Design of a novel multi-objective mixed stochastic optimization based neuro-controller for reactor coolant pressure control system

In this paper,a novel multi-objective mixed stochastic optimization (nmomso) based neuro- controller (nc) is synthesized for multi-input multi-output (mimo) reactor coolant pressure control system (rcpcs) of pressurized heavy water reactor (phwr)-type nuclear power plant (npp). in rcpcs,a new mimo modified particle swarm optimization algorithm (mpsoa) based nonlinear adaptive feedforward neural network (afnn) model (mimo mpsoa-afnn) of the reactor coolant pressure system (rcps) is developed mapping five inputs and two outputs. a highly intelligent nmomso based nc is designed using afnn and mixed stochastic optimization techniques (msot) in a mimo framework. the nmomso is comprised of five multi-input single-output (miso) afnn optimized by msot. three different stochastic techniques are used for the optimization of weight matrices of five miso intelligent networks. a modified particle swarm optimization algorithm (mpsoa) is used for the optimization of two intelligent miso networks parameters for two feed valve positions. an ant colony optimization algorithm (acoa) is implemented for the optimization of two intelligent miso networks for two bleed valve positions and a bee colony optimization algorithm (bcoa) is used for the optimization of one intelligent miso network for one spray valve position of rcpcs. in the proposed nmomso based nc,a multi-objective problem is formulated based on five parallel operating miso intelligent networks using new configuration of reactor coolant and steam pressure signals. the proposed mimo mpsoa-afnn model and nmomso based neuro-controller of rcps is designed in matlab and a graphical user interface (gui) is developed for variables transfer and simulations in visual c. the performance of model based novel neuro-controller is compared with conventional coupled controller (ccc) using pid controlled reactor coolant pressure and on-off controlled surge tank level and found remarkable. © pakistan academy of sciences.