Application of probabilistic networks for decision support in power system analysis

Application of Bayesian method for electrical power system transient stability assessment

“…Electrical power systems is not an exception: Bayesian inference is applied for risk analysis of power components and systems [27], analysis of unavailability due to aging failures in power systems [28], some research work based on Bayesian networks theory [29,30] and others. All aforementioned research work is based on Bayesian approach for estimating the parameters of a stationary process when more statistical data become available.…”

Detecting Small Fluid Leaks in Pipe-Type Cable Installations

“…The methodology developed by the authors and presented in [1] and [2] uses an expert system employing neural and probabilistic networks for detection of dielectric fluid leaks. Section II briefly reviews the philosophy behind the original approach, describes the main module for modeling the pressure-temperature relationship, and provides an overview of the probabilistic network as a decision-making tool.…”

“…The decision calculation is the final stage of the detection process and is discussed in [1] and [2] and briefly reviewed in this chapter. The values of the markers have to be properly combined in order to produce one Boolean value, which will indicate the presence or absence of a leak.…”

“…Later, the fluid thermal energy is dissipated to the surroundings as heat. According to the laws of thermodynamics [5], the increase in thermal energy manifests itself in the proportional rise of the temperature (2) where is the number of molecules, is the number of degrees of freedom, and is the Boltzmann's constant. The behavior of the fluid in the pipe may be described by the Van der Waals equation.…”

“…The valve and pump characteristics constantly change, different conditions are present along the pipe length, and sensor readings are not always representative for the whole system. These false leak indicators are accounted for by the decision module using probabilistic networks discussed in [2] and briefly described in Section II.…”