Extracting polybrominated biphenyls (PBB) from electronic and electrical equipment is of high concern due to RoHS directive. PBBs were so toxic for human and environment, they are applied largely in electronic and electrical equipment as flame retardant. In this thesis, a new method was developed to predict the optimal conditions of semivolatile organic polybrominated biphenyls (PBB) extraction from plastic for Enviromental protection. A feed forward type of artificial neural network (ANN) model design was used to investigate the effects of four independent variables, namely, the ratio of solvent, stirring speed (rpm), extraction temperature (°C), extraction time (h) on the response, the acquired ratio of PBB. The independent variables were coded at four levels and their actual values selected on the basis of results of single-factor experiment. The model was initially trained by the analytical data with function approximation principle in MATLAB environment to reveal the real engineering world of extract process. Then dimensions of the trained result were reduced from n-D to 2D, In which, a visual contour plot and simulated curve were displayed, an optimal extract processing is achieved with 9.654% maximal acquired ratio of PBB. In order to validate the method, at the optimal point, the simulated result generated by proposed model was checked with the real experiment, it founded they kept a good agreement with each other. Thus proves that the mathematical model developed for resolving the PBB extraction from plastics is very effective and accurate. It is also a useful tool to reveal the real parameters effect on productivity. Modern computational and experimental tools have matured to a stage where they can provide substantial insight into engineering processes involving extracting some contents from herbs, vegetable oil, rubber, plastics etc, and the successful application of optimization design can help to improve the process productivity. Extracting the flame retardant as polybrominated biphenyls (PBBs) from electronic and electrical equipment is of high concern due to RoHS directive [1]. PBBs were so toxic for human and environment, it is applied in electronic and electrical equipment as flame retardant. When the EE wastes disposed unsoundly at the end of life, some hazardous substances emit into air, water, soil which are a strict threaten to human. So to determine such substances for EE industries has a profound meaning. Many methods to determine such hazardous substance such as GC-MS, ICP-OES, AES, IC etc. are developed[2–4], however, there is a very high requirement for sample preparation. Soxhlet extraction is a good digest technology to prepare samples, in this thesis, a new method was developed for predicting the optimal digest conditions of semivolatile organic compounds PBB in plastic with soxhlet extrat technology. During experiment, Uniform design is accepted to array data due to the uniformation consideration. Conventional optimization design was to hold one but all other variables as constant while methodically changing one at a time which is called one variable at a time or OVAT [5]. But the major pitfall in this procedure is that it cannot quantitatively explore the interaction among all variables and does not describe the net effect of the various combined conditions on the response [6]. The rapid and continuous development in processing design will require a new proposal to meet goals for increased performance, robustness and visualization. To date, the majority of efforts in optimization of extraction process have relied on gradient-based search algorithms, polynomial-based response surface methodologies (RSM), local gradient-based method and so on [7]. However, the challenge of these approaches are that it renders to search insufficiently due to the objective functions discontinuous over the broad design space or too resource-intensive due to unrestricted “brute force” search schemes. So here, based on global optimization problems especially for multidisciplinary ones, a matrix method was developed to predict the extract process of PBB from plastics which are compositions of packaging, sealed, adiabatic, insulating materials in electronic and electrical equipment (EEE). By establishing a functional artificial neural network (ANN) which maps the relationship between the response and variables, using a global optimization line-up competing algorithm (LCA) to train the network [8], simulation curve and the optimal contour plots were produced in 2-D plane. The model would not only serve as a visual aid to have a clearer picture about the effect of different variables on the responses in the form of animation but also enable to locate the region where the properties are optimized. It is also possible to predict the combination of independent variables which will result in optimal acceptance. It helps to understanding how the result is achieved and increases the reliability to the result.