MONITORAMENTO DE FALHAS EM SISTEMAS DINÂMICOS UTILIZANDO UM MÉTODO DE ANÁLISE DE DADOS BASEADO NO ALGORITMO DE SELEÇÃO NEGATIVA

This work presents the application of a method for monitoring and diagnosing failures in mechanical structures based on the theory of vibration signals and on Artificial Immune Systems to assist in data processing. It uses the Negative Selection Algorithm as a tool to identify fault samples extracted from the laboratory simulated signals of a dynamic rotor. This methodology can help mechanical structure maintenance professionals, facilitating decision-making. The data set used in the processing of the intelligent system was generated through experiments. For normal (base-line) conditions, the signals of the rotor in free operation were used, that is, without the addition of unbalance mass, and for the fault conditions, unbalance masses were added to the system. The results are satisfactory, showing precision and robustness.

Pengenalan Pola Angka Menggunakan Pendekatan Optimisasi Sistem Kekebalan Buatan (Artificial Immune System)

The general election in Indonesia itself still experiences technical and non-technical problems where the technical problems occur in the recapitulation of votes from sheet C1 which are still incorrectly inputted and done manually. The problem occurred with the difference in the uploaded C1 data and the data in the KPU Situng and the C1 sheet uploaded was blurry, unclear, sheet C1 which was crossed out or folded in the KPU Situng. The purpose of this research is to reduce errors in data input and change the work that is done manually to the system, create a number pattern recognition system using an Artificial Immune System optimization approach, test and analyze the work of the system by taking into account the level of accuracy, preciseness and speed in recognize number patterns. The system created to applies an artificial immune system optimization approach with the Artificial Immune System using the Randomized Real-Valued Negative Selection Algorithm algorithm.

A new approach experimental to diagnosis of the failures in mechanical structures using the artificial immune algorithm with negative selection / Uma nova abordagem experimental para o diagnóstico das falhas nas estruturas mecânicas utilizando o algoritmo de imunidade artificial com seleção negativa

In this paper we present a new experimental approach to diagnose failures in mechanical structures using as decision tool an artificial immune algorithm with negative selection. This method is divided into two modules, and the acquisition and data processing module and analysis, detecting and characterizing flaws module. The module for data acquisition and processing of the experimental apparatus is constituted as sensors and actuators, so as to capture the signals in the structure and store it in the computer. From the signal acquisition executed if the negative selection algorithm to identify and characterize flaws in the structure. The main application of this methodology is to assist in the inspection process of mechanical structures in order to identify and characterize the flaws, as well as perform the decisions in order to avoid accidents. To evaluate the proposed methodology, experiments were performed in the laboratory where a real signs database was captured in a structure of the beam type, made of aluminum. The results obtained in the tests show robustness and efficiency when compared to literature.

An improved real-valued negative selection algorithm based on the constant detector for anomaly detection

Compared with the traditional negative selection algorithms produce detectors randomly in whole state space, the boundary-fixed negative selection algorithm (FB-NSA) non-randomly produces a layer of detectors closely surrounding the self space. However, the false alarm rate of FB-NSA is higher than many anomaly detection methods. Its detection rate is very low when normal data close to the boundary of state space. This paper proposed an improved FB-NSA (IFB-NSA) to solve these problems. IFB-NSA enlarges the state space and adds auxiliary detectors in appropriate places to improve the detection rate, and uses variable-sized training samples to reduce the false alarm rate. We present experiments on synthetic datasets and the UCI Iris dataset to demonstrate the effectiveness of this approach. The results show that IFB-NSA outperforms FB-NSA and the other anomaly detection methods in most of the cases.