(512b) The Use of Intelligent Mathematical Modeling and Optimization Agents Towards Process Intelligence

Process, manufacturing, and service industries currently face a large number of non-trivial challenges ranging from product conception, going through design, development, commercialization, and delivering in a customized market environment. These industries can benefit from integrating new technologies in their day-by-day tasks, thus gaining profitability. A crucial job consists of how decision support systems can improve effective problem management by using new technologies. It is also clear the importance of mathematical modeling that represents diverse aspects of the real world, for example, in the modeling of the actual system and subsystems, their interaction, and their dynamic behavior.

This work focuses on the integration of mathematical models, mathematical functions, and systems of equations, as well as their temporal-space scales. Besides, the proposed integration follows the operations research architecture, which makes use of mathematical expressions to support decision-making. Moreover, the present approach aims to automate the mathematical modeling task based on knowledge management and intelligent agents for improving the whole decision process ranging from problem conception to solution implementation.

The use of semantic models in the form of formal knowledge models allows capturing and exploiting domain knowledge, human knowledge, and expert knowledge. The present work makes use of a process system ontology, a mathematical modeling ontology, and an operations research ontology. First, the Enterprise Ontology Project strictly follows ANSI/ISA 88 and 95 Standards, Supply Chain Management books, Life Cycle Environment approaches, among others. This ontology is in charge of formalizing the process domain and those functionalities within process systems engineering, considering operational, tactical, and strategic functions. Next, Mathematical Modeling Ontology captures and formalizes algebraic and logic knowledge. Thus, this ontology handles the mathematical equations used to obtain explicit-concept-object oriented mathematical modeling. Its main task is to support the correct structure of mathematical expressions, formalize the relationship among mathematical elements and mathematical operations, and finally conceptualize mathematical models. Finally, Operation Research Ontology allows capturing and structuring the elements involved in the creation of decision support systems. As a result, one obtains a system of equations in the form of objective function and constraints. It is essential to point out that the proposed framework is capable of managing data and information to align the data requirements to the model. In this sense, semantics allows us to quickly realizing tasks such as data gathering, data conciliation, data structuring, and data transferring.

The proposed framework uses Java, Python, and Jython as main programming languages. Within this program code, OWLAPI allows managing formal knowledge models (Ontologies in OWL format), exploiting their knowledge in the form of classes, relations, data properties, axioms, and domain rules found within them. Next, the framework makes use of a set of agents, having the functions of problem definition, model development, model analysis, model solution, and problem conclusion. Besides, the design of intelligent agents provide additional reasoning capabilities and create case rules for model creation, problem application storage, and usage.

A case study is performed using a scheduling modeling and solution example. Specifically, a systematic approach to scheduling model selection and solution implementation is achieved, thus enabling the bridge between theoretical developments and industrial practice. The integration and management of data and information flow among different process operation levels is a common topic in the planning and scheduling areas, having a close relationship with the field of the supply chain. To improve the management of data and information flow among organizations, tools based on knowledge models seem to be a promising approach.

As a result, the decision-maker obtains substantial support in the modeling process, reaching the right solutions with limited knowledge on the solution techniques. An automated modeling platform is created, which encompasses mathematical models, their formal representation, and the semantic model of the represented system. Besides, the use of intelligent agents specifically designed provide additional reasoning capabilities and build case rules for model creation, problem application, storage, and usage. Future work contemplates the setting-up of this framework for supporting the decision making of tasks at different levels or process areas, ranging from process definition to process intelligence as part of a Wide Intelligent Management project.

References

International Society for Measurement and Control. (1995). Batch control. Part 1. Models and terminology. International Standard, ISA Society.

International Society for Measurement and Control. (2003). Batch control. Part 3. General and Site Recipe Models and Representation. International Standard, ISA Society.

International Society for Measurement and Control. (2008). Enterprise-Control System Integration. Part 1. Models and terminology. International Standard, ISA Society.

234. Muñoz et al. (2012). Ontological framework for enterprise-wide integrated decision-making at operational level. Comp. & Chem. Eng., 42, 217–234.
235. Muñoz et al. (2013). Integration of enterprise levels based on an ontological framework, Chem. Eng. Res.&Des., 91, 1542-1546.
236. Muñoz et al. (2015). , Operations Research Ontology for the Integration of Analytic Methods and Transactional Data. In Trends and Applications in Software Engineering, 139 - 145.

Tanvir, M. (2012). Semantics: Advances in Theories and Mathematical Models. InTech.

711. Varma et al. (2007). Enterprise-wide modeling & optimization - An overview of emerging research challenges and opportunities. Comp. & Chem. Eng., 31, 692–711.

Research Interests:

Edrisi Muñoz: Computing and Systems Engineering, Process Systems Engineering, Process Knowledge Management, Process Design and Development. Information Management and Intelligent Systems

José Miguel Laínez-Aguirre: Computing and Systems Engineering, Bussines and Supply Chain Optimization, Process Systems Engineering, Process Knowledge Management, Process Design and Development. Information Management and Intelligent Systems.

Elisabet Capón-García: Computing and Systems Engineering, Process Systems Engineering, Process Knowledge Management, Process Design and Development. Process Control Systems. Information Management and Intelligent Systems

Luis Puigjaner: Computing and Systems Engineering, Process Systems Engineering, Process Knowledge Management, Process Design and Development. Information Management and Intelligent Systems.