Neural Network Algorithm for Accuracy Control in Modelling of Structures with Changing Characteristics

The Use of Fuzzy Clustering in Solving Problem in Predicting the Durability of Corrosive Structures

“…The subject area of work is predicting the durability of a structure with changing geometric characteristics, which functions in an aggressive external environment (direct task) ( fig. 1) [11,20,21].…”

“…The considered error will depend on the following factors: initial geometric characteristics of the element (area 0 A and perimeter 0 P ), initial 0 σ and ultimate [ ] σ voltages in it, parameters of the corrosion process and the maximum permissible error value [ ] ε . Therefore, the error of the numerical solution of the differential equation describing the corrosion wear (4) can be represented as a function of several variables: [11,20,21]. As a control module, neural network algorithms are offered that have proven themselves quite well.…”

“…As a control module, neural network algorithms are offered that have proven themselves quite well. However, the use of neural networks has its advantages and disadvantages [9,20]. Neural networks (NN), as you know, are universal approximating systems and are based on a specific mathematical apparatus.…”

“…In addition to neural networks, combined approaches using genetic algorithms and NN are proposed for solving such tasks [8,10,20]. As a rule, the use of one or another approach is determined by the class of tasks being solved and computational costs that are spent on obtaining a solution [21].…”

PLC Implementation of Piecewise Affine PI Controller Applied to Industrial Systems with Constraints

“…This difficulty can be circumvented by finding explicit control laws using controlled invariant contractive sets as the solution of a multiparametric programming problem which results in a piecewise affine (PWA) law over a polyhedral set satisfying the constraints. Additionally, new controller techniques such as predictive (Short and Abugchem 2017), fuzzy (Khan et al 2015) and neural (Zelentsov and Denysiuk 2019) are not as frequently used in the industry as PID.…”