Based on the principle of chaos suppression with nonresonant parametric drive in autonomous chaotic systems, detection of weak square wave signals in strong noise is realized. Firstly a square wave signal whose frequency is far higher than the system characteristic frequency is introduced as the system internal drive signal. According to the parametric equivalent relation between the controlled system and the original system obtained by the averaging method, the critical value of the detection parameter is determined, which marks the sudden change from chaos state to periodic orbit of the system. Numerical simulations show that extremely low signal-to-noise ratio limit in detection can be attained by the proposed system. Compared with the known weak signal detection methods based on the parametric resonant perturbation principle, this method gives more accurate estimates of the detection parameter through strict theoretical analysis, which enables its generalization and application in related fields.