Particle Swarm Optimization Algorithm And The Application In The Optimized Questions Of The Stock Market Forecasts

This article mainly introduced recent year proposed newly the optimized algorithm, Particle Swarm Optimization. This algorithm succinct, easy to realize, the understand easily, does not need majorized function merits and gradient information and so on. PSO restrains quickly, specially in algorithm early time, but also has the precision to be low, easy to disperse and shortcomings and so on . If the celerating factor, maximum speed and so on are too big, particle swarm possibly misses the optimal solution, the algorithm does not restrain; But in restraining situation, because all particles fly to the optimal solution direction, The particles trend identical (has lost multiplicity), causes later period the convergence rate obviously to slow down, when simultaneously the algorithm restrains to certain precision, is unable to continue to optimize, can achieve lowerly precision also compared to Genetic Algorithm. Therefore this article has made the improvement to the particle swarm optimical algorithm, proposed a grouping grain of subgroup optimizes the algorithm. In a grouping grain of subgroup optimize algorithm , a grain of subgroup divide into several small groups, each small group will have the different evolution parameter, and each small group will evolve separately, In gap certain time carries on the group the variation and the reorganization operation, And carries on a grain of subgroup during reorganization to various groups parameter to optimize, The simulation result showed that compares the ordinary grain of subgroup algorithm, regardless of has the enhancement in the convergence rate in the perusal and in the operation conveniences.Because this algorithm succinct, easy to realize, the understand easily, does not need majorized function merits and gradient information and so on., therefore the algorithm p widely is applied, In 2007, Hassan and Nath proposed Hidden Markovian model and the ANN algorithm, The Genetic Algorithm combination model - AGHWA model, and uses in the stock price forecasting, the discovery forecasting result is optimistic, but because The Hidden Markovian model matrix operation adds on The Genetic Algorithm the code and the decoding causes entire process suitable complex. In view of the fact that the PSO merit, this article proposed optimized the algorithm with a grain of subgroup to come to carry on the optimization to the hidden Markovian model initial parameter, and has carried on the forecast using the model to the stock price, The simulation result showed after optimizing the model has the quite good performance.