OBRABOTKAMETALLOV TECHNOLOGY Vol. 24 No. 3 2022 It is apparent that the specifi c wear rate increases with the normal load, interface temperature, and sliding velocity. However, the increase in specifi c wear rate will become more noticeable at higher process parameters. The normal load followed by sliding velocity and interface temperature can be seen as most signifi cant parameters affecting the wear rate. This can be also confi rmed by the higher exponent value for the normal load followed by for sliding speed and then for interface temperature in Eq. (4). This study fi nds that wear is prominently affected by the normal load, especially at higher values of interface temperature and sliding speed. Artifi cial neural network (ANN) is a computational technique that can model relationships between input parameters and output responses. A typical MLP architecture which is most commonly used is shown in Fig. 3. MLP is characterized by three different layers namely input layer, hidden layer, and output layer, which consist of an interconnected group of artifi cial neurons. The number of neurons present in the input layer and output layer is equal to the number of input variables and corresponding output values. To predict output with higher accuracy, training of the developed network is essential. In the training process of a model, the synaptic weights of the network are modifi ed in an orderly fashion to attain the desired output. Most used training algorithms is the error backpropagation algorithm. For a typical ANN algorithm, at the fi rst step the weights and thresholds are initialized. Then, the output of each neuron is calculated Fig. 2. 3-D plots showing specifi c wear rate varying with: a – Normal load and interface temperature; b – Normal load and sliding speed; c – Interface temperature and sliding speed a b c Fig. 3. Typical ANN architecture
RkJQdWJsaXNoZXIy MTk0ODM1