IEEE Power Electronics Magazine - December 2020 - 21

This subsystem responds during the problem solving by the
HELP, WHY and HOW commands of the user. The ES user is
usually a semi-skilled person and does not understand in
depth the complexity of problem solving. The ES explains to
the user the technical features of the problem with explanatory texts and figures when HELP command is used. The
WHY command explains to the user why the ES is asking the
relevant information from the user, and HOW command
explains how the problem has been solved. Often, user education is the sole purpose for designing the ES.
The structure of a knowledge base with the matrix of
16 rules, for example, can be read as
Rule 1: IF X = X 1 AND Y = Y1, THEN Z = A ........ (1)
Rule 2 : IF X = X 2 AND Y = Y2, THEN Z = B ........ (2)
h

Rule 16 : IF X = X 4 AND Y = Y4, THEN Z = P ...... (16)
where X, Y and Z are defined as the rule parameters, and X 1,
Y1, X 2, Y2, etc. are the respective parameter values. The data
base that supports the knowledge base embeds the parameter values which can be in the form of data (logical, numeric,
facts or statements). A rule has a conditional (premise or
antecedent) part in the IF statement and the action (consequent or conclusion) part in the THEN statement. The logical
connectives can be AND, OR and NOT operations for drawing
conclusion. A rule can be fired or executed if the conditional
part is true (called forward chaining), and then the action part
guided by the THEN statement is executed. In backward
chaining, the system starts with the desired conclusion and
then finds the rules that could have caused the conclusion. In
a practical system, there can be large number of rules and the
parameters with the values may also be large and complex.
The knowledge base can be simple or adaptive in nature,
depending on the system changes or technology advancement. Knowledge can be defined as shallow or deep. Shallow
knowledge base can be directly obtained from the present
knowledge of the system. A deep knowledge can be derived
by machine learning from the system model and simulation
responses based on designer's or researcher's knowledge.
The ES knowledge is normally structured or represented
in the form of a tree with the help of a number of frames. A
frame consists of a cluster of characteristic rules and their
associated parameters. The advantage of frame-based structure of knowledge base is the logical organization of a large
amount of knowledge in the modular form. The root frame is
the core of the knowledge base. It may have child frames and
grandchild subframes. Each subframe can be considered as
a subdomain of expert knowledge. Assume, for example, a
customer wants to select a certain commercial drive product
[2] from a vendor for a certain application with the consultation of an ES program. In this case, the root frame embeds the
expertise of a general sales engineer. Consider that there are
two child subframes which embed the application engineer's
expertise of cage type induction motors (IM) and PM synchronous motors (PMSM), respectively. The user interfaces the

root frame in the beginning, and based on the user dialog and
then consultation with IM and PMSM drive child subframes,
the IM drive appears to be the choice. Once the type of drive
is selected, the details of the converter and the machine will
be calculated by the ES based on the specifications supplied
by the user. The iterations of computation can be done by the
ES until the user is fully satisfied with the ratings and performances of the drive. A grandchild subframe will then provide
the auxiliary features of IM drive, such as price, delivery and
installation considerations.
An ES SHELL is a software environment platform [3] for
efficient and user-friendly development of ES program. The
SHELL can interface with external programs, such as data
base files, graphical files, simulation files, interface routines,
and mathematical files. A limited amount of data, logical and
arithmetic capabilities can be directly embedded in the ES
program, but for large data, such as product catalog file, database files should be constructed. For complex calculations,
such as solving differential equations, the ES program can
access external programs. Similarly, the SHELL can control
and transfer data to and from simulation programs. A powerful feature of shell is that it can integrate pictures with the
knowledge base which are compressed as files. Thus, a fully
designed power electronic system can be the output of the ES
in the form of a circuit schematic showing all the numerical
values of the parameters. A number of ES shells are available
to design an ES program. One example is PC Plus [3] developed by Texas Instruments. It uses PC SCHEME language
which is a dialect of LISP language. The designer should have
familiarity with PC SCHEME although English-like Abbreviated Rule Language (ARL) is used for fast development of the
program. For time-critical real-time applications, C language
should be used. If the program is resident in the SHELL, the
developer can easily alter or update it, but program modification is not possible in client computer environment.

Expert System Based Control of Smart Grid
In the past, ES has been applied extensively in many power
electronic systems [1]. The applications include automated
P-I tuning of drive, power electronic system fault diagnostics, selection of commercial drive product, drive system
configuration selection, design and simulation that generates
real-time controller object code for a DSP, and control strategy development of smart grid. As an example of ES application, we will discuss briefly the preliminary control of smart
grid based on ES [4]. Suffice to say that control and protection of a modern smart grid are extremely complex, particularly if the grid is large. In the present status of the technology, the control and protection strategy of future large smart
grid is not yet well-defined.
As mentioned before, a smart grid is basically an advanced
electric power grid using state-of-the-art technologies that
will improve system availability, reliability, power quality,
energy efficiency, and security with optimum resource utilization and economical electricity to the consumers. Figure 2
shows a simplified control block diagram of smart grid based
December 2020

z IEEE POWER ELECTRONICS MAGAZINE

IEEE Power Electronics Magazine - December 2020

Table of Contents for the Digital Edition of IEEE Power Electronics Magazine - December 2020