Problems of Improving the Diagnostic Systems of Marine Diesel Generator Sets Iryna Hvozdeva National University “Odessa Maritime Academy” NU “OMA” Odesa, Ukraine onopchenko.im@gmail.com Vitalii Budashko National University “Odessa Maritime Academy” NU “OMA” Odesa, Ukraine bvv@te.net.ua Volodymyr Myrhorod National University “Odessa Maritime Academy” NU “OMA” Odesa, Ukraine v.f.mirgorod@gmail.com Valerii Shevchenko National University “Odessa Maritime Academy” NU “OMA” Odesa, Ukraine vash4891@gmail.com Abstract— The problem of transition from marine diesel- generator sets operation by the assigned useful life to that by their technical condition is considered. This Article includes a review of control and diagnostic systems of marine diesel-generator sets, establishes their features, analyzes the methods of trend control used and shows their identified shortcomings. A regression statistical model linking the temperature conditions and the load change is offered. The trends of deviations of time series of the temperature condition recorded parameters data from the statistical model in marine diesel generator sets long operation are analyzed. It is proposed to improve the methods of trend control and analysis for use in the systems of technical diagnosing of marine diesel generator sets, in order to increase their efficiency by taking into account the natural wear-and-tear. Keywords— diesel generator set, diagnostics, trend analysis, feature space, statistical model I. Introduction The current state and development prospects of power plants are characterized by an increasing level of intensity of energy conversion processes in the conditions of strict requirements to the technical and environmental safety. The need to improve technical and economic efficiency in a competitive environment contributes to the introduction of new processes and technologies that require continuous and objective monitoring at all stages. Therefore, in parallel with the increase in the intensity level of energy conversion processes in power plants, the intensity of information processes of continuous assessment and monitoring of their technical condition throughout the life cycle cumulatively increases. A separate and quite specific class of power plants are diesel generator sets (DGS), which are part of modern marine electric power systems (MEPS). Distinctive features of such technical facilities are autonomous operation, high maneuverability (continuous mode change), limited energy resources, high risk of emergency situations, high level of responsibility, limited control resources and other factors caused by operating conditions. The problematic issue of life DGS cycle control is to ensure the required operational reliability. The solution of this problem is achieved by the transition to the operation of MEPS according to their technical condition, which, in turn, requires improvement of both methods of monitoring of the MEPS current state and those for their technical condition long-term forecasting. A number of mutually contradictory requirements to diagnosis depth, decision-making speed and diagnosis and prognosis accuracy are placed on the methods of diagnosis used in the ship's DGS maintenance and repair planning. One of the main tasks of forecasting the technical condition of ship's DGS is the detection of dangerous defects at the initial stage of their development in order to prevent the occurrence of failures. Technologies of diagnosing and monitoring of technical condition of MEPS develop in two directions, the first of which is aimed at development of automatic systems for safe management of objects, and the second – at improvement of maintenance and repair processes. The key point in the convergence of the two directions of diagnosis is the necessity to install the stationary monitoring systems on the most important units to solve the problems of prevention of emergency situations arising from control errors, without premature shutdown of the unit. The time available for decision-making on such systems is several times longer than on emergency protection systems, which allows supplementing the system with various monitoring and prediction algorithms, while the amount of diagnostic information used increases several times. This, in turn, permits increasing the reliability of the ship's DGS technical condition forecasts. Methods of trend control used in modern ship systems of technical diagnostics (STD) allow establishing just the fact of deviation of the registered parameters from their nominal values [1–4], but do not allow giving a forecast assessment of