Predictive model for detection of imminent failures in industrial electric machines

• Grant number: 19/09086-1
• Support Opportunities: Research Grants - Innovative Research in Small Business - PIPE
• Start date: March 01, 2020
• End date: October 31, 2020
• Field of knowledge: Engineering - Electrical Engineering
• Principal Investigator: Nicolas Goulart Silva
• Grantee: Nicolas Goulart Silva
• Company: Dipsie Engenharia e Software S/A
• CNAE: Instalação de máquinas e equipamentos industriais ; Desenvolvimento de programas de computador sob encomenda ; Desenvolvimento e licenciamento de programas de computador customizáveis
• City: Vinhedo
• Pesquisadores principais: Vitor Augusto Silva Bueno
• Associated research grant(s): 20/13127-2 - Software platform prototyping for management and control of industrial plant asset maintenance integrated with predictive maintenance models, AP.PIPE

Abstract

With the increasing need to be more competitive, industries are forced to incorporate technologies into their processes to optimize costs and increase productivity and efficiency. Companies are now plugging their assets to make the most of their data to create a new, powerful business value. This great movement in the industry is known as the digital transformation, Industria 4.0 and Industrial Internet of Things (IIoT). Connecting devices is only the first step. The actual value is in the information generated through the data that is transmitted from these devices, such as the business insights that this data can enable. Maintenance costs correspond to most of the total operating expenses of manufacturing and / or process manufacturing plants. Depending on the type of industry, maintenance costs can represent 15% to 30% of the cost of goods produced. The systems and solutions that are commercialized today, which act in the treatment of the presented problem, make use of static mathematical models, which based on the recommendations of the manufacturers and norms, process data collected manually to generate the Insights, but in its great majority, not precise. The proposal of predictive models for maintenance has as great innovation the use of computational intelligence techniques that enable the ability to analyze in real time, parameters and real variables of machine operation as well as to learn several reading patterns and to make a comparison between the data collected with a database contemplating various operating states of the machine. (AU)