Quantum Particle Swarm Optimization Algorithm And Its Application In CBM Production Forecast

This paper combined the CBM production forecast and particle swarm optimization.First,a new neural particle swarm optimization algorithm has been proposed.Than introduced a mixed neural network model and algorithm.Eventually established a neural network forecast model based on PSO by combining these two algorithms.In order to forecast the CBM production precisely,using this model to approximate the nonlinear mapping relationships between the geographical factors and the productivity.The specific research results of this paper are as follows.First,proposed a new neural particle swarm optimization algorithm.Particle swarm optimization(PSO)has been introduced in 1995.It has successfully got millions of scholars' attention by its search efficiency,quick convergence,fewer parameter settings and many other traits.The improvements of this algorithm are abundant.This paper considering from the encoding of the individual,proposed a multi-qubits probability amplitude PSO algorithm,which named MQPAPSO for short.In this method,the particles are encoded by the probability amplitudes of the basic states of the multi-qubits system.The rotation angels of multi-qubits are determined based on the local optimum particle and the global optimal particle,and the multi-qubits rotation gates are employed to update the particles.Therefore,the updating of every qubit can lead to the updating of all probability amplitudes of the corresponding particles,so that the optimization ability has improved significantly.Second,by combining the quantum computing and the neural computing,proposed a hybrid quantum inspired neural network model,named HQINN for short.In this model,the hidden layer is composed by quantum inspired neurons.The quantum weights are updated by quantum rotation gates,by which the adjustment of this model is more exquisite.Therefore the mapping ability of this network has been improved.To improve the global searching ability of the quantum neural network further,this paper used MQPAPSO to optimize the weights of HQINN,proposed a new model named MQPAPSO-HQINN.The experimental comparison and analysis showed that both the approaching ability and the generalization ability of this model have been improved significantly.The effectiveness of this method has been verified.At last,in the aspect of the algorithm application,selected the datum of 8 CBM wells from JinCheng County,Shanxi Province,China.Established the CBM production forecast model with MQPAPSO-HQINN network,and compared with other algorithms and method.The experiment results showed that forecast precision of MQPAPSO-HQINN network model was better.Also,this paper made the sensitivity analysis of 5 geographic factors that affected CBM production.The different effects of five geographic factors on CBM production forecast have been analyzed.