**4. Predictive maintenance using computational (fuzzy logic) decision support tool in preload dispatch**

An application of fuzzy logic is justified by the ability to anticipate the possibilities of making the pre-dispatch time of the load on the operational tasks of the equipment.

This study deals with the application of fuzzy logic to load dispatch, but with a particularity that is to perform the said pre-dispatch of load taking into account the technical state of the engines, evaluated by different variables related to maintenance. In the first part, the development of the fuzzy rules and of the whole procedure of inference is exposed, and in the second part, all the tests evaluate the maintenance and the technical state of the motors. This tool served as the basis for the resolution of the real problem of pre-dispatch of cargo to satisfy the rationalized methods of just in time of the thermal plant on the operational conditions of the equipment (**Figure 2**).

A system based on fuzzy logic, as shown in **Figure 2**, can have its action schematized by the following constituent elements: fuzzifier; rules, or knowledge base; inference, or logical decision-making, and Defuzzifier [13].

**Figure 2.** *System based on fuzzy logic. Source: Authors.*

#### *Maintenance Management*

In the first part the development of the fuzzy rules and of the whole procedure of inference is exposed and in the second part all the tests to evaluate the maintenance and the technical state of the motors. This tool served as the basis for the resolution of the real problem of pre-shipment of cargo to satisfy the rationalized methods of just in time of the thermal plant on the operational conditions of the equipment [14, 15].

The fuzzy system models the style of reasoning, imitating the capability to make decisions in an environment of uncertainty and imprecision. In this way, fuzzy logic is an intelligent technology, which provides a mechanism to manipulate imprecise information—concepts of small, high, good, very hot, cold—and that is able to infer an estimated answer to a question based on an inexact, incomplete knowledge, or not fully reliable information.

The development of a computational tool supports the load dispatch according to the location of motors and generators for thermal energy, analyzing the main thermoelectric generation variables for the entire predictive maintenance process.

All variables are inserted considering the intervals determined in the rules of inference as shown below.

The computational interface was useful for the search of some preselected characteristics to enable its implementation. **Tables 6–12** show such characteristics and respective purposes.

In this context, the following groups of information and data are abstracted: the input values, called crisp, the linguistic variables, and the fuzzy variables. The fuzzy logic is justified in the solution of this case study in function of the input variables with better representation in fuzzy sets. The variables due to the dimension of the universe of study were divided in 04 (three) and 03 (two) inputs and 01 (one) output, all independent of each other.
