Diagnostyka, 20xx, Vol XX, No X e-issn

S1	S2	…
Sn
D1	R11	R12	…	R1n
D2	R21	R22	…	R2n
…	…	…	…	…
Dm	Rm2	Rm2	…	Rmn

In this case, the inference system will be built on the basis of the model proposed by Mamdani, according to the MIMO scheme (Multiple In – Multiple Out), i.e., will contain several inputs and several output variables. The values of B _j will be used as input variables, and the values of A_i will be used as output variables.

Thus, the fuzziness of the variable B _j may be due to the subjectivity of a person's opinion about the presence of a certain diagnostic symptom. To determine the fuzzy value of the variable B_j, depending on the specified parameters, one should introduce such linguistic variables as OE – “assessment of the presence of symptom S _j is objective” and QoS - "The quality of the communication channel is good. "These linguistic variables take values from the previously introduced term-set CON . It should be noted that either only OE or only QoS can affect the value of B_j at the same time. In cases where the quality of the communication channel of the remote diagnostic system or the objectivity of the assessment of the presence of the symptom S_j do not affect the value of B_j , the value of vt is assigned to the variable B_j. In all other cases, the value of B_j should be equal to OE or QoS , depending on which of these linguistic variables is used, i.e. depending on what exactly affects the objectivity of assessing the presence of a symptom S_j. It should be noted that, B_j as a rule, has the value vt. If information about the presence of a defect B _j has not been received by the system, B_j is assigned the value un.[5]
In the general case, it is not enough for a diagnostic system to know the value of one linguistic variable B_j , i.e. when information about the presence of one diagnostic symptom enters the diagnostic system, there is a lack of initial information, which has a significant impact on the reliability of the decision made about the type of defect. In this case, the system should be able to request an assessment of the presence of symptoms closest to the original, in terms of belonging to one type of defect. Thus, it is possible to form a complete set (vector) of input variables {B_j}, j=1, ..., n, used to decide on the type of defect.[6]
To implement a fuzzy inference system, it is necessary to organize a rule base based on the values of the linguistic variable R_ij, which will contain “if-then” type inference rules that express experts’ knowledge about the belonging of diagnostic symptoms to a specific type of defect. Based on these rules, the system will decide on the type of defect.[3]
To reduce the number of calculations in the diagnostic system, the matrix M describing the relationship between symptoms S_j and defects D_i can be divided into submatrices that will contain defects grouped according to certain characteristics (for example, defects in transmission systems, defects in switching and routing systems, etc.). P.).
In the diagnostic system, based on a survey of experts, a term set CON is formed , the elements of which have membership functions. Also, based on a survey of experts in the system, the following matrix M was formed and graphs of the membership functions of linguistic variables (Fig. 1) were constructed, which characterizes the relationship between symptoms and defects:[14]