INTERNATIONAL SCIENTIFIC JOURNAL "INNOVATIONS" WEB ISSN 2603-3771; PRINT ISSN 2603-3763

LIMITATIONS TO SUSPENSION PERFORMANCE

IN A TWO-DEGREE-OF-FREEDOM CAR ACTIVE SUSPENSION

Assoc.Prof. Katerina Hyniova

Faculty of Information Technology –Czech Technical University in Prague, the Czech Republic
[email protected]
Abstract: It is often assumed that if practical difficulties are neglected, active systems could produce in principle arbitrary ideal behavior.
This paper presents the factorization approach that is taken to derive limitations of achievable frequency responses for active vehicle
suspension systems in terms of invariant frequency points and restricted rate of decay at high frequencies. The factorization approach
enables us to determine complete sets of such constraints on various transfer functions from the load and road disturbance inputs for typical
choices of measured outputs and then choose the “most advantageous” vector of the measurements from the point of view of the widest class
of the achievable frequency responses. Using a simple linear two degree-of-freedom car suspension system model it will be shown that even
using complete state feedback and in the case of in which the system is controllable in the control theory sense, there still are limitations to
suspension performance in the fully active state.

Keywords: VEHICLE, ACTIVE SUSPENSION, CONTROL, LIMITATIONS, PARAMETRIZATION

Note that if the one-quarter-car model contains also a passive

1. Introduction suspension system (a sprung of stiffness k and a shock absorber of
Two major performance requirements of suspension are to damping quotient b then the active suspension force u(t) involves
improve ride and handling quality when random road and load also the adequate force generated by the passive suspension system.
disturbances from the environment act upon running vehicles.
On the base of the Youla-Kucera parametrization, complete sets
Automotive suspensions are designed to provide good vibration
of limitations were derived for transfer functions from the road
insulation of the passengers and to maintain adequate adherence of
disturbance input (load f(t) is absent):
the wheel for braking, accelerating and handling, i.e. the purpose of
active suspensions in terms of performance is to improve both of 1 (𝑠) = 𝑍𝑠 (𝑠)
𝐻𝑧𝑤 / (to the sprung mass position) (1)
these conflicting requirements. 𝑍𝑟 (𝑠) 𝑓=0

In this paper, it will be shown the factorization approach taken

to derive limitations of achievable frequency responses for active 2 𝑍𝑠 (𝑠)−𝑍𝑢 (𝑠)
vehicle suspension systems. As we will see, limitations derived for 𝐻𝑧𝑤 (𝑠) = /𝑓=0 (to the suspension deflection)
𝑍𝑟 (𝑠)
a traditional one-quarter-car model (Fig.1) in the frequency domain (2)
arise in the form of invariant frequency points and restricted rate of
decay at zero and infinite frequencies.
3 (𝑠) = 𝑍𝑢 (𝑠)−𝑍𝑟 (𝑠)
Youla-Kucera factorization approach to feedback system 𝐻𝑧𝑤 /𝑓=0 (to the tire deflection) (3)
𝑍𝑟 (𝑠)
stability has been shown in [2], [3] to derive achievable dynamic
responses for active suspension systems of vehicles. Complete sets and analogically for the load disturbance input for various
of constraints on various transfer functions from the road and load choices of measured outputs even for full state feedback.
disturbance inputs were derived for typical choices of measured
outputs.
2. Comprime Factorization
Consider the standard feedback configuration shown in Fig.2,

Fig. 1. One-quarter-car model Fig. 2 Standard feedback configuration

The approach was illustrated for the one-quarter-car model
shown in Fig.1, where: where w is the exogenous input, typically consisting of
u(t) control input (active suspension force) [N] disturbances and sensor noises, u is the control signal, z is the
mu weight of the unsprung mass (wheel) [kg] output to be controlled, and y the measured output. In general, u, w,
ms weight of the sprung mass supported by each wheel and y, and z are vector-valued signals.
taken as equal to a quarter of the total body mass [kg]
The transfer matrices G(s) and K(s) are, by assumption, real-
kt stiffness of the tire [N/m]
rational and proper: G represents a generalized plant, the fixed part
zr(t) road displacement (road disturbance) [m]
of the system, and K the controller [4]. Partition G(s) as:
zs(t) displacement of the sprung mass [m]
zu(t) displacement of the unsprung mass [m] 𝐺11 (𝑠) 𝐺12 (𝑠)
𝐺(𝑠) = (4)
f(t) load disturbance [N] 𝐺21 (𝑠) 𝐺22 (𝑠)

111 YEAR V, ISSUE 3, P.P. 111-114 (2019)

INTERNATIONAL SCIENTIFIC JOURNAL "INNOVATIONS" WEB ISSN 2603-3771; PRINT ISSN 2603-3763

Then Fig.2 stands for the following algebraic equations: It is obvious from (19) and (20) that the sprung mass position
transfer function HSP has an invariant “tire-hop” frequency at
𝑍(𝑠) = 𝐺11 (𝑠)𝑊(𝑠) + 𝐺12 (𝑠)𝑈(𝑠) (5)
𝜔1 = 𝑘𝑡 /𝑚𝑢 , where:
𝑌(𝑠) = 𝐺21 (𝑠)𝑊(𝑠) + 𝐺22 (𝑠)𝑈(𝑠) (6)
𝐻𝑆𝑃 (𝑠)/𝑠=𝑗 𝜔 1 = −𝑚𝑢 /𝑚𝑠 (22)
𝑈(𝑠) = 𝐾(𝑠)𝑌(𝑠) (7)
Similarly, from (16) and (18) the suspension deflection transfer
Manipulating the equations listed above, the following transfer function HSD has an invariant “rattle-space” frequency at
function Tzw(s) from w to z as a linear-fractional transformation of 𝜔2 = 𝑘𝑡 /(𝑚𝑢 + 𝑚𝑠 ) and:
K(s) can be derived:
−1 𝐺 𝐻𝑆𝐷 (𝑠)/𝑠=𝑗 𝜔 2 = −(1 + 𝑚𝑢 /𝑚𝑠 ) (23)
𝑇𝑧𝑤 = 𝐺11 + 𝐺12 𝐾 𝐼 − 𝐺22 𝐾 21 =
= 𝐺11 + 𝐺12 𝐼 − 𝐾𝐺22 −1 𝐾𝐺21 (8) Finally, from (20) and (21), the tire deflection transfer function
HTD does not have any invariant frequency point except 𝜔3 = 0,
It is shown in [1] that the set of all proper real-rational matrices
where:
K(s) stabilizing G(s) is parametrized by a free parameter 𝑄(𝑠) ∈
𝑅𝐻∞ as follows: 𝐻𝑇𝐷 (𝑠)/𝑠=𝑗 𝜔 3 = 0 (24)
−1
𝐾 = 𝑌 − 𝑀𝑄 𝑋 − 𝑁𝑄 =
−1
= 𝑋 − 𝑄𝑁 𝑌 − 𝑄𝑀 (9) 4. Transfer Functions Limitations
where: In the next, consider the standard block diagram shown in
M(𝑠), 𝑁(𝑠), 𝑋(𝑠), 𝑌(𝑠) ∈ 𝑅𝐻∞ , and Fig.2.
𝑀(𝑠), 𝑁(𝑠), 𝑋(𝑠), 𝑌(𝑠) ∈ 𝑅𝐻∞ can be found by comprime As an example, consider:
factorization approach of G22(s): w=zr , z=zs 𝑦 = 𝑧𝑢 , 𝑧𝑠 − 𝑧𝑢 , 𝑧𝑢 − 𝑧𝑟 𝑇 .
Then:
𝐺22 (𝑠) = 𝑁(𝑠)𝑀−1 (𝑠) = 𝑀 −1 (𝑠)𝑁(𝑠) (10)
𝐺11 (𝑠) 𝐺12 (𝑠)
𝑋(𝑠) −𝑌(𝑠) 𝑀(𝑠) 𝑌(𝑠) 𝐺(𝑠) = =
=𝐼 (11) 𝐺21 (𝑠) 𝐺22 (𝑠)
−𝑁(𝑠) 𝑀(𝑠) 𝑁(𝑠) 𝑋(𝑠) 0
1
𝑚 𝑠𝑠2
Substituting the equation (9) into (8) we obtain the transfer 𝑠𝑘 𝑡 −𝑠
matrix Tzw(s) from w to z in terms of the free parameter 𝑄(𝑠) ∈ 𝑚 𝑢 𝑠 2 +𝑘 𝑡 𝑚 𝑢 𝑠 2 +𝑘 𝑡
 = −𝑘 𝑡 (𝑚 𝑢 +𝑚 𝑠 )𝑠 2 +𝑘 𝑡 (25)
𝑅𝐻∞ :
𝑚 𝑢 𝑠 2 +𝑘 𝑡 𝑚 𝑠 𝑠 2 (𝑚 𝑢 𝑠 2 +𝑘 𝑡 )
𝑇𝑧𝑤 (𝑠) = −𝑚 𝑢 𝑠 2 −1
𝑚 𝑢 𝑠 2 +𝑘 𝑡 𝑚 𝑢 𝑠 2 +𝑘 𝑡
= 𝐺11 (𝑠) + 𝐺12 (𝑠)𝑀(𝑠) 𝑌(𝑠) − 𝑄(𝑠)𝑀(𝑠) 𝐺21 (𝑠) =
   = 𝐺11 (𝑠) + 𝐺12 (𝑠) 𝑌(𝑠) − 𝑀(𝑠)𝑄(𝑠) 𝑀(𝑠)𝐺21 (𝑠) The limitations of all achievable closed-loop transfer functions
(12) 𝑇𝑧𝑤 (𝑠) = 𝐻𝑆𝑃 (𝑠) are derived from the right and left comprime
factorization of G22(s), i.e.:
As the parameter Q(s) varies over the set of all stable proper
functions, the equation (12) parametrizes all achievable transfer −𝑚𝑠 𝑠 3
functions Tzw(s). 𝑝4 (𝑠)
−1
If it is assumed that the tire does not leave the ground, for the (𝑚𝑢 + 𝑚𝑠 )𝑠 2 + 𝑘𝑡 𝑚𝑠 𝑠 2 (𝑚𝑢 𝑠 2 + 𝑘𝑡 )
one-quarter car model (Fig.1) the linear differential equations of 𝐺22 = =
𝑝4 (𝑠) 𝑝4 (𝑠)
motion are:
−𝑚𝑠 𝑠 2 𝑀 −1 (𝑠)
𝑚𝑠 𝑧𝑠 = 𝑢 − 𝑓 (13) 𝑝4 (𝑠)
𝑁(𝑠)
𝑚𝑢 𝑧𝑢 = −𝑢 + 𝑘𝑡 (𝑧𝑟 − 𝑧𝑢 ) (14) −1
𝑚 𝑠𝑠 𝑚 𝑠𝑠2
0 1
where zu and zs are measured from the static equilibrium position. 𝑝 2 (𝑠) 𝑝 2 (𝑠)
𝑝 2 (𝑠)
𝑚 𝑢 𝑚 𝑠𝑠2 −𝑚 𝑠 𝑘 𝑡
First, let the load disturbance is absent (f=0). Adding equations   = 0 𝑚 𝑢 +𝑚 𝑠 (26)
𝑝 2 (𝑠) 𝑝 2 (𝑠) 𝑝 2 (𝑠)
(13) and (14) we obtain the invariant equation of: −1 𝑠
0 0
𝑝 2 (𝑠) 𝑝 2 (𝑠)
𝑚𝑠 𝑧𝑠 + 𝑚𝑢 𝑧𝑢 = 𝑘𝑡 (𝑧𝑟 − 𝑧𝑢 ) (15) 𝑀 −1
𝑁 (𝑠)

that is independent on the suspension force u. The following where p2(s) and p4(s) are Hurwitz polynomials of degree 2 and
transfer functions will be investigated: 4, respectively. Then:
𝐻𝑆𝑃 (𝑠) = 𝑍𝑠 (𝑠)/𝑍𝑟 (𝑠) (16) 𝐻𝑆𝑃 (𝑠) =
𝐻𝑆𝐷 (𝑠) = 𝑍𝑠 (𝑠) − 𝑍𝑢 (𝑠) /𝑍𝑟 (𝑠) (17) = 𝐺11 (𝑠) + 𝐺12 (𝑠)𝑀(𝑠) 𝑌(𝑠) − 𝑄(𝑠)𝑀(𝑠) 𝐺21 (𝑠) =
−𝑠𝑘𝑡
𝐻𝑇𝐷 (𝑠) = 𝑍𝑢 (𝑠) − 𝑍𝑟 (𝑠) /𝑍𝑟 (𝑠) (18) 1 𝑠(𝑚 𝑢 𝑠 2 +𝑘 𝑡 ) ∗
=− 𝑌(𝑠) 𝑘𝑡 + 𝑄 (𝑠) (27)
𝑝 4 (𝑠) 𝑝 4 (𝑠)
𝑚𝑢 𝑠 2
3. Invariant Properties
𝑄3 𝑠
Manipulating the equation (15) we can derive the following where𝑄 ∗ 𝑠 = , 𝑄(𝑠) = 𝑄1 𝑠 𝑄2 𝑠 𝑄3 (𝑠) .
𝑝2 𝑠
invariant identities:
It follows from (27) that thanks to the term 𝑌(𝑠) − 𝑄(𝑠)𝑀(𝑠)
(𝑚𝑢 + 𝑚𝑠 )𝑠 2 + 𝑘𝑡 𝐻𝑆𝑃 (𝑠) − 𝑚𝑢 𝑠 2 + 𝑘𝑡 𝐻𝑆𝐷 (𝑠) = 𝑘𝑡 (19) bounded for large s:
𝑚𝑠 𝑠 2 𝐻𝑆𝑃 (𝑠) + 𝑚𝑢 𝑠 2 + 𝑘𝑡 𝐻𝑇𝐷 (𝑠) = −𝑚 𝑢 𝑠 2 (20) 𝑙𝑖𝑚𝑠 2 𝐻𝑆𝑃 (𝑠) < ∞ (28)
𝑠→∞
(𝑚𝑢 + 𝑚𝑠 )𝑠 2
+ 𝑘𝑡 𝐻𝑇𝐷 (𝑠) + 𝑚𝑠 𝑠 2 𝐻𝑆𝐷 (𝑠) = (21)
i.e., the resulting rate of decay is of second degree:
= −(𝑚𝑢 + 𝑚𝑠 )𝑠 2

112 YEAR V, ISSUE 3, P.P. 111-114 (2019)

INTERNATIONAL SCIENTIFIC JOURNAL "INNOVATIONS" WEB ISSN 2603-3771; PRINT ISSN 2603-3763

−2
𝐻𝑆𝑃 (𝑠)/𝑠→∞ = 𝑂(𝑠 ) (29) and suspension deflection velocity are measured, is as follows
[2],[3]:
𝑠(𝑚 𝑢 𝑠 2 +𝑘 𝑡)
It is obvious from (27) that the member has two 1
𝑝 4 (𝑠) 𝐻𝑧𝑤 (𝑠)/𝑠→∞ = 𝑂(𝑠 −3 ) (infinite frequency constraint)
imaginary axis zeros - at s=0 and 𝑠 = 𝑗𝜔1 = 𝑗 𝑘𝑡 /𝑚𝑢 - which can
not be canceled by the denominator of 𝑄 ∗ (𝑠) ∈ 𝑅𝐻∞ . With respect
1
H zw ( s ) / s 0  1  O( s 2 )
(zero frequency constraint)
to (27) and the Bezout identity (11) it follows, that:
1 𝑚𝑢 𝑘𝑡
−𝑠𝑘𝑡 𝐻𝑧𝑤 (𝑗𝜔1 ) = − for 𝜔1 =
1 𝑚𝑠 𝑚𝑢
𝐻𝑆𝑃 (𝑠)/𝑠=0 = − 𝑌(𝑠) 𝑘𝑡 /𝑠=0 =
𝑝4 (𝑠) and analogically:
𝑚𝑢 𝑠 2
𝑘𝑡 𝑘𝑡
=− 𝑌 (𝑠)/𝑠=0 = 1 (30) 2 (𝑠)/
𝐻𝑧𝑤 𝑠→∞ = − 𝑠 −2 + 𝑂(𝑠 −3 ) (infinity freq. constraint)
𝑝 4 (𝑠) 2 𝑚𝑢

and similarly: 2 (𝑠)/

𝐻𝑧𝑤 2
𝑠→0 = 𝑂(𝑠 ) (zero frequency constraint)
−𝑠𝑘𝑡
1 2 (𝑗𝜔 ) = −(1 +
𝐻𝑧𝑤
𝑚𝑢
) for 𝜔2 =
𝑘𝑡
and
𝐻𝑆𝑃 (𝑠)/𝑠=𝑗 𝜔 1 = − 𝑌(𝑠) 𝑘𝑡 /𝑠=𝑗 𝜔 1 = 2 𝑚𝑠 𝑚 𝑢 +𝑚 𝑠
𝑝4 (𝑠)
𝑚𝑢 𝑠 2
3 (𝑠) 𝑘𝑡
= −𝑚𝑢 /𝑚𝑠 𝐻𝑧𝑤 = −1 + 𝑠 −2 + 𝑂 𝑠 −3 (infinity
𝑠→∞ 𝑚𝑢
(31)
freq. constraint)
This result endorses (25).
3 (𝑠)/ (𝑚 +𝑚 )
Thanks to the first order of the imaginary axis zero at s=0, the 𝐻𝑧𝑤 𝑠→0 = − 𝑢 𝑠 𝑠2 + 𝑂(𝑠 4 ) (zero freq. constraint)
𝑘𝑡
first derivative of the transfer function HSP(s) does not have any 3 (𝑗𝜔 ) = 0 for 𝜔 = 0.
similar restrictions at this point. Expressions (29), (30) and (31) 𝐻𝑧𝑤 3 3
create the complete set of limitations which any admissible transfer It is often assumed that if practical difficulties are neglected,
function 𝐻𝑆𝑃 (𝑠) ∈ 𝑅𝐻∞ must satisfy. Another words, if any active systems could in principle produce arbitrary ideal behavior.
complex transfer function satisfies the mentioned limitations, there This paper presents the factorization approach that is taken to derive
exists a stabilizing controller K(s) so that 𝑇𝑧𝑤 (𝑠) = 𝐻𝑆𝑃 (𝑠). It does limitations of achievable frequency responses for active vehicle
not depend on what variables are chosen as the measured output - suspension systems in terms of invariant frequency points and
the limitations always arise in the form of invariant frequency restricted rate of decay at high frequencies.
points as was shown in paragraph 3 and in the form of restricted
rate of decay at infinite frequencies as shown in (29). In control law design for active suspension system of vehicles it
is demanded to prevent magnitudes of the road and load frequency
Complete sets of limitations for the transfer functions HSD(s) responses from being too large. There are some frequency points
and HTD(s) can be similarly carried out from the corresponding and frequency ranges where the transfer functions have modulus
transfer functions G(s) or using (26), (27), (28) and the strictly greater than one i.e. where road and load disturbance
corresponding invariant equation stated above. That way the amplification occur. On the base of the Bode integral it can be
following complete sets of limitations can be derived: shown that the transfer functions must be greater in modulus to at
least the same extend that it is less than one, when measured in
𝐻𝑆𝐷 (𝑠)/𝑠→∞ = 𝑂(𝑠 −2 ) (32)
terms of the area on a Bode magnitude plot [1], [2] [3]. In such a
𝐻𝑆𝐷 (𝑠)/𝑠=0 = 0, 𝐻𝑆𝐷 (𝑠)/𝑠→0 = 𝑂(𝑠) (33) case there is a possibility to shift frequency ranges where
𝐻𝑆𝐷 𝑠 =− 1+
𝑚𝑢
, where disturbance amplification occurs to a ”more advantageous place” or
𝑠=𝑗 𝜔 2 𝑚𝑠 to make magnitudes lower spreading the frequency range.
𝜔2 = 𝑘𝑡 /(𝑚𝑢 + 𝑚𝑠 ) (34)
𝐻𝑇𝐷 (𝑠)/𝑠→∞ = −1 + 𝑂(𝑠 −2 ) (35)
5. Analysis of the Complete Sets of Limitations
𝐻𝑇𝐷 (𝑠)/𝑠=0 = 0,  
𝐻𝑇𝐷 (𝑠)/𝑠→0 = −(𝑚𝑢 + 𝑚𝑠 )𝑠 2 /𝑘𝑡 + 𝑂(𝑠 3 ) In context with transfer functions 𝐻(𝑠) of the one-quarter-car
(36) model given in Section 1, the generalized Bode integral theorem
can be modified as follows [1] :
Note, that even though it is desirable to prevent amplitudes of
the frequency responses 𝐻𝑆𝑃 (𝑠), 𝐻𝑆𝐷 (𝑠), and 𝐻𝑇𝐷 (𝑠) being too Theorem 1. Let 𝑅𝐻∞ is a set of rational transfer functions that
large in any frequency domain, a brief analysis of the expressions are stable (their poles lie in the open right half-plane) and proper
(29) - (36) enables to find out that the investigated transfer (the numerator degree of these functions is less than or equal to the
functions must have modulus strictly grater that one at some denominator degree). Let 𝐻(𝑠) belongs to 𝑅𝐻∞ and satisfies
frequencies what indicates the fact that the road disturbance signal 𝐻(𝑠)/𝑠→∞ = −1 + 𝑂(𝑠 −2 ). Let 𝑧𝑖 , 𝑖 = 1, . . , 𝑛 𝑛 ∈ 𝑁 are zeros of
is amplified at these mentioned frequencies. 𝐻(𝑠) with 𝑅𝑒( 𝑧𝑖 ) > 0. Then:
∞ 𝑛
The same approach can be used to derive limitations for other 0
𝑙𝑜𝑔 𝐻(𝑗𝜔) 𝑑𝜔 = 𝜋 𝑖=1 𝑅𝑒( 𝑧𝑖 ) (37)
transfer functions and various choices of the measured outputs.
In control law design, it is desirable to prevent amplitudes of the
It has been shown that the limitations always arise in the form dynamic responses 𝐻𝑧𝑤 1 (𝑠), 𝐻 2 (𝑠) and 𝐻 3 (𝑠) from being too
𝑧𝑤 𝑧𝑤
𝑘𝑡 large.
of invariant frequency points (for example 𝜔1 = for
𝑚𝑢
𝑘𝑡
A brief examination of the results stated in Section 1 shows that
1 (𝑗𝜔 ), 𝜔 = 2 (𝑗𝜔 ) and 𝜔 = 0 for
𝐻𝑧𝑤 1 2 for 𝐻𝑧𝑤 2 3 the suspension deflection transfer function 𝐻𝑧𝑤 2 (𝑠) has modulus
𝑚 𝑢 +𝑚 𝑠
3 (𝑗𝜔 )) and in the form of restricted rate of decay at frequencies 𝑘𝑡
𝐻𝑧𝑤 3 strictly grater than one for 𝜔2 = 2 (𝑗𝜔 ) =
( 𝐻𝑧𝑤 2
𝑚 𝑢 +𝑚 𝑠
tending to zero and infinity.
𝑚𝑢
− 1+ ), i.e. at this frequency an amplification grater than one
As an example, the complete sets of constraints for transfer 𝑚𝑠
1 (𝑠), 𝐻 2 (𝑠) and 𝐻 3 𝑠 , when suspension deflection
functions 𝐻𝑧𝑤 𝑧𝑤 𝑧𝑤
occurs. This amplification can be made less only and only by

113 YEAR V, ISSUE 3, P.P. 111-114 (2019)

INTERNATIONAL SCIENTIFIC JOURNAL "INNOVATIONS" WEB ISSN 2603-3771; PRINT ISSN 2603-3763

adjusting the ratio of the unsprung and sprung masses. From the 6. Results
result:
Using a simple linear two-degree of freedom car suspension model
3 𝑘𝑡
𝐻𝑧𝑤 (𝑠)/𝑠→∞ = −1 + 𝑠 −2 + 𝑂(𝑠 −3 ) (38) in Fig.1, it was shown that there are still some limitations to
𝑚𝑢
suspension performance even in the fully active state. It has been
3
it is evident, that 𝐻𝑧𝑤 (𝑗𝜔) must tend to one from above as 𝜔 shown in the paper that there are some frequency points and
3
tends to ∞ and it turns out that 𝐻𝑧𝑤 (𝑗𝜔) cannot be made less than frequency ranges where the transfer functions have modulus strictly
or equal to one at all frequencies. Since the right-hand side of (37) greater than one i.e. where road and load disturbance amplification
3 (𝑗𝜔) to be less than occurs. On the base of the Bode integral theorems, it has been
is non-negative then it is not possible for 𝐻𝑧𝑤
or equal to one at all frequencies since that would make the left- shown that the transfer functions must be greater in modulus to at
hand side of the equation (37) negative. It has been shown in [3], least the same extent that it is less than one when measured in terms
that no matter what signals are used for feedback, the tire deflection of the area on a Bode magnitude plot. In such a case there is a
transfer function must amplify road disturbances at some possibility to shift frequency ranges where disturbance
frequencies. This fact is valid even for full state feedback used in amplification occurs to a ”more advantageous place” or to make
the control loop. magnitudes lower spreading the frequency range.
A similar theorem is valid for transfer functions where Acknowledgment
𝐻(𝑠)/𝑠→0 = 1 + 𝑂(𝑠 2 )
This research has been supported by MSMT project INTER-
Theorem 2. Let 𝐻(𝑠) belongs to 𝑅𝐻∞ and satisfies𝐻(𝑠)/𝑠→0 = VECTOR 17019.
1 + 𝑂(𝑠 2 ).Let 𝑧𝑖 , 𝑖 = 1, . . , 𝑛 𝑛 ∈ 𝑁 are zeros of 𝐻(𝑠) with
𝑅𝑒( 𝑧𝑖 ) > 0. Then:
∞
𝑙𝑜𝑔 𝐻(𝑗𝜔)
𝑑𝜔
=𝜋 𝑛 𝑅𝑒 (𝑧 𝑖 )
(39)
References
0 𝜔2 𝑖=1 𝑧 2
𝑖
[1] Freudenberg, J. S., Looze, D. P., Right-half Plane Poles and
Similarly to the consequences of Theorem 3, the result Zeros and Design Tradeoffs in Feedback Systems, IEEE
1 (𝑠)/ 2
𝐻𝑧𝑤 𝑠→0 = 1 + 𝑂(𝑠 ) from Section 1 is the case when Transactions on Automatic Control, 30/6, 1985
1
𝐻𝑧𝑤 (𝑗𝜔) cannot be less than or equal to one at all frequencies
since that would make the left-hand side of (39) negative. This fact [2] Hyniova, K., Frequency Response Limitations for Active
is again valid even when full state feedback is introduced in the Suspension of Vehicles, Micro-CAD 99 – International Computer
control loop [3]. Science Conference, Miskolc, pp. 209-217, 2006,

In such cases that were mentioned above designers have the [3] Hyniova, K., Achievable Dynamic Responses for Active
only possibility ”to shift” the frequencies where the amplifications Suspension System of Vehicles, Habilitation thesis, Prague, 2000
occur to ”more advantageous places” or ”to spread” the ranges [4] Zhou, K., Doyle, J.C., Essentials of Robust Control, Prentice
where amplifications occur making the amplification lower the Hall, London, 1998
positive area of the Bode magnitude plot, i.e. the area where
𝐻(𝑗𝜔) is greater than one (is grater to at least the same extent than
the negative area where 𝐻(𝑗𝜔) is less than one) by choosing a
proper feedback controller. Analogically, similar results can be
derived for arbitrarily chosen measurements and load disturbances.

114 YEAR V, ISSUE 3, P.P. 111-114 (2019)