PSI - Issue 39

Nabam Teyi et al. / Procedia Structural Integrity 39 (2022) 608–623 Author name / Structural Integrity Procedia 00 (2019) 000–000

614

7

prediction. The most basic design is a feedforward neural network with a three-layer RBF. The first layer contains the network’s inputs, while the second layer contains a number of random-basis-function nonlinear activation units, and the third layer contains the network’s final output. Gómez et al. (2016) used RBF and Wavelet Packets ANNs to detect cracks in a rotating shaft. The shaft rotated under its own weight and had several different types of cracks. There were nine flaws in the shaft (with depths from 4 percent to 50 percent of the shaft diameter). Its parameters were optimised to maximise success rates. These probabilities were near 100 percent on the ‘Probability of Detection’ curves. Castejón et al. (2014) MRA to search for patterns in vibration data from shafts with a transverse crack. The depiction of a signal in more than one resolution/scale is known as MRA. The feature vector of each fault condition was fed into an ANN RBF that performed supervised learning. Online monitoring and diagnostics were provided by MRA and RBF. The results indicated that ANN and wavelets is effective to diagnose rotor problems. 4.2. Fuzzy Logic Fuzzy logic is a type of many-valued logic in which the truth value of variables can be any real integer between 0 and 1, in logic. It is used to deal with the concept of partial truth, where the truth value can be somewhere between true and false. The truth values of variables in Boolean logic, on the other hand, can only be the integer values 0 or 1 (Fig. 7.).

Fig. 7. Fuzzy logic.

Behera et al. (2018) developed an accurate hybrid intelligent model in an aluminium beam construction with free free boundary conditions by investigating vibration properties using mathematical models and experiments. Combinations of Fuzzy logic, GA and Rule-based technique made up the hybrid intelligent model. They concluded that crack depth generated local flexibility at the crack position, affecting structural integrity sensitive measures such relative mode shapes and natural frequency. Jena (2018) investigated a FRC beam with variable fibre direction for a fixed-simple end condition using a non-destructive neural-fuzzy hybrid technique. The quick reaction (natural frequency) had changed due to variations in crack depth, crack location, and fibre orientation. This technique could be used to monitor the condition of FRC structural beams. Fuzzy logic was used by Talekar et al. (2016) to analyse the vibration of a broken beam. When compared to other membership functions like triangular and trapezoidal, the fuzzy controller constructed with a Gaussian membership function was more exact. Gowd et al. (2018) did a comprehensive comparison between ANN and fuzzy logic approaches for crack detection in a beam-like structure. 4.3. Genetic Algorithm and Evolutionary Algorithm A GA is a search heuristic based on Charles Darwin’s natural selection hypothesis. This algorithm mimics natural selection, in which the fittest individuals are chosen for reproduction in order to create the following generation’s children. As in nature, an individual that is best suited to its surroundings is formed through generations of gradual change. The method begins with a population of individuals who represent potential problem solutions. After numerous rounds, this population evolves into distinct populations (generations). As a solution, the algorithm returns the population’s fittest (if not best) individual. The algorithm evaluates, selects, and recombines population members in order to generate offspring and build new populations. A problem-dependent fitness function is used to evaluate each individual encoding a proposed solution. Crossover and mutation are used to simulate reproduction. Mutation