Development of Sparse Time Frequency Distribution Reconstruction Using a Gradient Slime Mould Renyie Entropy Shrinkage Model

Time–frequency distributions (TFD) can provide a great set of tools for studying transient signals. Although TFD can overcome constraints on signal representation, the most widely used TFDs produce artifacts known as cross terms, which could pose a challenge when used on real-world signals. The paper proposes a sparse Time–Frequency Distribution (TFD) reconstruction method employing a gradient slime Renyi entropy shrinkage model. Each time–frequency slice within the TFD undergoes shrinkage utilizing distinct algorithms based on Renyi entropy, which accounts for both short-term and narrowband characteristics. Renyi entropy quantifies data presence in the time–frequency plane. By integrating the Renyi entropy-based shrinkage operator, the traditional hard threshold operator in the shrinkage process is replaced, enhancing TFD resolution. The reconstruction model parameters are fine-tuned using a gradient slime shape optimizer, employing a concentration minimization function and Mean Squared Error (MSE) metrics between initial and reconstructed TFD modules for optimization. The simulation results prove that the proposed method achieved a reduced MSE value of 1.95 as compared with other existing methods.