information. Another problem is how to improve simulation efficiency, especially for intelligent optimization algorithms
which need iterative searching. That is the reason why multi-core environment are highly integrated in the platform. IMSL
is defined to support the M&S of systems through describing intelligent model structures, behaviors, facts, knowledge and
rules and can be applied to a wide range of systems. And IMSL is designed to tackle with knowledge engineering, expert sys-
tem, qualitative reasoning and intelligent optimization problems efficiently. Besides, this platform has a user-friendly inter-
face providing both text and graphic development environment and supports the simulation result display of intelligent
behaviors.
Related key technologies of Intelligent System M&S Platform are described in detail. The remainder of this paper is orga-
nized as follows. Section 2 summarizes related research. Section 3 gives an overview of Intelligent System M&S Platform.
Section 4 discusses design and integration of intelligent models. Section 5 presents some key techniques of implementing
the integrated development environment, including the definition of IMSL, translating and compiling technology, and sim-
ulation engine technology. Section 6 implements a C3I application instance. The functions of IMSL and the platform are ver-
ified through the modeling of air force combat rule sets and the simulation of cooperative attack target allocation. Finally, the
paper is concluded in Section 7.
2. Related work
The pertinent literature of this study mainly focuses on: (a) M&S language and platform; (b) solutions to intelligent sys-
tems; (c) intelligent optimization algorithms and application; (d) M&S of C3I system.
A lot of M&S languages and platforms have been developed and widely used in the whole community to support system
simulation, such as Modelica, MATLAB, UML and SRML. Modelica is an object-oriented language for modeling the complex
physical systems [1]. Models in Modelica are described by differential, algebraic and discrete equations [2]. It is suited for the
modeling of multi-domain cooperative and continuous/discrete systems, and works almost perfectly on that kind of prob-
lem. Marco Lovera have conducted the M&S of satellite dynamics and obtained the results based on the Modelica language
[3]. Victorino Sanz has developed a free Modelica library named DEVSLib to model discrete-event systems [4]. MATLAB is a
language for technical computing with excellent performance [5] providing enriched toolboxes for continuous/discrete sys-
tem modeling. It also provides different intelligent models and optimization algorithms in artificial intelligent toolbox. How-
ever, models in the artificial intelligent toolbox are not comprehensive to describe a complicated intelligent problem. For
examples, MATLAB does not support qualitative modeling, knowledge based expert system, etc. UML is a comprehensive
and universal modeling language [6]. Yves Vanderperren has applied UML to System-on-Chip and hardware-related embed-
ded systems design and combined UML tools with well-known simulation environments [7]. Francesco Basile has used UML
to formally express system’s requirements, model the uncontrolled system and design the controlled one [8]. SRML is an
XML application that can describe the behavior for distributed simulation models. Every SRML project builds a XML file
defining the specification of typical elements, interactions and operating environment of simulation system, which makes
it much flexible when modeling with SRML [9]. In addition, many scholars have analyzed related work of M&S and proposed
new M&S platforms or framework. Tsai has proposed a service-oriented distributed modeling and simulation framework for
the development and evaluation of large scale distributed systems. This framework offers specification for modeling via lan-
guage specification, automated code generation, model checking and policy enforcement to allow different application archi-
tectures (layered, bus or peer-to-peer) to be simulated [10]. Among these M&S languages and platforms mentioned above,
most are not suitable for M&S of intelligent systems or intelligent behaviors. Further application will be limited if a M&S
language lacks a mature development environment for building simulation systems. And there needs massive developing
work on simulation application if executable programs cannot be directly generated automatically.
The M&S of intelligent systems is to apply intelligent theory to solve intelligent problems on the base of M&S theory and
method. The research scope of intelligent systems mainly includes expert systems, artificial neural network, evolutionary
computing, knowledge engineering and data mining [11]. For solving these complex problems, artificial intelligent languages
such as LISP, Prolog, Smalltalk, CLIPS, have been coming out one after another. LISP was originally specified in 1958 and cre-
ated as a practical mathematical notation for computer programs [12]. And it quickly became the favored programming lan-
guage for artificial intelligence research. Durand and Schwer have used LISP to describe the reasoning process about
incomplete qualitative temporal information in artificial intelligence and natural language processing applications [13]. Pro-
log is a general purpose logic programming language associated with artificial intelligence and computational linguistics
[14]. PY Zhao and XY Huang have carried out some experiments to present the realization of artificial intelligence in math-
ematic problem solving by using Prolog and found that artificial intelligence could be brought into mathematics through Pro-
log [15]. Smalltalk is first published in 1980 by Alan Kay to underpin the ‘‘new world’’ of computing exemplified by ‘‘human–
computer symbiosis’’ [16]. CLIPS (C Language Integrated Production System) is written in ‘‘C language’’ by NASA in 1984 and
is a public domain software tool for building expert systems which provides a complete environment for the construction of
rule [17]. Furthermore, CLIPS provides interfaces for other high-level languages. Zhang has presents a knowledge-based sys-
tem named ‘‘EFDEX’’ and used CLIPS 6.1 to perform intelligent functional design of engineering systems [18]. These lan-
guages can deal with most of intelligent problems, but problem still stands. These languages are mainly oriented at
knowledge-based reasoning application and perform not quite well in optimization problems. Their capability of solving var-
ious intelligent problems is limited due to the lack of a unified modeling language.
150 N. Li et al. / Simulation Modelling Practice and Theory 29 (2012) 149–162