One of the most prevalent spread spectrum techniques is frequency hopping. Once implemented, it will dramatically enhance the difficulty of communication confrontation because to its high anti-interference ability, low probability of interception, and ease of networking. In the crowded environment of modern electronic warfare, it is critical to predict the parameters of frequency hopping signals and provide prior knowledge to jammers. To begin, the distribution properties of the frequency hopping signal in the fuzzy domain are completely leveraged to create a novel bilinear time-frequency distribution kernel function based on the existing nonlinear time-frequency approach. Second, entropy measurement theory is introduced to optimize the kernel function parameters, filter the cross-interference items to reduce interference, and keep the self-term as long as possible. Finally, the procedure of estimating the frequency hopping signal parameter is completely designed. The experimental results suggest that the kernel function proposed in this study is capable of accurately estimating signal properties in a -6 dB environment. The improved kernel function can provide larger performance benefits under the same information entropy conditions.