To solve the problems of high energy consumption, high operating cost, and slow temperature response in coupled temperature–pressure simulations of large-scale deep-water test chambers, a non-integrated temperature–pressure transfer device was developed. In this device, accurate pressure transfer is realized through elastic deformation of a rubber pressure-transmitting structure, while precise control of the surrounding temperature is achieved using a helical heat exchange tube. To improve the performance of the device, the pressure resistance and heat transfer behavior under deep-water high-pressure conditions were systematically analyzed, followed by a multi-objective optimization study. First, a radial deformation model was established to describe the mechanical response of the pressure-transmitting structure and ensure structural safety. A calculation method was proposed to determine the critical pressure of the helical heat exchange tube under yielding and buckling failure modes. The sensitivity of critical pressure to key structural parameters was also examined. Second, to evaluate heat transfer performance without the influence of specimen size, a comprehensive evaluation factor defined as the heat transfer coefficient per unit pressure drop was introduced. This factor was used to analyze the thermal sensitivity of heat transfer performance with respect to major design parameters. Then, a multi-objective optimization model was built by selecting the outer diameter, wall thickness, and helical radius of the heat exchange tube as design variables. The optimization aimed to maximize both the critical pressure and the comprehensive evaluation factor. The NSGA-II algorithm was used to obtain the Pareto front, and the optimal design was selected by combining the entropy weight method with the TOPSIS decision method. Experimental results show that the optimized device operates reliably under a pressure of 35 MPa. The internal pressure of the device matches the chamber pressure, indicating good dynamic pressure-following behavior. The internal ambient temperature was controlled from 23 °C to 3 °C, with a temperature response time (TRF) of 83 min and a relative steady-state temperature fluctuation (RTF) of 4%. The device demonstrates efficient heat transfer performance and high temperature control accuracy. Through a complete "design-analysis-optimization-verification" process, this work provides an efficient solution for deep-water environmental simulation centered on performance optimization.