in Figure 1) that will directly measure the 4+1 tooth with crank angle perceptron 12 is an example, and noise added in the bent axle signal, estimate the result of crank angle and rotating speed with explanation the present invention, reach the result who obtains with the direct measurement of traditional perceptron and do one relatively, as described below old:
In the design method one of above-mentioned feedback gain value matrix L, can set observation system free frequency parameter ω
n=5 (rad/s), damping ratio parameter ζ=0.707; And set the amount of variability parameter Q=10 and perceptron noise amount of variability parameters R=10 of system interference in the design method two and three of feedback gain value matrix L
-6In view of the above:
At first carry out the simulation (as Fig. 6, shown in Figure 7) that engine cold (just) starts, the crank angle that discloses actual measurement in Fig. 6 is not that the position is in stroke initial value of zero degree position, but the stroke initial value of mistake is estimated the variation situation of crank position to observe the above-mentioned three kinds of methods of traditional method for measurement and the present invention by this.In other words, in the operation cycle of first 720 degree of bent axle, tradition measures and the above-mentioned three kinds of stroke judgement initial values that method read of the present invention and incorrect (as shown in Figure 6), but can judge correct stroke gradually when second 720 operation cycle of spending of bent axle begin; And because method of the present invention is to estimate the crank position that according to the traditional time basis method that measures, carry out the estimation correction of loop circuit, so the convergence rate of position estimation can be slower than conventional method, speed of crankshaft estimation simultaneously also can be subjected to traditional location to measure the influence (as shown in Figure 7) of change; But after the operation cycle of the 3rd 720 degree of bent axle, crank angle and rotating speed all can be restrained: and because the tradition measurement is directly to utilize equation (1) to calculate speed of crankshaft, so can not be subjected to the influence of its position calculation.
When noise added the crank angle perceptron in 0.48 second, for crank angle and rotating speed estimation result's influence (as Fig. 8, shown in Figure 9).At first demonstrate the estimation result of crank angle among Fig. 8, wherein tradition measures when running into a bigger noise of intensity and get involved the bent axle signal, and its estimation will be the result will make a mistake.And the present invention all can reduce the influence of noise to the crank angle estimation effectively when having noise to get involved.In addition, when noise is got involved, show that traditional method for measurement can produce the deviation of about 4000rpm when the estimation rotating speed, and estimation departure of the present invention has only 500rpm (as shown in Figure 9) approximately.Be all can reduce the influence of noise effectively, so be better than traditional method for measurement to crank angle and rotating speed estimation with, the present invention.