CN103149838B

CN103149838B - The self-adaptation control method of methanol self-heating reforming hydrogen manufacturing process

Info

Publication number: CN103149838B
Application number: CN201310066499.1A
Authority: CN
Inventors: 卢建刚; 王学飞; 陈金水; 施英姿; 庄宏; 游杰; 王新立
Original assignee: Zhejiang University ZJU
Current assignee: Zhejiang University ZJU
Priority date: 2013-03-01
Filing date: 2013-03-01
Publication date: 2015-10-14
Anticipated expiration: 2033-03-01
Also published as: CN103149838A

Abstract

Aiming at the uncertainty of model parameters in the methanol autothermal reforming hydrogen production process, the self-adaptive control method for the methanol autothermal reforming hydrogen production process proposed by the present invention does not depend on the precise mathematical model of the controlled object, and remains stable when the model is mismatched. A good control effect can be obtained. In the present invention, an adaptive controller is used to control the flow rate of the reaction raw material methanol aqueous solution according to the expected hydrogen production, the actual hydrogen production and the actual hydrogen production change rate, and at the same time, a variable ratio controller with reforming temperature constraints is used to control the flow rate of the other reaction raw material air. flow. The invention can adapt to the coupling between model parameter uncertainty and input in the hydrogen production process of methanol autothermal reforming, and realizes the advanced control of the hydrogen production process of methanol autothermal reforming.

Description

Adaptive control method for methanol autothermal reforming hydrogen production process

技术领域technical field

本发明涉及甲醇自热重整制氢过程的控制方法，特别涉及甲醇自热重整制氢过程的自适应控制方法。The invention relates to a control method for the hydrogen production process of methanol autothermal reforming, in particular to an adaptive control method for the hydrogen production process of methanol autothermal reforming.

背景技术Background technique

随着能源危机的加剧，寻求可以替代化石燃料的新能源已经成为一个研究热点。能源问题对国民经济和国家发展有着重大影响，因此我国也将开发新能源作为一个战略。With the intensification of the energy crisis, seeking new energy sources that can replace fossil fuels has become a research hotspot. Energy issues have a major impact on the national economy and national development, so my country also takes the development of new energy as a strategy.

氢气被认为是一种高效的清洁能源。燃料电池技术的快速发展也在推动着氢能的开发。燃料电池是一种将化学能直接转化为电能的能源转换装置。燃料电池因其具有很高的能量转换效率而受到关注。燃料电池的一个非常合适的燃料就是氢气。但由于氢气分子小、易燃易爆等性质，导致氢气的存储和运输存在困难。甲醇自热重整制氢工艺是直接为燃料电池现场供氢的优选方案之一。Hydrogen is considered to be an efficient and clean energy. The rapid development of fuel cell technology is also promoting the development of hydrogen energy. A fuel cell is an energy conversion device that converts chemical energy directly into electrical energy. Fuel cells have attracted attention due to their high energy conversion efficiency. A very suitable fuel for fuel cells is hydrogen. However, due to the small size of hydrogen molecules, flammability and explosive properties, it is difficult to store and transport hydrogen. The methanol autothermal reforming hydrogen production process is one of the preferred solutions for directly supplying hydrogen to fuel cells on site.

甲醇自热重整制氢过程是一个涉及多个化学反应的复杂过程，是一个多输入多输出系统，并且模型参数具有不确定性，输入之间还存在着耦合。另外，燃料电池对输入气体的压力有严格的要求，频繁的大幅波动会引起燃料电池故障和缩短燃料电池的寿命。这些都给控制带来了困难，因此经典控制算法难以取得良好的控制效果。自适应控制因其具有适应变化的能力，成为一种可以应对系统不确定性的有效工具。The hydrogen production process of methanol autothermal reforming is a complex process involving multiple chemical reactions. It is a multi-input and multi-output system, and the model parameters are uncertain, and there are couplings between inputs. In addition, the fuel cell has strict requirements on the pressure of the input gas, and frequent large fluctuations will cause fuel cell failure and shorten the life of the fuel cell. All of these bring difficulties to the control, so the classical control algorithm is difficult to achieve good control effect. Adaptive control is an effective tool to cope with system uncertainty because of its ability to adapt to changes.

发明内容Contents of the invention

为解决甲醇自热重整制氢过程的控制问题，本发明提供一种甲醇自热重整制氢过程的自适应控制方法。In order to solve the control problem of the hydrogen production process of methanol autothermal reforming, the invention provides an adaptive control method for the hydrogen production process of methanol autothermal reforming.

甲醇自热重整制氢过程的自适应控制方法，其特征是采用自适应控制器根据期望氢气产量y_d、实际氢气产量y和实际氢气产量变化率x₂来操纵反应原料甲醇水溶液的流量u₁，同时采用具有重整温度T约束的变比值控制器来操纵另一反应原料空气的流量u₂，具体步骤包括：An adaptive control method for the hydrogen production process of methanol autothermal reforming, which is characterized in that an adaptive controller is used to manipulate the flow rate u of methanol aqueous solution as the reaction raw material according to the expected hydrogen production y _d , the actual hydrogen production y and the actual hydrogen production change rate x ₂ _1. At the same time, use a variable ratio controller with a constraint on the reforming temperature T to manipulate the flow rate u ₂ of another reaction raw material air. The specific steps include:

所述自适应控制器按以下实际氢气产量y和甲醇水溶液的流量u₁之间的模型来操纵甲醇水溶液的流量u₁：The adaptive controller manipulates the methanol _- water flow u1 according to the following model between the actual hydrogen production y and the methanol _- water flow u1:

$\{\begin{matrix} {\overset{\cdot &Center Dot;}{x x}}_{11} = = {x x}_{22} \\ {\overset{\cdot &Center Dot;}{x x}}_{22} = = - - {a a}_{11} {x x}_{11} - - {a a}_{22} {x x}_{22} + + {b b}_{11} {u u}_{11} + + {b b}_{22} {\overset{\cdot &Center Dot;}{u u}}_{11} + + d d \\ y the y = = {x x}_{11} \end{matrix} - - - - - - ((11))$

式中，x₁为表征实际氢气产量的状态变量，为x₁的导数；为x₂的导数；为u₁的导数；d为干扰，其绝对值小于等于干扰的上限值d_M；为x₁、x₂、u₁、与d的线性函数，a₁、a₂、b₁、b₂为所述线性函数的正系数。In the formula, x ₁ is the state variable representing the actual hydrogen production, is the derivative of x ₁ ; is the derivative of x ₂ ; is the derivative of u ₁ ; d is interference, its absolute value is less than or equal to the upper limit value d _M of interference; for x ₁ , x ₂ , u ₁ , A linear function with d, a ₁ , a ₂ , b ₁ , b ₂ are positive coefficients of the linear function.

所述自适应控制器中，定义误差e₁、误差e₂和误差r为In the adaptive controller, define error e ₁ , error e ₂ and error r as

$\{\begin{matrix} {e e}_{11} = = {x x}_{11} - - {y the y}_{d d} \\ {e e}_{22} = = {x x}_{22} - - {\overset{\cdot &Center Dot;}{y the y}}_{d d} \end{matrix} - - - - - - ((22))$

r=λe₁+e₂ (3)r=λe ₁ +e ₂ (3)

式中，λ为大于0的系数。In the formula, λ is a coefficient greater than 0.

所述自适应控制器中，甲醇水溶液的流量u₁的导数为In the adaptive controller, the derivative of the methanol-water flow u ₁ for

${\overset{\cdot &Center Dot;}{u u}}_{11} = = - - k k ((t t)) r r - - [[k k ((t t)) δ δ + + {k k}_{v v} {v v}^{22}]] sgn sgn ((r r)) + + {\overset{^^}{W W}}^{T T} Φ Φ ((Z Z)) - - - - - - ((44))$

式中，δ、k_v为大于0的系数；其中为y_d的二阶导数；sgn(r)为符号函数；k(t)为In the formula, δ and k _v are coefficients greater than 0; in is the second derivative of y _d ; sgn(r) is a sign function; k(t) is

$k k ((t t)) = = σ σ [[11 + + \frac{11}{{ω ω}_{11}} {| | {\overset{^^}{W W}}^{T T} Φ Φ ((Z Z)) | |}^{22} + + \frac{11}{{ω ω}_{22}} {| | h h (({x x}_{11},, {x x}_{22},, {u u}_{11})) | |}^{22}]] - - - - - - ((55))$

式中，σ、ω₁、ω₂为大于0的系数；h(x₁,x₂,u₁)≥|-a₁x₁-a₂x₂+b₁u₁|。为神经网络的输出；Φ(Z)为作用函数，其中Z=V^TX，V为4×N维的权值矩阵，N为神经元的数目，X=[x₁ x₂ u₁ 1]^T为神经网络的输入；为N×1维的理想权值向量W的估计值，的更新规则为In the formula, σ, ω ₁ and ω ₂ are coefficients greater than 0; h(x ₁ , x ₂ , u ₁ )≥|-a ₁ x ₁ -a ₂ x ₂ +b ₁ u ₁ |. is the output of the neural network; Φ(Z) is the action function, where Z=V ^T X, V is a 4×N-dimensional weight matrix, N is the number of neurons, X=[x ₁ x ₂ u ₁ 1] ^T is the input of the neural network; is the estimated value of the N×1-dimensional ideal weight vector W, The update rule for

$\overset{\cdot &Center Dot;}{\overset{^^}{W W}} = = - - ηΦ ηΦ ((Z Z)) r r - - γη γη | | r r | | \overset{^^}{W W} - - - - - - ((66))$

式中，η、γ为大于0的系数。In the formula, η and γ are coefficients greater than 0.

所述变比值控制器中，空气的流量u₂和甲醇水溶液的流量u₁的比值为K，即 _In the variable ratio controller, the ratio of the flow u2 _of the air to the flow u1 of the aqueous methanol solution is K, that is

u₂=K(n)u₁ (7)u ₂ =K(n)u ₁ (7)

式中，K(n)为第n个采样周期空气的流量u₂和甲醇水溶液的流量u₁的比值；K(n)的取值根据重整温度T进行更新，重整温度T偏高时K(n)减小，重整温度T偏低时K(n)增大。In the formula, K(n) is the ratio of the flow rate u ₂ of air and the flow rate u ₁ of methanol aqueous solution in the nth sampling period; the value of K(n) is updated according to the reforming temperature T, when the reforming temperature T is too high K(n) decreases, and K(n) increases when the reforming temperature T is low.

甲醇自热重整制氢过程的自适应控制方法，其特征是所述实际氢气产量变化率x₂采用高增益观测器进行估计，得到实际氢气产量变化率x₂的估计值所述高增益观测器为An adaptive control method for the hydrogen production process of methanol autothermal reforming, characterized in that the actual hydrogen production rate of change x ₂ is estimated by a high-gain observer to obtain an estimated value of the actual hydrogen production rate of change x ₂ The high gain observer is

$\{\begin{matrix} {\overset{\cdot &Center Dot;}{\overset{^^}{x x}}}_{11} = = {\overset{^^}{x x}}_{22} + + {p p}_{11} k k ((y the y - - \overset{^^}{y the y})) \\ {\overset{\cdot \cdot}{\overset{^^}{x x}}}_{22} = = {p p}_{22} {k k}^{22} ((y the y - - \overset{^^}{y the y})) \\ \overset{^^}{y the y} = = {\overset{^^}{x x}}_{11} \end{matrix}- - - - - - - ((88))$

式中，为实际氢气产量x₁的估计值，为的导数；为的导数；为y的估计值；k为观测器增益；p₁、p₂为系数，取值必须满足如下方程In the formula, is the estimated value of actual hydrogen production x ₁ , for derivative of for derivative of is the estimated value of y; k is the observer gain; p ₁ and p ₂ are coefficients, and the values must satisfy the following equation

z²+p₁z+p₂=0 (9)z ² +p ₁ z+p ₂ =0 (9)

具有互不相同根，其中z为所述方程的未知数。have mutually distinct roots, where z is the unknown of the equation.

本发明的有益效果是，提供了一种甲醇自热重整制氢过程的自适应控制方法。针对甲醇自热重整制氢过程的模型参数不确定性，本发明利用的自适应控制不依赖于被控对象的精确数学模型，这使得控制方法在模型失配时，仍然可以取得良好的控制效果。本发明采用所述自适应控制器根据期望氢气产量y_d、实际氢气产量y和实际氢气产量变化率x₂来操纵反应原料甲醇水溶液的流量u₁，同时采用所述具有重整温度T约束的变比值控制器来操纵另一反应原料空气的流量u₂。本发明可以适应甲醇自热重整制氢过程的模型参数不确定性和输入之间的耦合，实现对甲醇自热重整制氢过程的先进控制。The invention has the beneficial effects of providing an adaptive control method for the hydrogen production process of methanol autothermal reforming. Aiming at the uncertainty of model parameters in the hydrogen production process of methanol autothermal reforming, the adaptive control used in the present invention does not depend on the precise mathematical model of the controlled object, which enables the control method to still achieve good control when the model is mismatched Effect. In the present invention, the self-adaptive controller is used to manipulate the flow rate u ₁ of the reaction raw material methanol aqueous solution according to the expected hydrogen production y _d , the actual hydrogen production y and the actual hydrogen production rate x ₂ , and at the same time, the above-mentioned reforming temperature T constraint is adopted A variable ratio controller is used to manipulate the flow rate u ₂ of another reaction raw material air. The invention can adapt to the coupling between model parameter uncertainty and input in the hydrogen production process of methanol autothermal reforming, and realizes the advanced control of the hydrogen production process of methanol autothermal reforming.

附图说明Description of drawings

图1为本发明对应的甲醇自热重整制氢过程的控制框图。Fig. 1 is a control block diagram of the hydrogen production process corresponding to the autothermal reforming of methanol in the present invention.

具体实施方式Detailed ways

甲醇自热重整制氢过程的自适应控制方法以反应原料甲醇水溶液的流量u₁和另一反应原料空气的流量u₂为操纵变量，以氢气产量y和重整温度T为被控变量。The self-adaptive control method of methanol autothermal reforming hydrogen production process uses the flow rate u1 of the reaction raw material methanol water and the flow rate u2 _of the other reaction raw material air as the manipulated variables, and takes the _hydrogen production y and the reforming temperature T as the controlled variables.

所述甲醇水溶液中，水和甲醇的摩尔比一般为(1.0～1.5):1，本实施例采用1.2:1。In the aqueous methanol solution, the molar ratio of water to methanol is generally (1.0-1.5):1, and 1.2:1 is used in this embodiment.

$\{\begin{matrix} {\overset{\cdot \cdot}{x x}}_{11} = = {x x}_{22} \\ {\overset{\cdot \cdot}{x x}}_{22} = = - - {a a}_{11} {x x}_{11} - - {a a}_{22} {x x}_{22} + + {b b}_{11} {u u}_{11} + + {b b}_{22} {\overset{\cdot \cdot}{u u}}_{11} + + d d \\ y the y = = {x x}_{11} \end{matrix} - - - - - - ((11))$

式中，x₁为表征实际氢气产量的状态变量，为x₁的导数；为x₂的导数；为u₁的导数；d为干扰，其绝对值小于等于干扰的上限值d_M；为x₁、x₂、u₁、与d的线性函数，a₁、a₂、b₁、b₂为所述线性函数的正系数，正系数a₁、a₂、b₁、b₂的取值通过系统辨识实验获得。In the formula, x ₁ is the state variable representing the actual hydrogen production, is the derivative of x ₁ ; is the derivative of x ₂ ; is the derivative of u ₁ ; d is interference, its absolute value is less than or equal to the upper limit value d _M of interference; for x ₁ , x ₂ , u ₁ , A linear function with d, a ₁ , a ₂ , b ₁ , b ₂ are the positive coefficients of the linear function, and the values of the positive coefficients a ₁ , a ₂ , b ₁ , b ₂ are obtained through system identification experiments.

$\{\begin{matrix} {e e}_{11} = = {x x}_{11} - - {y the y}_{d d} \\ {e e}_{22} = = {x x}_{22} - - {\overset{\cdot \cdot}{y the y}}_{d d} \end{matrix} - - - - - - ((22))$

r=λe₁+e₂ (3)r=λe ₁ +e ₂ (3)

${\overset{\cdot \cdot}{u u}}_{11} = = - - k k ((t t)) r r - - [[k k ((t t)) δ δ + + {k k}_{v v} {v v}^{22}]] sgn sgn ((r r)) + + {\overset{^^}{W W}}^{T T} Φ Φ ((Z Z)) - - - - - - ((44))$

式中，σ、ω₁、ω₂为大于0的系数；h(x₁,x₂,u₁)≥|-a₁x₁-a₂x₂+b₁u₁|。为神经网络的输出；Z=V^TX=[z₁ z₂ … z_N]^T，V为4×N维的权值矩阵，N为神经元的数目，X=[x₁ x₂ u₁ 1]^T为神经网络的输入；本实施例的作用函数In the formula, σ, ω ₁ and ω ₂ are coefficients greater than 0; h(x ₁ , x ₂ , u ₁ )≥|-a ₁ x ₁ -a ₂ x ₂ +b ₁ u ₁ |. is the output of the neural network; Z=V ^T X=[z ₁ z ₂ … z _N ] ^T , V is a 4×N-dimensional weight matrix, N is the number of neurons, X=[x ₁ x ₂ u ₁ 1] ^T is the input of neural network; The action function of the present embodiment

Φ(Z)=[φ₁(z₁) φ₁(z₂) … φ₁(z_N)]^T Φ(Z)=[φ ₁ (z ₁ ) φ ₁ (z ₂ ) … φ ₁ (z _N )] ^T

$φ φ (({z z}_{i i})) = = \frac{{e e}^{{z z}_{i i}} - - {e e}^{{- - z z}_{i i}}}{{e e}^{{z z}_{i i}} + + {e e}^{{- - z z}_{i i}}},, i i = = 1,2 1,2,, \cdot &Center Dot; \cdot \cdot \cdot &Center Dot;,, N N$

为N×1维的理想权值向量W的估计值，的更新规则为 is the estimated value of the N×1-dimensional ideal weight vector W, The update rule for

$\overset{\cdot \cdot}{\overset{^^}{W W}} = = - - ηΦ ηΦ ((Z Z)) r r - - γη γη | | r r | | \overset{^^}{W W} - - - - - - ((66))$

所述自适应控制器中有系数λ，σ，ω₁，ω₂，δ，k_v，η和γ，它们对性能存在影响。There are coefficients λ, σ, ω ₁ , ω ₂ , δ, k _v , η and γ in the adaptive controller, which have an influence on the performance.

λ主要影响动态性能。λ的取值越大，动态响应速度加快，但同时操纵变量的变化也更为剧烈，这提高了对执行器的要求。因此，λ的取值需要综合考虑对动态性能的要求和执行器的能力来设定。λ mainly affects the dynamic performance. The larger the value of λ, the faster the dynamic response, but at the same time the change of the manipulated variable is more severe, which increases the requirements for the actuator. Therefore, the value of λ needs to be set in consideration of the requirements for dynamic performance and the ability of the actuator.

σ，ω₁和ω₂的作用是对k(t)进行调节。k(t)的取值越大，系统的动态响应越快，但同时操纵变量的变化也更为剧烈，因此会提高了对执行器的要求。除此以外，当k(t)的取值过大时，也容易给系统带来超调现象。因此，σ，ω₁和ω₂的取值需要综合考虑对动态性能的要求和执行器的能力来设定。在实际中，如果对h(x₁,x₂,u₁)的先验知识不足，可以将作为一个整体进行考虑。The role of σ, _ω1 and _ω2 is to adjust k(t). The larger the value of k(t), the faster the dynamic response of the system, but at the same time the change of the manipulated variable is more severe, so the requirements for the actuator will be increased. In addition, when the value of k(t) is too large, it is easy to bring overshoot to the system. Therefore, the values of σ, ω ₁ and ω ₂ need to be set in consideration of the requirements for dynamic performance and the ability of the actuator. In practice, if the prior knowledge of h(x ₁ ,x ₂ ,u ₁ ) is insufficient, the Consider it as a whole.

[k(t)δ+k_vv²]sgn(r)一项对干扰和噪声有抑制作用，但是过大会产生抖动。因此，δ和k_v的取值也需要综合考虑对性能的要求和执行器的能力来设定。[k(t)δ+k _v v ² ] sgn(r) can suppress interference and noise, but if it is too large, it will cause jitter. Therefore, the values of δ and k _v also need to be set in consideration of the performance requirements and the ability of the actuator.

η和γ影响的更新速度。η和γ的取值越大，的更新速度越快，但更新过快会加重执行器的负荷，并可能产生超调。因此，η和γ的取值同样需要结合实际情况来设定。η and γ effects update speed. The larger the values of η and γ are, the The faster the update speed is, but too fast update will increase the load on the actuator and may cause overshoot. Therefore, the values of η and γ also need to be set in combination with the actual situation.

所述变比值控制器中，空气的流量u₂（单位：升/分钟）和甲醇水溶液的流量u₁（单位：毫升/分钟）的比值为K，即In the variable ratio controller, the ratio of the air flow rate u ₂ (unit: liter/minute) to the methanol aqueous solution flow rate u ₁ (unit: ml/minute) is K, namely

u₂=K(n)u₁ (7)u ₂ =K(n)u ₁ (7)

式中，K(n)为第n个采样周期空气的流量u₂和甲醇水溶液的流量u₁的比值；K(n)的取值根据重整温度T（单位：摄氏度）进行更新，重整温度T偏高时K(n)减小，重整温度T偏低时K(n)增大。本实施例采用的更新规则为In the formula, K(n) is the ratio of the flow rate u ₂ of air and the flow rate u ₁ of methanol aqueous solution in the nth sampling period; the value of K(n) is updated according to the reforming temperature T (unit: Celsius), and reforming When the temperature T is high, K(n) decreases, and when the reforming temperature T is low, K(n) increases. The update rule adopted in this embodiment is

K(n)=K(n-1)+△KK(n)=K(n-1)+△K

$ΔK ΔK = = \{\begin{matrix} - - 0.015 0.015,, T T > > 550550 \\ 0.015 0.015,, T T < < 510510 \\ 0,510 0,510 \leq \leq T T \leq \leq 550550 \end{matrix}$

z²+p₁z+p₂=0 (9)z ² +p ₁ z+p ₂ =0 (9)

本发明对应的甲醇自热重整制氢过程的控制框图如图1所示。甲醇自热重整制氢过程的自适应控制方法采用自适应控制器根据期望氢气产量y_d、实际氢气产量y和实际氢气产量变化率x₂的估计值来操纵反应原料甲醇水溶液的流量u₁，同时采用具有重整温度T约束的变比值控制器来操纵另一反应原料空气的流量u₂。The control block diagram of the methanol autothermal reforming hydrogen production process corresponding to the present invention is shown in FIG. 1 . The adaptive control method of methanol autothermal reforming hydrogen production process adopts the adaptive controller according to the estimated value of expected hydrogen production y _d , actual hydrogen production y and actual hydrogen production change rate x ₂ to manipulate the flow rate u ₁ of methanol aqueous solution as the reaction raw material, and at the same time use the variable ratio controller with the constraint of reforming temperature T to manipulate the flow rate u ₂ of the other reaction raw material air.

上述具体实施方式用来解释说明本发明，仅为本发明的优选实施例而已，而不是对本发明进行限制，在本发明的精神和权利要求的保护范围内，对本发明作出的任何修改、等同替换、改进等，都落入本发明的保护范围。The specific implementation above is used to explain the present invention, and it is only a preferred embodiment of the present invention, rather than restricting the present invention. Within the spirit of the present invention and the scope of protection of the claims, any modifications and equivalent replacements made to the present invention , improvements, etc., all fall within the protection scope of the present invention.

Claims

1. the self-adaptation control method of methanol self-heating reforming hydrogen manufacturing process, is characterized in that adopting adaptive controller according to expectation hydrogen output y _d, actual hydrogen output y and actual hydrogen Yield change rate x ₂carry out the flow u of control response material benzenemethanol aqueous solution ₁, adopt the variable-ratio control device with reforming temperature T constraint to handle the flow u of another reaction raw materials air simultaneously ₂, concrete steps comprise:

The flow u of following actual hydrogen output y and methanol aqueous solution pressed by described adaptive controller ₁between model handle the flow u of methanol aqueous solution ₁:

\{\begin{matrix} {\overset{\cdot}{x}}_{1} = x_{2} \\ {\overset{\cdot}{x}}_{2} = {- a}_{1} x_{1} - a_{2} x_{2} + b_{1} u_{1} + b_{2} {\overset{\cdot}{u}}_{1} + d \\ y = x_{1} \end{matrix} - - - (1)

In formula, x ₁for characterizing the state variable of actual hydrogen output, for x ₁derivative; for x ₂derivative; for u ₁derivative; D is interference, and its absolute value is less than or equal to the higher limit d of interference _m; for x ₁, x ₂, u ₁, with the linear function of d, a ₁, a ₂, b ₁, b ₂for the positive coefficient of described linear function;

In described adaptive controller, definition error e ₁, error e ₂with error r be

\{\begin{matrix} e_{1} = x_{1} - y_{d} \\ e_{2} = x_{2} - {\overset{\cdot}{y}}_{d} \end{matrix} - - - (2)

r＝λe ₁+e ₂(3)

In formula, λ be greater than 0 coefficient;

In described adaptive controller, the flow u of methanol aqueous solution ₁derivative for

{\overset{\cdot}{u}}_{1} = - k (t) r - [k (t) δ + k_{v} v^{2}] s g n (r) + {\hat{W}}^{T} Φ (Z) - - - (4)

In formula, δ, k _vfor being greater than the coefficient of 0; wherein for y _dsecond derivative; Sgn (r) is sign function; K (t) is

k (t) = σ [1 + \frac{1}{ω_{1}} | {\hat{W}}^{T} Φ (Z) |^{2} + \frac{1}{ω_{2}} | h (x_{1}, x_{2}, u_{1}) |^{2}] - - - (5)

In formula, σ, ω ₁, ω ₂for being greater than the coefficient of 0; H (x ₁, x ₂, u ₁)>=|-a ₁x ₁-a ₂x ₂+ b ₁u ₁|; for the output of neural network; Φ (Z) is action function, wherein Z=V ^tx, V are the weight matrix of 4 × N dimension, and N is neuronic number, X=[x ₁x ₂u ₁1] ^tfor the input of neural network; for the estimated value of the desirable weight vector W that N × 1 is tieed up, update rule be

\overset{\cdot}{\hat{W}} = - η Φ (Z) r - γ η | r | \hat{W} - - - (6)

In formula, η, γ be greater than 0 coefficient;

In described variable-ratio control device, the flow u of air ₂with the flow u of methanol aqueous solution ₁ratio be K, namely

u ₂＝K(n)u ₁(7)

In formula, K (n) is the flow u of the n-th sampling period air ₂with the flow u of methanol aqueous solution ₁ratio; The value of K (n) upgrades according to reforming temperature T, and when reforming temperature T is greater than 550 DEG C, K (n) reduces, and when reforming temperature T is less than 510 DEG C, K (n) increases.