阅读说明：本技术 柴油机Urea-SCR系统氨喷射速率和状态同步估计方法 (Ammonia injection rate and state synchronous estimation method for Urea-SCR system of diesel engine ) 是由 魏丽 孙川 余毕超 程晓敏 王建宇 颜伏伍 于 2019-10-15 设计创作，主要内容包括：公开了一种柴油机Urea-SCR系统氨喷射速率和状态同步估计方法,本公开首先对一般非线性系统提出高增益观测器设计方法,能同时获得系统的状态及未知参数估计,该方法不需要对输出求导,仅假设这些输入动态性是有界的,对这些未知输入如何变化没有任何假设。根据上述方法对Urea-SCR系统进行了高增益观测器设计,能获得Urea-SCR系统的氨喷射速率和NOx浓度、NH<Sub>3</Sub>浓度、氨覆盖率等状态的同步估计。ETC测试循环仿真及误差分析结果表明本发明所设计的高增益观测器具有较高的估计精度,这些估计信息可用于Urea-SCR系统故障诊断,也可用于全状态反馈控制策略或自适应控制策略,能省掉安装传感器的成本。(The method firstly provides a high-gain observer design method for a general nonlinear system, can simultaneously obtain the state of the system and unknown parameter estimation, does not need to differentiate the output, only assumes that the input dynamics is bounded, and has no any assumption on how the unknown input changes. By designing a high-gain observer for the Urea-SCR system according to the method, the ammonia injection rate, the NOx concentration and NH of the Urea-SCR system can be obtained 3 Simultaneous estimation of concentration, ammonia coverage, etc. ETC test cycle simulation and error analysis results show that the high-gain observer designed by the invention has higher estimation precision, and the estimation information can be used for Urea-SCR system fault diagnosis and can also be used for a full-state feedback control strategy or a self-adaptive control strategy, so that the cost for installing the sensor can be saved.)

1. A method for designing a nonlinear system high-gain observer is characterized by comprising the following steps:

1.1 the nonlinear system is:

wherein

State x ∈ RⁿIs an n-dimensional real vector with x^k∈R^nkK is 1, …, q, and q is n₁≥n₂≥…≥n_q，

1.2 the conditions are assumed to be:

(1) the state of the nonlinear system, the unknown input, the known input are bounded;

for each function f^k(u, v, x), k 1, …, q-1, the rank of which is required to satisfy

In addition, for two positive scalars α_fAnd beta_fIs provided with

(2) measuring the value x¹Is divided into two parts

Wherein

meanwhile, the following condition one and condition two need to be satisfied:

the first condition is as follows:

the second condition is as follows:α_vand beta_vAre two positive numbers;

(3) the time partial derivative of the unknown input is defined as ξ (t), and the time partial derivative is bounded;

1.3 converting the nonlinear system into an augmentation system shown in a formula (3) by using state division and input;

wherein the content of the first and second substances,

1.4 designing a high-gain observer of the nonlinear system according to the 1.1-1.3 to obtain unknown parameters and state estimation of the nonlinear system, wherein the high-gain observer is modeled by an equation (4);

wherein

2. A method for synchronously estimating the ammonia injection rate and the state of a Urea-SCR system of a diesel engine is characterized in that a high-gain observer of the Urea-SCR system is designed by using the nonlinear system high-gain observer design method of claim 1, and the ammonia injection rate and the internal NO of the Urea-SCR system are obtained simultaneously_X、NH₃Estimating the concentration and ammonia coverage rate state;

the Urea-SCR system state equation is as follows:

will be provided withAs unknown input, choose

Wherein the content of the first and second substances,

The designed Urea-SCR system high-gain observer is as follows:

the formulae (6a), (6c) and (6d) are each used for the internal state variable NH of the SCR catalyst₃Estimation of concentration, NOx concentration and ammonia coverage, said equation (6b) being used to estimate the unknown input

Wherein the content of the first and second substances,are respectively

3. The method of claim 2, wherein the ammonia injection rate and state synchronization estimation is obtained by converting the ammonia input concentration to ammonia injection rate according to the following equation (6e)

Wherein the content of the first and second substances,

Technical Field

The disclosure relates to the field of engines, in particular to a synchronous estimation method for ammonia injection rate and state of a Urea-SCR system of a diesel engine.

Background

Selective Catalytic Reduction (SCR) technology is widely used for NOx emission control of medium and heavy diesel vehicles due to its advantages of high efficiency, high selectivity, high economy, sulfur resistance, etc. In order to meet the latest national six-emission regulation requirements, a closed-loop feedback control strategy is required to be adopted by a diesel engine aftertreatment system so as to reduce the calibration workload and improve the transient control performance.

Closed-loop feedback control strategies for Urea-SCR systems typically require feedback of catalyst outlet state information to modify the control variables. On the other hand, the ammonia injection rate is an important parameter in the control of the Urea-SCR system, but as the service time of the Urea-SCR system increases, the actual ammonia injection rate and the ammonia storage capacity change due to the aging of the system, the change is usually slowly time-varying and uncertain, and the obtaining of the actual ammonia injection rate information is crucial to the design of the control strategy of the Urea-SCR system.

In engineering applications, the measured values of the sensors are generally used as feedback information, but all the states required by the controller are not possible to obtain by using the sensor measurements and the cost is high. Although NOx sensors have been widely used in diesel engines, the ammonia sensors currently on the market are not widely popularized due to immature technology, few varieties and high cost; meanwhile, ammonia coverage is an important state for Urea-SCR system control, but no device or sensor can directly obtain the measured value of ammonia coverage at present. In a non-linear observable system, the design observer can provide an estimate that asymptotically approaches an unmeasured state, while eliminating the cost of installing sensors, if not all states are measurable, but only some of the output states.

Most state observers of the research design are mainly used to obtain ammonia coverage, NOx concentration, NH inside the catalyst₃Concentration, etc., state, the proposed observer, in which the ammonia injection rate of the Urea-SCR system is usually taken as a known parameter,however, research finds that as the service time of the Urea-SCR system increases, the actual ammonia injection rate may change along with the aging of the system, the change may be slowly time-varying and uncertain, and generally needs to be regarded as unknown parameters, and no research relates to the synchronous estimation of the ammonia injection rate and the internal state of the catalyst.

Disclosure of Invention

Firstly, a high-gain observer design method is provided for a general nonlinear system, the state and unknown parameter estimation of the system can be obtained simultaneously, the method does not need to differentiate the output, only the dynamics of the input is assumed to be bounded, and no assumption is made on how the unknown input changes. Then, the Urea-SCR system is subjected to high-gain observer design according to the method, so that the ammonia injection rate, the NOx concentration and NH of the Urea-SCR system can be obtained₃Simultaneous estimation of concentration, ammonia coverage, etc.

According to an aspect of the embodiments of the present disclosure, a method for designing a nonlinear system high-gain observer is provided, including:

(1) the high-gain observer is designed by adopting a nonlinear system, the input of the nonlinear system is divided into a known input and an unknown input, the unknown input is only required to be a bounded absolute continuous function, no assumption is needed for how the unknown input changes, the first p vectors in the state vectors of the nonlinear system are measurable vectors, and the output of the nonlinear system is measurable output.

(2) Designing the high gain observer to estimate the unknown input takes into account the following assumptions:

1) the state of the nonlinear system, the unknown input, the known input are bounded;

2) the measurement of the nonlinear system can be divided into two parts;

3) the time partial derivative of the unknown input is bounded.

(3) Converting the nonlinear system into an augmented system by using state division and input; if the assumption is satisfied, the observer of the augmented system can ensure that the estimation error converges to a small value.

(4) And designing the nonlinear system high-gain observer.

According to another aspect of the embodiment of the disclosure, a method for synchronously estimating the ammonia injection rate and the state of the Urea-SCR system of the diesel engine is provided, the high-gain observer of the Urea-SCR system is designed by utilizing the nonlinear system high-gain observer design method, and the ammonia injection rate and the internal NO of the Urea-SCR system are obtained simultaneously_X、NH₃Concentration, ammonia coverage status estimation.

According to the method for synchronously estimating the ammonia injection rate and the state of the Urea-SCR system of the diesel engine, the established high-gain observer is subjected to ammonia injection rate estimation and SCR system state estimation effect verification in Matlab by using ETC transient test cycle data. And comparing the estimated value of the observer with the model value, and verifying the estimation accuracy of the observer under the ETC circulation. NOx concentration and NH downstream of SCR system of contrast ratio observer₃Comparing the concentration, the ammonia coverage rate estimated value and the model value with an absolute error, and analyzing the cause of the error; compared with the estimated value, the model value and the absolute error of the ammonia injection rate of the SCR system, the observer can still ensure higher estimation precision.

The advantages of the present disclosure are as follows:

(1) the Urea-SCR system high-gain observer established by the disclosure can not only obtain the estimation of the NOx concentration, NH3 concentration, ammonia coverage and other states in the Urea-SCR catalyst, but also obtain the synchronous estimation of the ammonia injection rate.

(2) The high gain observer designed by the present disclosure does not need to output derivation when estimating unknown parameters, only assumes that the dynamics of the inputs are bounded, and does not assume how the unknown inputs change.

(3) ETC transient test cycle simulation and error analysis results show that the high-gain observer designed by the method has high estimation precision, the estimation information can be used for Urea-SCR system fault diagnosis, and can also be used for a full-state feedback control strategy or a self-adaptive control strategy, and the cost for installing the sensor can be saved.

(4) The high-gain observer design method has general applicability to nonlinear system unknown parameter and state synchronous estimation, and can also be used for estimation of unknown parameters such as ammonia storage capacity of a Urea-SCR system.

Drawings

The present disclosure is described in further detail below with reference to the attached drawings and the detailed description.

FIG. 1 illustrates engine speed and torque under an ETC test cycle according to one embodiment of the present disclosure.

FIG. 2 illustrates exhaust gas flow and catalyst temperature at an ETC test cycle according to one embodiment of the present disclosure.

FIG. 3 illustrates NO upstream of an SCR catalyst under ETC test cycles according to one embodiment of the present disclosure_XConcentration profile.

FIG. 4 illustrates SCR downstream NOx concentration estimates and model value versus absolute error at ETC test cycles according to one embodiment of the disclosure.

FIG. 5 illustrates SCR downstream NH under ETC test cycles according to one embodiment of the present disclosure₃And comparing the concentration estimated value with the model value and absolute error.

Fig. 6 illustrates SCR system ammonia coverage estimates and model value versus absolute error under an ETC test cycle according to one embodiment of the disclosure.

FIG. 7 illustrates ammonia injection rate estimates and model value versus absolute error at ETC test cycles according to one embodiment of the present disclosure.

Detailed Description

The method can simultaneously obtain the state variable estimation and the ammonia injection rate estimation of the Urea-SCR system, can simulate the designed high-gain observer through Matlab, qualitatively and quantitatively analyzes the error between the estimated value and the model value, and verifies the accuracy of the estimated value of the observer under the ETC test cycle.

The method does not need to differentiate the output, only assumes that the input dynamics are bounded, and has no any assumption on how the unknown input changes. The following steps 1.1-1.3 disclose a nonlinear system high-gain observer design method.

Step 1.1: the input of the nonlinear system is divided into a known input and an unknown input, the unknown input is only required to be a bounded absolute continuous function, no assumption is needed for how the unknown input changes, the first p vectors in the state vector of the nonlinear system are measurable vectors, and the output of the nonlinear system is measurable output.

The nonlinear system is as follows:

wherein

State x ∈ RⁿIs an n-dimensional real vector with x^k∈R^nkK is 1, …, q, and q is n₁≥n₂≥…≥n_q，Unknown input v ∈ R^mIs an m-dimensional real vector, which is a bounded absolute continuous function; knowing the input u ∈ R^s-mIs an s-m dimensional real vector, and controls the input total dimension to be s;

is an identity matrix; system output

Is n₁A dimension real vector; the first p vectors in the state vector x are measurable vectors.

Step 1.2: designing a high gain observer to estimate the unknown input takes into account the following assumptions.

(1) The state of the nonlinear system, the unknown input, the known input are bounded;

for each function f^k(u, v, x), k 1, …, q-1, the rank of which is required to satisfy

In addition, for two positive scalars α_fAnd beta_fIs provided with

Is an identity matrix.

(2) Measuring the value x¹Divided into two parts, see formula (2)

Wherein

Is m₁A real vector of the dimension(s),

is p-m₁Vector of real dimension, m is less than or equal to m₁< p, and

meanwhile, the following condition one and condition two need to be satisfied:

the first condition is as follows:I_mis an identity matrix.

The second condition is as follows:

α_vand beta_vAre two positive numbers.

(3) The time partial derivative of the unknown input is defined as ξ (t) and is bounded.

Step 1.3: converting the nonlinear system into an augmentation system shown in formula (3) by using state division and input;

wherein the content of the first and second substances,

is a state quantity, component, of the augmentation system

Step 1.4: designing a high-gain observer of the nonlinear system for estimating unknown parameters and states, wherein the high-gain observer is modeled by an equation (4);

wherein

(where i is 1,2, … q, j is q),

and I is an identity matrix.

Then, the high-gain observer of the Urea-SCR system is designed by utilizing the design method of the high-gain observer of the nonlinear system, and the ammonia injection rate and the internal NO of the Urea-SCR system are obtained simultaneously_X、NH₃Concentration, ammonia coverage status estimation.

The Urea-SCR system state equation is as follows:

will be provided with

As unknown input, choose

Is provided with

Wherein the content of the first and second substances,

and

respectively catalyst inlet NH₃And NOx gas concentration;

andrespectively the catalyst outlet NH₃And NOx gas concentration;is the ammonia coverage;

is the ammonia storage capacity; k is a radical of_iAnd E_i(i ═ ads, des, red, ox are subscripts for adsorption, desorption, SCR reaction, NH3 oxidation reaction, respectively) are pre-factors and activation energies. Definition of

R is the cosmic gas constant, T is the catalyst temperature; f is the exhaust gas volumetric flow rate; v is the catalyst volume.

The designed Urea-SCR system high-gain observer is as follows:

the formulae (6a), (6c) and (6d) are each used for the internal state variable NH of the SCR catalyst₃Concentration, NOx concentration, and ammonia coverage, equation (6b) for estimating unknown inputs

It is the ammonia input concentration in mol/m³。

Wherein the content of the first and second substances,

are respectively

And

an estimated value of (d); the observer gain beta needs to be designed₁And beta₂To ensure that the observer error asymptotically converges to zero. Beta is a₁And beta₂The principle of choice is that it should be positive and suitably large to compensate for the uncertainty of the system. However, note that too large a β₁And beta₂Will cause the observer to degrade the estimated performance and be sensitive to measurement noise.

Since the ammonia injection rate is typically used as a control quantity in the controller, an estimate of the ammonia injection rate is obtained by converting the ammonia input concentration into the ammonia injection rate by the following equation (6 e).

Wherein the content of the first and second substances,

is an ammonia injection rate estimate; r_s,EGIs the exhaust gas constant; p_ambIs at atmospheric pressure;

is the exhaust gas mass flow rate.

And finally, carrying out ammonia injection rate estimation and SCR system state estimation effect verification on the established high-gain observer in Matlab by utilizing ETC transient test cycle data. And comparing the estimated value of the observer with the model value, and verifying the estimation accuracy of the observer under the ETC circulation. NOx concentration and NH downstream of SCR system of contrast ratio observer₃Comparing the concentration, the ammonia coverage rate estimated value and the model value with an absolute error, and analyzing the cause of the error; compared with the estimated value, the model value and the absolute error of the ammonia injection rate of the SCR system, the observer can still ensure higher estimation precision.

Carrying out ETC transient test circulation on an engine bench, wherein the rotating speed and the torque of the engine are shown in figure 1, the idle speed is about 650r/min, the maximum rotating speed is about 1350r/min, and the maximum torque is about 800 N.m under the ETC transient test; corresponding exhaust gas flow and catalyst temperature are shown in fig. 2, the ETC transient test cycle temperature range is 450-650K, and the temperature is over 473K (urea injection temperature threshold 200 ℃) under most working conditions; FIG. 3 shows a NOx concentration profile upstream of the SCR catalyst under ETC test cycles.

And (3) verifying the ammonia injection rate and the state estimation effect of the designed high-gain observer in Matlab, and comparing and analyzing the simulation result.

FIGS. 4 and 5 compare NOx concentration and NH, respectively, downstream of the SCR system of the high gain observer₃Concentration estimates and model values and absolute error, defined as the estimate minus the model value. From FIGS. 4(a) and 5(b), the NOx concentration and NH concentration of the high gain observer can be seen₃The concentration estimate and the model value both fit better in whole. In FIG. 4(b), the downstream NOx concentration error is less than 1X 10 as a whole^-4mol/m³Only slightly higher in the middle about 400-500 s, but the maximum error peak value is 3 multiplied by 10^-4mol/m³Within. Downstream NH in FIG. 5(b)₃The concentration error is not more than 2.5 multiplied by 10 as a whole^-5mol/m³。

Fig. 6 and 7 show the ammonia coverage and comparison of ammonia injection rate estimates and model values and the absolute error, respectively, for a high gain observer. It can be seen from FIGS. 6(a) and 7(a) that the ammonia coverage and ammonia injection rate estimates are better in integral agreement with the model values. FIG. 6(b) shows that the ammonia coverage error is within 0.05 as a whole, except that the ammonia coverage error is slightly increased in the last 100s, and the maximum error does not exceed 0.1; the ammonia injection rate error in FIG. 7(b) fluctuates significantly due to the transient nature of the ETC test cycle, but most errors are 5X 10^-4Within mol/s, error peak value is not more than 10^-3mol/s。

Further, the estimation effect of the high-gain observer is analyzed by table 1, the Mean Absolute Percentage Error (MAPE) of the downstream NOx concentration, ammonia leakage, ammonia coverage and ammonia injection rate are respectively 4.29%, 5.07%, 3.9% and 6.61%, and the MAPE error is less than 7%, which indicates that the unknown input and state estimation values of the high-gain observer can be better fit model values; meanwhile, the mean absolute error (MAD) of the downstream NOx concentration is 4.04X 10^-5mol/m³Root Mean Square Error (RMSE) of 7.27X 10^-5mol/m³(ii) a The average absolute error and the root mean square error of the ammonia leakage are both less than 1.2 multiplied by 10^-5mol/m³(ii) a It can be seen from FIGS. 4(b) and 5(b) that the downstream NOx concentration and the ammonia slip error peak do not exceed 3X 10, respectively^-4And 2.5X 10^-5mol/m³The downstream NOx concentration and ammonia leakage error are both on an acceptable level as a whole; meanwhile, the errors of the ammonia coverage rate MAD and the RMSE are both less than 0.02, the error of the peak value is less than 0.1, and the integral error is at an acceptable level; the MAD and RMSE errors for ammonia injection rates were 1.42 × 10, respectively^-4mol/s and 2.12X 10^-4mol/s, local error peak of ammonia injection rate of not more than 10 in FIG. 7(b)^-3mol/s。

TABLE 1 error between estimated value and model value of high gain observer

ETC transient test cycle simulation and error analysis results show that the high-gain observer designed by the invention has higher estimation precision, and the estimation information can be used for Urea-SCR system fault diagnosis and can also be used for a full-state feedback control strategy or a self-adaptive control strategy, so that the cost for installing the sensor can be saved.