Scribd
Upload
281 views4 pages

Principles of Discretization

The document discusses principles of discretization when approximating continuous functions at discrete time intervals for numerical integration. It presents: 1) Using a polynomial approximation pn(t) of order n to approximate the function f(t), with higher-order polynomials providing better accuracy but requiring more coefficients and data points. 2) Zero-order approximations that estimate f(t) as a constant value at the previous or next time instant, resulting in forward or backward rectangular approximations. 3) A first-order (straight-line) approximation that minimizes error compared to zero-order, known as trapezoidal or bilinear approximation. 4) The relationship for converting a transfer function from the continuous s-domain

Uploaded by

manmohancq
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOC, PDF, TXT or read online on Scribd
281 views4 pages

Principles of Discretization

The document discusses principles of discretization when approximating continuous functions at discrete time intervals for numerical integration. It presents: 1) Using a polynomial approximation pn(t) of order n to approximate the function f(t), with higher-order polynomials providing better accuracy but requiring more coefficients and data points. 2) Zero-order approximations that estimate f(t) as a constant value at the previous or next time instant, resulting in forward or backward rectangular approximations. 3) A first-order (straight-line) approximation that minimizes error compared to zero-order, known as trapezoidal or bilinear approximation. 4) The relationship for converting a transfer function from the continuous s-domain

Uploaded by

manmohancq
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOC, PDF, TXT or read online on Scribd
You are on page 1/ 4

PRINCIPLES OF DISCRETIZATION

Suppose that we have a continuous function f(t) whose values are known only at discrete instants of time, say f(T), f(2T), , f(kT). Now, if we want to integrate f(t) for one period, it can be done as follows:
k T

( k )T 1

f (t )dt

k T

( k )T 1

(t ) d t

---------- (1)

where, pn(t) is a polynomial of order n that approximates f(t) i.e. pn(t) is of the form,
p n (t ) = a 0 + a1t + ... + a n t n

----------- (2)

Now, higher the order of the polynomial, better is the accuracy. But, higher the order, greater is the number of co-efficients to be found and consecutively higher is the number of points on f(t) that is to be known. Different p n(t) gives different discretization methods. Integration:
k T

The process of integration given by, under the curve f(t) as shown below:

( k )T 1

f (t )dt

can be visualized as the area

An approximation to this can be arrived at by using the polynomial approximation outlined above. Following are some methods that are generally used in practice.

Zero order approximations: Taking a zero order polynomial to approximate f(t), we have from (2) above,
f (t ) p 0 (t ) = a 0

Now, a 0 can be any of the following: 1. p 0 (t ) = a 0 = f [( k 1)T ] 2. p 0 (t ) = a 0 = f [( k )T ] (Forward Rectangular Approximation) (Backward Rectangular Approximation)

The above two have been obtained from the known value at (k-1) th and kth instants respectively. Equivalently, the approximations can be represented as follows. As seen in the diagrams, there is an error associated with the approximation.

First order approximation: To minimize the error, we can go in for a higher order polynomial approximation. Taking a straight-line approximation (i.e. an approximation of the form a 0 + a1t ), we have a diagram as shown below:

As shown in the diagram above, the error is less compared to the zero-order approximation. This approximation is called trapezoidal or bilinear approximation in mathematics. Relationship between s and z under bilinear approximation: The process of integration can be represented as follows in s domain: I(s)
1 s

O(s)

The equivalent z domain representation should be as follows: I(z)


H ( z)

O(z)

The approximation for the integration process using the bilinear transformation can be written mathematically (from Trapezoidal Approximation diagram) as,
O[ kT ] O[( k 1)T ] + T {I [( k 1)T ] + I [ kT ]} 2

i.e. O[ k ] O[ k 1] +

T ( I [ k 1] + I [k ]) 2

Transforming to Z-domain, we have,


T 1 [ z I ( z ) + I ( z )] 2 O ( z ) T (1 + z 1 ) = i.e. H ( z ) = I ( z) 2(1 z 1 ) O ( z ) z 1O ( z ) +

i.e. the required substitution to convert s domain transfer function to z domain is,
s 2(1 z 1 ) T (1 + z 1 )

Now, whenever we require a discrete domain equivalent of an s domain transfer function, this relation can be used.

You might also like