, (5)
where and accordingly . Then, with the account of equation of reference model (2) and parametrical equation of plant (4), will make expression for an error of following as
, (6)
where - an estimation of an error , and it will be formed lower; is function of deviation; is bounded function of time in the force Hurwitz of polynomial and assumption A2 and A4. And on the base of the equation (6) will form the state filters
(7)
Here ; - a numbering matrix in the Frobenius form with characteristic polynomial ; . Then vector for the equation (6) and filters (7) will form as a vector . If to set the law of control as , where - vector of adjusted parameters, than the equation (6) can be rewrite as
, (8)
where - vector of unknown parameters, depending on coefficients of polynomials , and .
As the possibility of measuring of derivative of an output of control plant (1) and reference model (2) is absent, so for realization of evaluation - derivative of function can use any of observers [for example, 6-8]. Let’s make use of scheme, which offered in [7, 8].
, , (9)
where ; , - single matrix of order ; , where are chosen from the condition of stable of matrix , where ; ; - small enough value. In these case in the law of control variables from the observer (9) will be used, it means that in conditions of measuring only, the law of control is technical realized. How it will be shown lower, for realization of the closed-loop control system the order of observer (9) can be chosen equal , but for the proof the order of observer is used equal to estimate - derivatives of function .
So, let’s enter into consideration vector of deviation , where , . Having taken the derivative from vector of deviation on time, will receive . The equation of deviation what in the force of equation (9) and vector of deviation will be as . Let’s transform the equation of deviation and vector of deviation to the equivalent equation relative an output
, . (10)
Where . On the base of equations (9) and (10) rewrite the equation (8) as
, (11)
where - unknown constant coefficients, and depend on coefficients of polynomial .
Statement. If is SPR-function, than there is a number and algorithm of adaptation
, (12)
such that if , and , than the system of control (7), (8), (9) and (12) dissipative, if movement of the system begins in some domain and target condition (3) is fulfilled, where - domain, and determined by the entry conditions of plant (1).
Proof. Let’s write down the equations (10) and (11) as
(13)
where . Use with a lemma [10].
Lemma [10]. If the system is described by the equation , , , where - an continuous Lipshec function on and if has the bounded closed domain of dissipative , where - undermined piece-smooth positively determined function in will exist such that if than the initial system has domain of dissipative , when for some numbers and when is fulfilled condition
, (14)
when .
Let's take Lypunov function for equation (13) as
,
where ; matrix is determined from the equation , . Then, having taken a derivative on time when will receive
(15)
Having used with identity , with estimations , , , and with equation (12), will receive
. (16)
From fulfillment of inequality (16) follows the boundedness of functions , and . As is bounded, so from (13) quantities , and are bounded too. Function will be bounded in the force of Hurwitz of matrix and boundedness . As the vector-function - bounded, than -derivatives of signal also will be bounded in the force of ways of formation of the vector . Function according to the assumption A3, is also the bounded function. Further, will transform the second equation (7)
,
from where follows the boundedness , because , matrix has a characteristic polynomial , which according to the assumption A2 is Hurwitz. So vector is bounded. As , and is bounded vector in the force of boundedness , than controlling influence is bounded function too.
Уважаемый посетитель!
Чтобы распечатать файл, скачайте его (в формате Word).
Ссылка на скачивание - внизу страницы.