A Study On Multi-objective Optimization Design Algorithm Of Radar Waveform

The waveform design of radar system is the key technique of the radar detection.Based on different radar task requirements,different waveforms are proposed,including the lowest sidelobe emission waveform,cognitive waveform and orthogonal waveforms.The optimization design of multi-objective radar waveform is aimed at the waveform design problem under multi-requirements.The optimization of waveform is mainly divided into two directions in the thesis,one is to optimize the spectrum function of waveform first,then search for the time-domain constant modulus waveform fitted by the spectrum function,the other is to optimize time-domain waveform directly.In this thesis,a balanced model is proposed for the design of single transmitting waveform of space borne radar,which takes into account the mutual information of waveform and the weighted sum of integral sidelobe and integral main lobe.Compared with the two traditional models,it can obtain large mutual information and has good pulse compression performance.In this thesis,three different methods are used to solve the problem for different requirements.One is to get the linear constrained quadratic programming model through the first-order approximation of Taylor expansion without considering the clutter,which is convenient to solve and requires less calculation,but the obtained solution is approximate.The second is to optimize the complex model with noise and clutter by improving the differential evolution algorithm.More accurate solutions are obtained,but the amount of calculation is increased.The third is to get multi-group solutions for providing more choice,and the NSGA-II algorithm is used.There are more alternative solutions,but the disadvantage is that each solution is not smooth enough,and the calculation is large.Finally,based on the phase sojourn principle,a time-domain constant modulus waveform is obtained.Furthermore,momentum algorithm is used to optimize the phase to reduce the spectrum leakage,so that the waveform spectrum function is similar to the expected spectrum function.For the design of multi transmit waveform for MIMO radar,in addition to the autocorrelation properties of the waveform,the orthogonality between the waveforms is also very important to the detection performance.The traditional LFM signal has good orthogonality and Doppler tolerance,but only two groups can be generated.To get more groups,piecewise method can be used.The traditional piecewise methods are all fixed piecewise number.In this thesis,the piecewise number is also taken as the optimization variable to increase the degree of freedom.A variable piecewise LFM waveform is proposed,and genetic algorithm is used to optimize,so as to obtain multiple sets of orthogonal waveforms.Compared with other traditional orthogonal waveforms,the cross-correlation peak level of the waveforms in this thesis is the lowest,so the orthogonality between waveforms is the best,but the autocorrelation sidelobe level is higher.By analyzing the ambiguity function,we can see that the waveform has good Doppler tolerance.Then,combining with mismatch filtering method,the sidelobe is greatly reduced under the condition that the main lobe is not widened and the output SNR loss is constrained.Finally,aiming at the disadvantage of low ratio of the main lobe and the side lobe of the variable piecewise LFM waveform under the matched filter,a variable piecewise NLFM waveform is obtained by replacing the LFM signal with the NLFM signal.Compared with the former,under the slight loss of cross-correlation performance,the autocorrelation sidelobe level is greatly suppressed,and there is no loss of output SNR caused by windowing.But after matched filtering,the main lobe of the waveform is widened and the Doppler tolerance is reduced.