Título

SIMULATION-INFORMED ARTIFICIAL NEURAL NETWORKS FOR CALCULATING ELECTRICAL RESISTIVITY OF LOW ALLOYED CU: CASES CUCRZR AND CUAGCR

SIMULATION-INFORMED ARTIFICIAL NEURAL NETWORKS FOR CALCULATING ELECTRICAL RESISTIVITY OF LOW ALLOYED CU: CASES CUCRZR AND CUAGCR

Autoria

ANJOS, PATRICK QUEIROZ DOS; GRILLO, FELIPE FARDIN; MACHADO, MARCELO LUCAS PEREIRA; QUARESMA, LUCAS DE ALMEIDA

ANJOS, PATRICK QUEIROZ DOS

Eu sou esse autor

GRILLO, FELIPE FARDIN

Eu sou esse autor

MACHADO, MARCELO LUCAS PEREIRA

Eu sou esse autor

QUARESMA, LUCAS DE ALMEIDA

Eu sou esse autor

DOI

10.5151/2594-5327-40969

Downloads

Resumo

LOW ALLOYED CU ARE USED IN A RANGE OF APPLICATIONS IN THE ELECTRICAL INDUSTRY THAT OVER THE YEARS REQUIRE METALS WITH OPTIMIZED PROPERTIES SUCH AS HIGH STRENGTH, PLASTICITY AND ELECTRICAL CONDUCTIVITY. AS MACHINE LEARNING IS AN INSTRUMENT CAPABLE OF ASSISTING IN THE DEVELOPMENT OF ALLOYS WITH OPTIMIZED PROPERTIES, NEURAL NETWORKS WERE USED TO MODEL THE ELECTRICAL RESISTIVITY OF LOW ALLOYED CU THROUGH A SET OF DATA. THE NEURAL NETWORK WITH THE BEST STATISTICAL EVALUATIONS WAS THE ONE WITH THE GREATEST COMPLEXITY IN TERMS OF NUMBER OF NEURONS AND HIDDEN LAYERS (NN 13-28) AND EXHIBITS COLLINEARITY BETWEEN THE TEST DATA AND THE PREDICTED DATA. SIMULATIONS WERE CARRIED OUT TO DEMONSTRATE THE VALIDATION OF NN 13-28 IN RELATION TO THE VARIATION IN ELECTRICAL RESISTIVITY AT TEMPERATURE IN THE CUCRZR ALLOY AND IN THE RELATIONSHIP BETWEEN AGING TIME AND ELECTRICAL CONDUCTIVITY IN THE CUAGCR ALLOY AT CONSTANT TEMPERATURE. BOTH SIMULATIONS PRESENTED PROMISING RESULTS IN RELATION TO NN 13-28, LITERATURE DATA AND EXPERIMENTALLY DETERMINED DATA.

LOW ALLOYED CU ARE USED IN A RANGE OF APPLICATIONS IN THE ELECTRICAL INDUSTRY THAT OVER THE YEARS REQUIRE METALS WITH OPTIMIZED PROPERTIES SUCH AS HIGH STRENGTH, PLASTICITY AND ELECTRICAL CONDUCTIVITY. AS MACHINE LEARNING IS AN INSTRUMENT CAPABLE OF ASSISTING IN THE DEVELOPMENT OF ALLOYS WITH OPTIMIZED PROPERTIES, NEURAL NETWORKS WERE USED TO MODEL THE ELECTRICAL RESISTIVITY OF LOW ALLOYED CU THROUGH A SET OF DATA. THE NEURAL NETWORK WITH THE BEST STATISTICAL EVALUATIONS WAS THE ONE WITH THE GREATEST COMPLEXITY IN TERMS OF NUMBER OF NEURONS AND HIDDEN LAYERS (NN 13-28) AND EXHIBITS COLLINEARITY BETWEEN THE TEST DATA AND THE PREDICTED DATA. SIMULATIONS WERE CARRIED OUT TO DEMONSTRATE THE VALIDATION OF NN 13-28 IN RELATION TO THE VARIATION IN ELECTRICAL RESISTIVITY AT TEMPERATURE IN THE CUCRZR ALLOY AND IN THE RELATIONSHIP BETWEEN AGING TIME AND ELECTRICAL CONDUCTIVITY IN THE CUAGCR ALLOY AT CONSTANT TEMPERATURE. BOTH SIMULATIONS PRESENTED PROMISING RESULTS IN RELATION TO NN 13-28, LITERATURE DATA AND EXPERIMENTALLY DETERMINED DATA.

Palavras-chave

Low alloyed Cu; Electrical Resistivity, Neural Networks; Simulation

Low alloyed Cu; Electrical Resistivity, Neural Networks; Simulation;