Research On The Estimation Performance Of Adaptive Weighted Random Pool Network

In the development process of the random signal processing field,the processing of random noise has been mainly based on reducing and suppressing.With the discovery of the phenomenon of stochastic resonance,people realized that the effect of noise on signal processing is not necessarily negative,and it can even improve the accuracy of signal estimation.Therefore,the research on stochastic resonance phenomenon is further deepened.Suprathreshold stochastic resonance is how stochastic resonance behaves under certain conditions.A number of studies have shown that processing signals through a stochastic pooling network will generate stochastic resonance when noise is added.Therefore,we use the stochastic pooling network as a carrier to explore ways to improve the signal estimation performance.In order to further improve the signal estimation performance of the stochastic pooling network,this paper adds weight coefficients and adaptive algorithm modules to the traditional stochastic pooling network.The mean square error between the real signal and the estimated signal is mainly used as the evaluation network estimation performance.The theoretical expression of the minimum mean square error of the adaptive weighted stochastic pooling network under the optimal weight coefficient is theoretically derived.The correctness of the theoretical expression is verified by simulation.The relationship between the number of network nodes and the signal estimation performance is studied,and the conclusion is drawn that the estimation performance of the adaptive weighted stochastic pooling network is continuously improved as the number of nodes increases.Next,the performance of the network performance under multiple thresholds is studied,and the conclusion that the network performance is better as the number of thresholds increases is obtained.Combined with the knowledge of the Cramer-Rao bound and the error bound in the sense of Fisher information,the performance of the multi-threshold adaptive weighted stochastic pooling network is further evaluated,and it is of great significance to improve the estimated performance of the stochastic pooling network.