A Study On Complexity Evaluation Of Electromagnetic Environment Signal Based On Mathematical Morphology

With the rapid development of science and technology,radar,communication,strong electromagnetic radiation interference and other equipment are used in large quantities,a wide variety,large number and complex function of electromagnetic signals filled the battlefield space,the battlefield electromagnetic radiation energy from weak to strong,the spectrum from narrow to wide,the electromagnetic environment tends to be complex,which has a great impact on the safety of electromagnetic equipment and personnel in the electromagnetic environment,so the complexity needs to be assessed and studied to take corresponding measures to ensure the integrity of equipment and personnel safety.The complex electromagnetic environment signal has significant nonlinear nature,and the traditional linear signal processing techniques are not effective for it.Mathematical morphology theory is a very representative method in the field of nonlinear signal processing and analysis,and its algorithm is simple,fast,easy and can effectively extract and retain the edge information and morphological features of the signal.Therefore,this paper aims at the complexity assessment and classification of complex electromagnetic environment signals,takes mathematical morphology theory as the main technical method,combines it with other theoretical methods such as non-negative matrix decomposition and neural network,and proposes a feature extraction and classification method for complex electromagnetic environment signals based on mathematical morphology,which provides a new theoretical and technical path for the complexity assessment of electromagnetic environment signals.The main results of the paper are as follows.(1)A variety of communication signals,radar signals and interference signals are simulated,and the amplitude and emergence duration of the simulated signals are randomly generated considering the characteristics of complex electromagnetic environment,thus four kinds of electromagnetic environment signals of different complexity are constructed:simple,mildly complex,moderately complex and heavily complex.(2)A feature extraction method based on adaptive multi-scale morphological gradient filtering and non-negative matrix decomposition is proposed for the electromagnetic environment signal.The adaptive multi-scale morphological gradient filtering is used to process the electromagnetic environment signals in response to the problem of one-sided signal processing structure by using single-scale structural elements.The results show that the adaptive morphological gradient operation is more effective for signal processing.After filtering the signal,the non-negative matrix decomposition method is selected to extract the features of the electromagnetic environment signal,and the results show that the extracted features provide better distinguishing information for different complexity of the electromagnetic environment signal,and can better distinguish between different complexity of the electromagnetic environment signal.(3)A feature extraction method of electromagnetic environment signal based on logarithmic morphological gradient spectrum is proposed.The method addresses the statistical bias phenomenon of the mathematical morphological spectrum and the problem that the morphological gradient spectrum mainly highlights the impact information of the signal,and carries out logarithmic processing based on the mathematical morphological gradient spectrum.Applying the logarithmic morphological gradient spectrum to the feature extraction of electromagnetic environment signals,and extract the feature parameters of electromagnetic environment signals of different complexity.The results show that the feature extraction method can well provide feature information with good differentiation and lay the foundation for the evaluation and classification of electromagnetic environment signals of different complexity.(4)A method for evaluating the complexity of electromagnetic environment signals based on dendritic morphological neurons is proposed.For the evaluation and classification of electromagnetic environment signals of different complexity,a dendritic morphological neuron DMN-SGD trained by stochastic gradient descent method is used.The algorithm improves the training process by introducing a stochastic gradient descent method to optimise the dendritic morphological neuron model for its overfitting problem.The DMNSGD is applied to the classification of complex electromagnetic environment signals,and high classification accuracy is achieved.The effectiveness of the feature extraction method of electromagnetic environment signals proposed in this paper is further verified.