Analog Circuits
Description
Each file, named File nameCircuit.txt is used for training and testing a system (SVM in our case) which performs the classification of the faults. Therefore It contains a set of data formatted for being the input of a classification software, and containing the following lines: n1 n2 varying components: c1 c2 ...... cNC mag_1,1 pha_1,1 mag_1,2 pha_1,2 ........ mag_1,nf pha1,nf FC_1 mag_2,1 pha_2,1 mag_2,2 pha_2,2 ........ mag_2,nf pha2,nf FC_2 ............. mag_Ne,1 phaNe,1 mag_Ne,2 pha_Ne,2 ........ mag_Ne,nf pha_Ne,nf FC_Ne ------------------- The exact meaning of any element is: n1 = number of input lines (dimension of the input vector) n2 = number of output lines (for this application = 1, because it is a classification problem and we have just one output integer which assumes the value of various classes) mag_i,j = it is the magnitude of i-th sample of the frequency response (the absolute value of the complex frequency response term at that given frequency) mag_i,j = it is the phase in rad of i-th sample of the frequency response (the phase of the complex frequency response term at that given frequency, that arctan(ImF/ReF) in rad) nf = it is the numer of sampling frequencies Ne = it is the number of examples FC_i = it is the correct Fault Class for that sample; the class 0 is "no faulty", otherwise it is an integer which represents the component which is out of tolerance (or, in case of ambiguity groups with a dim > 1, it represents an ambiguity group). as consequence to previous format we can also say that: n1 = 2 x nf (magnitude and phase each sample) mag_i,j are abs values with no normalization in the file and so they are always positive pha_i,j are the original phases in rad, so they are values ranging in the interval [-pi pi], not normalized in the file