Research And Implementation Of Digital Noise Model Based On Particle Swarm Optimization Algorithm

In the process of industrial development,in order to reduce the cost of R & D,testing and manufacturing,it is necessary to establish a simulation model that restores the real situation for simulation testing.The establishment of traditional simulation models is costly and time-consuming,and in some fields,there are security problems in the establishment of simulation models.With the advent of the "intelligence +" era,the technology in the process of industrial development is deeply integrated with the new generation of information technology,among which the simulation technology and the new generation of information technology fusion produced a digital simulation model,this model compared to the traditional simulation model,low cost,high security.The most important is that the digital simulation model can simulate more comprehensive states than the traditional simulation model,and these states are controllable.In order to establish a digital model consistent with the actual situation,it is necessary to consider all possible influencing factors in the environment where the device is located,including the noise signal suffered by the device.Therefore,how to generate noise signal consistent with the real noise characteristics is particularly important.This thesis studies the problem of using swarm intelligence algorithm to generate noise signals.The main research contents and contributions are as follows:1)Two-Layer Particle Swarm Optimization(TWOPSO)algorithm was proposed to optimize the tap coefficient of the filter.The first layer optimized the order of the filter,and the second layer optimized the tap coefficient based on the order found in the first layer.This solves the problem that the particle swarm optimization algorithm cannot find the optimal filter tap coefficient because the filter order is fixed.2)Generate uniform random sequence by combining shift register principle and chaos principle,and map the uniform random sequence to Gaussian random sequence by combining Lagrange mean value theorem and linear interpolation method.The mean value and variance of the generated Gaussian random number are-0.0049 and 1.0392.And this method solves the problem that the period characteristic of data is affected by data bit width.3)Using the two-layer particle swarm optimization algorithm in MATLAB,the tapped coefficients of the filters in the noise model are optimized for four types of noises in Noise X-92 library.The correlation between the power spectral density of the noise signal generated by the tapped coefficients and that of the target noise is higher than 0.87.4)The tapped coefficient obtained by the two-layer particle swarm optimization algorithm was configured on the FPGA platform,and the correlation between the power spectral density of the noise data obtained and that generated by MATLAB was 0.8276.