Fish-swarm Optimization Algorithm And Its Application In Turbine Fault Diagnosis

In recent years, artificial fish-swarm algorithm has gradually become a research hotspot intelligent algorithm.It has many advantages such as parallel, simple, overall and so on.lt also has achieved good result in the applications in power system, signal processing and parameter estimation.In this paper, the basic theory is based on the artificial fish-swarm algorithm combined with the probabilistic neural network which has network training simple, reliable, strong jamming of ability.Based on the analysis of two intelligent algorithms,selects smoothing factor of probabilistic neural network as the breakthrough point to the combination of two algorithms.Searching for the best smoothing factor by the use of foraging behavior, clustering behavior, rear-end behavior of fish-swarm algorithm.Then applied to probabilistic neural network training. Simulation verified the algorithm for pattern classification which commonly used. Probabilistic neural network training correct rate was around 89%,and fish-swarm optimization algorithm of training correct rate was up to 98%. The former detection accuracy rate of only 82.5%, the latter, compared with 95%. It can be seen that fish optimization algorithm in pattern recognition has good recognition accuracy.Turbine is an important equipment in industrial production, and its diagnostics is the key to ensure the normal operation.In this paper, fish-swarm optimization algorithm has been applied to the steam turbine fault diagnosis,and also has been compared with the fault diagnosis which based on probabilistic neural network.The result shows that, the diagnostic accuracy rate of fish-swarm optimization algorithm was 96%,and the rate of probabilistic neural network was only 86%.The network which training by fish-swarm optimization algorithm has high fault diagnosis rate.At last,the paper has built a turbine fault diagnosis system by using Visual C# and MATLAB.The operation is simple,and the system also has a high diagnostic efficacy and good stability.