The Topology Optimization Of Oil & Gas Gathering And Transportation Network Based On The Neural Network Techniq

The design of optimization to oil-gas gathering and transportation system can lead to remarkable economic result. Since the costs of connecting mediums between any nodes in the net are very expensive, generally from 50 to 80 percent of the whole system costs, it's important for the design of topology optimization to reduce the costs of the whole network system. Scholars both at home and abroad have mostly adopted traditional optimization theories such as the classification optimization method about the topology optimization to the oil-gas gathering and transportation system for a long time, and modem optimization algorithms including neural network technique begin to be applied in oilfield exploration decision-making as a result of its advantage to traditional optimization methods.In the face of this situation, the mathematical model of the topology optimization for star gathering system is established, and solved with the neuron network method in this paper, which seeks a new way to solve topology optimization to oil-gas gathering and transportation pipeline network in oilfield.This paper sets up the optimization mathematical model with the crude transport costs as objective function, unique relation between wells and stations, enough crude tank volume on stations, and locality space limitation as constraint conditions. The model is discretized for the convenience of solving the problem. When the neuron network method is applied, Hopfield neural network is chosen as design proposal of optimization methods on the basis of neuron network's fundamental, then its topological and operational principle is discussed in detail. Subsequently, the energy function is constructed based on the penalty function method, and its structural parameters are deduced, thus its corresponding power system equation is obtained. Consequently, the original optimization mathematical model is translated into neural network structure, and proven strictly its stability in math. In the end, we program an oil & gas gathering and transportation network topological optimization design software with C++. Considering the shortcoming and deficiency that Hopfield neural network is apt to be trapped in local minima, we try to improve on this method with genetic algorithms helpfully.The calculated instances indicate that the network model is stable, and the result solved by the neural network method is better than that by traditional optimization methods, thus validating the correctness and validity of the new optimization method. The software can optimize and plot the pipeline-station topological graph, and provide the convenience to guide the conceptual design and production run in oilfield.