The time-frequency domain analysis and detection method of a weak harmonic signal cannot identify the target signal in a scene which has strong noise in the wide spectrum environment. Thus this paper first introduces a stochastic resonance principle of a chaotic system, and when the signal component is completely unknown, uses random noise to enhance the weak target signal for frequency blind detection. It introduces the Duffing chaotic system further, uses its high sensitivity and noise resistance to the amplitude of specific frequency periodic signals, and accurately detects the amplitude of each harmonic frequency component that has been detected. Using the classical stochastic resonance model, the system parameters are improved by normalization. Combined with the Duffing chaotic system model, a real-time detection control strategy is designed. The particle swarm optimization algorithm is used to find the optimal system parameters by taking the correlation number of input and output signals of the system as the fitness function. Simulation experiments are conducted on harmonic signals with multiple integer harmonics and inter harmonics mixed in a strong random noise background. The results show that the method proposed can achieve accurate frequency detection and high-precision amplitude detection of various harmonic components in harsh noisy environments.