konsep design mikro hidro

8/10/2019 Konsep design mikro hidro

1/6

De La Salle University Manila, ECE Technical Report, September 1, 2004

1

CONCEPTUAL DESIGN OF A RULE-BASE FOR A MICRO-HYDRO

POWER PLANT FEEDBACK CONTROL SYSTEM

Emmanuel A. Gonzalez1, Jingel Tio

2and Felicito S. Caluyo

3

Department of Electronics and Communications Engineering

College of Engineering

De La Salle University, Manila2401 Taft Ave., Malate, Manila 1004, Philippines

1 [email protected] [email protected]

[email protected]

ABSTRACT

This paper presents the conceptual design of a rule basefor a feedback control system for the regulation of a

micro-hydro power plant. Included in this paper is thediscussion on the feedback control system based on the

use of machine intelligence for the generator terminal

voltage and frequency regulation. The analysis and

discussion of the feedback control system is also

presented. Rule base models and modeling techniques arediscussed with a design example using a Mamdanian

Fuzzy Logic system.

I. INTRODUCTION

The stability of a small power system, such as a micro-

hydro power plant has been an issue that arises to

engineers within its development. With a small amount of

power that is fed to an electricity grid, which is usually

less than 100kW, it has been proven that the plantsstability is a serious problem and is a major concern

especially in small networks of application [1].

The regulation of the terminal voltage and its frequencyhas been developed to impede the causes of instability

using different methods such as feedback control [1], and

the application of electronic load controllers [2], [3].Feedback control which has the advantage of controlling

the flow rate of the water intake is an efficient method ofpower plant regulation. On the other hand, application of

dump loads with some circuitries can be less difficult to

implement.

In this paper, the conceptual design of a feedback control

system is discussed with its impact in the non-linearoperation of the micro-hydro power plant. In Section II,

the overall micro-hydro power plant system is introduced

and discussed in detail. In Section III, the general

feedback control system diagram is discussed with furtherdiscussion on some important parts of the system. In

Section IV, the machine intelligence that is implemented

in the control system is discussed. A design exampleusing Mamdani-type Fuzzy Logic is discussed in Section

V. Conclusions are discussed in Section VI.

II. THE MICRO-HYDRO POWER PLANT

The micro-hydro power plant used in this research is

composed of a pump and a water tank that serves as thesource of water flow and energy that is used to run the

turbine and generator. A tailwater reservoir is alsoincluded to ensure that the water is recirculated throughthe entire system. The water from the pump and tank is

fed to a turbine via a gate valve that controls the flow rate

of water. The turbine is connected to a synchronous

generator that converts the mechanical energy of theturbine to an electrical energy that can be utilized by

connecting a load through the terminals of the generator.

The block diagram of the micro-hydro power plant is

shown in Fig. 1.

Fig. 1. Components of the micro-hydro power plant.


2/6


2

Problems arise with the use of a micro-hydro power plant

which involves: (1) the decrease in the generator terminal

voltage with resistive loads, (2) decrease in the generator

terminal voltage with lagging loads, (i.e. capacitive loads),and (3) increase in the generator terminal voltage with

leading loads, (i.e. inductive loads), which happens when

the generator is loaded. Experimentally, it has been

proven that the decrease in the terminal voltage for a

resistive and capacitive load is about 8 to 20 per centbelow its no-load value, and about 25 to 50 per cent

below its no-load value, respectively [3].

III. THE FEEDBACK CONTROL SYSTEM

The feedback control system is shown in Fig. 2,emphasizing on its integration to the micro-hydro power

plant. The output voltage generated by the turbine-

synchronous generator tandem is a function mainly of the

input water flow rate to the turbine, which is controlled bya gate valve. However, it is also expected that the output

voltage generated by the plant is non-linear with respect

to the input flow rate because of two reasons: (1) waterflow rate saturation due to the diameter of the pipes, and

(2) backlash which results for the non-linear

characteristics of the turbine-synchronous generator

tandem.

The gate valve, on the other hand, is controlled by a servomotor which is driven by the servo motor interface circuit

inside the feedback control system which is controlled by

a microcontroller. The servo-motor acts as a counter-part

of a human hand that rotates the valve for controlling thewater flow rate to the turbine.

The control system consists of a signal conditioning

system that converts the measured terminal voltage to be

understood by the processor. The heart of the feedbackcontrol system consists of a microcontroller that decides

on the control of the servo motor based on measured

terminal voltage.

The control system acquires two inputs from the generator:

the (1) present voltage [ ]x n , and its (2) rate-of-change

[ ]x n . It is used to determine not only the presentcharacteristics of the terminal voltage, but also its rate-of-

change. These values are inputted to a digital control

system that samples these data to digital values through

an analog-to-digital converter built inside the

microcontroller.

The output of the control system is the signal that will

turn the gate valve to a desired angle, , which in turn willdirect the desired water flow rate, rF(t), to the turbine.

Hence, the characterization of the servo-gate-valve

system by determining the characteristics of the flow ratewith respect to the angle of turn is needed. This is done by

setting the gate valve at a certain angle, and then

measuring the water flow rate using any conventional

Fig. 2. Block diagram of the proposed feedback control system for theregulation of the micro-hydro power plant.

Fig. 3. Block diagram of the inference engine system.

Fig. 4. An approach for the modeling of a rule-base.


3/6


3

method. With this, a characteristic curve of the water flow

rate with respect to the valve turn angle is obtained.

IV. INCORPORATION OF MACHINE

INTELLIGENCE

The implementation of machine intelligence is done witha microcontroller. Before implementing intelligent

algorithms, a model must first be constructed. In this

section the modeling approach is discussed in the firstpart of this section. The second part discusses three rule-

base models that can be used as a control approach.

A. Modelling Approach

The proposed individual steps in the modeling of the

control algorithm are shown in Fig. 4. Such steps in thedevelopment of the model are shown with the assumption

that the data gathered is proper and correct.

Step 1: Rule-base selection. Different types of rule-base structures and models can be used

depending on a situation where that rule-base

is best suited, such as the three examples

discussed in the second part of this section. It

is basically based from the designers point ofview on how the micro-hydro power plant can

be controlled, which is also based on the

designers experience.

Step 2: Initial rule-base formulation.The formulation

of the initial rule-base is also based onintuitiveness and uncertainties that rises duringthe design of the micro-hydro power plant.

The rules that constitute the model are

extracted in an automated way, in which, the

exact procedure depends on the type of modelused.

Step 3: Rule-base simplification. The initial rule-baseobtained from data may be redundant and

unnecessarily complex as it is based on

numerical optimization [6]. The complexity

can be reduced by using similarity analysis,

where sets representing compatible orredundant concepts are identified and merged.

Step 4: Rule validation. The final model is either

accepted as appropriate for the micro-hydropower plant or is rejected to evaluation and

validation. An important role in the rulevalidation procedure is the inspection of the

rule model that was used. The input and output

spaces can also be analyzed to determine

models validity with the rule model.

B. Rule-Base Models

Based on some experimentation, a set of rules is

developed, each of which describes the local input-outputrelation, typically in linear form:

Ri: ifx1isAi1and andxnisAinthenyiisBi (1)

where, i= 1,2,,K.

Here Ri is the ith rule, [ ]1,...,T

nx x=x is the input

(antecedent) variable, Ai1,Ain are sets defined in theantecedent space, yi is the rule output (consequent)

variable,Biis a set defined in the consequent space, andK

denotes the number of rules in the rule base.

The block diagram describing the inference engine systemthat uses the input-output relations is shown in Fig. 3. The

system is a two-input-single-output (TISO) system that

uses the present voltage value and the change of the

voltage as its inputs, and the desired flow rate at theoutput. The rule base of K rules is coupled to the

inference engine that functions as an intelligent watchdog.

The sets in the antecedent and consequent of the rules are

obtained from Eq. 1.

Ri: if [ ]x n isAi1and [ ]x n isAi2then [ ]y n isBi (2)

where [ ]x n is the measured present voltage at sample

n, [ ]x n is the change of the voltage which is defined asthe difference of the present value to the previous value,

also denote in Eq. 3.

[ ] [ ] [ ]1x n x n

x nn

= (3)

[ ]y n is the output desired flow rate of the water intake tothe turbine.

Since the micro-hydro power plant is comprised of somenon-linear characteristics, it is therefore necessary to

develop a non-linear intelligent scheme to solve this

problem. One way of implementing such non-linearintelligence is applying Mamdanian Fuzzy Logic to the

system, as shown in Fig. 5 [5]. Based on the set of rules

developed earlier, the local input-output relation is then

converted into a non-linear form, comprising of linguistic


4/6


4

variables. The structure for the fuzzy rules similar to Eq

(1) has fuzzy sets in linguistic form such as small,

medium, and high, both input and output, such that afuzzy rule would have the form

Ri: ifx1isAi1and andxnisAinthenyiisBi (4)

where the input variablesAi1,,Ainand the output variableBiare composed of fuzzy linguistic sets. An example of

this form is shown in Eq. 5.

R1: if [ ]x n is high and [ ]x n is high then [ ]y n islow.

(5)

Two modules were added to compensate the crispness of

the input and output values. The fuzzification and

defuzzification engines function as mapping devices thatconvert the crispness of the inputs to a fuzzy membership

value, and vise-versa.

However, the determination of the fuzzy rules and fuzzymembership function are developed through trial and

error. Model building by input-output data is

characterized by two things; one is a mathematical tool toexpress a system model and the other is the method of

identification [4].

Another method in applying Fuzzy Logic for machine

intelligence is the use of the Takagi-Sugeno approach [4],which has rules in the form

Ri: ifx1isAi1and andxnisAinthenyi=Bi0+

Bi1x1+ +Binxn (6)

where, i= 1,2,,K.

Here Ri is the ith rule, [ ]1,...,T

nx=x is the input

(antecedent) variable, Ai1,Ain are sets defined in theantecedent space, yi is the rule output (consequent)

variable which a function of input variable vector x,

Bi1,,Binare sets defined in the consequent space, and K

denotes the number of rules in the rule base.

V. DESIGN EXAMPLE

Let [ ]x n be the present sample voltage under an inputspace U, defined to be

{ }150,300U volts. (7)

Let [ ]x n be the rate-of-change of voltage value (i.e. thedifference between the present value and previous value),

under an input space V, defined to be

{ }20,20 V volts. (8)

Let [ ]y n be the desired control signal to the gate valvefor the control of the water flow rate, under the output

space Y, which is defined as

{ }0, 2Y degrees. (9)

It is assumed that a 0-degree of angular knob displace

corresponds to a 0 lb/sec flow rate, and a 2 -degree ofangular knob displace corresponds to a 1000 lbs/sec flow

rate. Another assumption is that, the water flow rate to the

turbine corresponds with the degree of turn of the valveknob, characterizing it to be linear. It is also assumed that

the nominal flow rate of water is achieved at a -degreeangular displacement.

By using the modeling steps in the previous section, the

following characteristics and model of the rule-base has

been chosen:

Rule-base model:Mamdani type.

Linguistic variables:

Fig. 5. A Mamdanian fuzzy logic system.


5/6


5

[ ]x n : {low,normal,high}

[ ]x n : {negative_large, negative_small, zero,positive_small,positive_large}

[ ]y n : {very_low,low, normal, high, very_high}.

The membership functions of the linguistic variables

under their respective spaces are shown in Fig. 6.

The developed rule-base in the form of Eq. 5 is composed

of 15 rules.

R1: if [ ]x n is low and [ ]x n is neg_large then

[ ]y n is very_high.

R2: if [ ]x n is low and [ ]x n

is neg_small then[ ]y n is high.

R3: if [ ]x n is low and [ ]x n iszero then [ ]y n ishigh.

R4: if [ ]x n is low and [ ]x n is pos_small then

[ ]y n is high.

R5: if [ ]x n is low and [ ]x n is pos_high then

[ ]y n is normal.R6: if [ ]x n is normal and [ ]x n is neg_large

then [ ]y n is very_high.

R7: if [ ]x n is normal and [ ]x n is neg_small

then [ ]y n is high.

R8: if [ ]x n is normal and [ ]x n is zero then

[ ]y n is normal.

R9: if [ ]x n is normal and [ ]x n is pos_small

then [ ]y n is low.

R10: if [ ]x n is normal and [ ]x n ispos_high then[ ]y n is very_low.

R11: if [ ]x n is high and [ ]x n is neg_large then

[ ]y n is normal.

R12: if [ ]x n is high and [ ]x n is neg_small then

[ ]y n is low.

R13: if [ ]x n is high and [ ]x n is zero then [ ]y n is low.

R14: if [ ]x n is high and [ ]x n is pos_small then

[ ]y n is low.

R15: if [ ]x n is high and [ ]x n is pos_high then

[ ]y n is very_low.

With the 15 rules running the fuzzy logic system and with

a defuzzification method by using centriodal method, the

surface plot is obtained in Fig. 7. Rule-base simplificationcan be done by merging rules 2, 3, 4, and merging rules

Fig. 7. A surface plot showing the control area of the rule-base.

Fig. 6. Input and output membership functions.


6/6


6

12, 13, 14, since their output are all high and low,

respectively.

VI. CONCLUSION

The authors had proposed a methodology for rule base

design of a feedback control system for a micro-hydropower plant. It is also shown that the application of

machine intelligence by using rules that describe the local

input-output relation is applicable and can be used as abackbone for intelligence tuning. However, the

implementation of the proposed controller can display

some irregularity especially since the micro-hydro power

plant constitutes non-linear parameters and characteristics.

One way to compensate such non-linearities is to applysome sort of non-linear machine intelligence such as the

use of Fuzzy Logic.

The author has also proposed a rule-base modeling

approach with three rule-base models that can be used inthe design of the control strategy. It has been defined that

these rules are likely in the form of if-then statementswhich vary depending on the design considerations and

data gathered. A design example by using the Mamdani

Fuzzy Model was used to describe the behavior of the

micro-hydro power plant control system, in which a

surface plot that shows the control area is presented.

REFERENCES

[1] L. A. Gan Lim, Implementation of fuzzy logic

control in micro-hydro power stabilization, DLSUEngineering Journal, Vol. 15, No. 1, pp. 134-141,March, 2002.

[2] V. Aravinthan, A. V. Jayadarshana, J. C. Jayakodi, K.

K. L. S. Kothalawa, J. Peiris, and I. Keerthiratne,Implementation of a portable micro-hydro power

plant using an induction generator controller.

[3] C. S. Siskind, Electrical Machines, New York:

McGraw-Hill, 1966.

[4] T. Takagi, and M. Sugeno, Fuzzy identification of

systems and its applications to modeling andcontrol,IEEE Trans. Syst., Man., and Cybern., vol.

SMC-15, no. 1, pp. 116-132, Jan./Feb. 1985.

[5] E. H. Mamdani, Application of fuzzy algorithms forcontrol of simple dynamic plant, Proc. IEEE, vol.

121, no. 12, pp. 1585-1588, 1976.

[6] M. Setnes and R. Babuska, Fuzzy modeling for

predictive control, Fuzzy control: synthesis and

analysis, New York: John Wiley & Sons, 2000.

[7] F. S. Caluyo, A PLD-based electronic loadcontroller for microhydroelectric power plant.

konsep design mikro hidro

Documents