Dynamic optimization analyses and algorithm design for the parallel heat exchange system in spacecraft

Abstract

Dynamic optimization of the fluid loop is critical of the active thermal control system (ATCS) for future spacecraft. In this paper, the transient heat current model of the parallel heat exchanger system is constructed by the thermoelectric analogy method to analyze the optimal control problem of dynamic flow allocation. The continuous free time optimal flow control problem is transformed into a fixed-time nonlinear programming problem which can be solved by the sequential quadratic programming (SQP) algorithm through the control variable parameterization (CVP) and time-scaling methods. The simulation showed that the optimized flow allocation significantly reduced the process time required for system heat dissipation. Compared with the non-gradient particle swarm optimization (PSO) algorithm, the SQP method constructed in this paper decreased the optimization index by about 4%∼6% and the calculation time by about one order of magnitude. To diminish the sensitivity of the SQP algorithm to the initial value, this paper simplifies the model according to the system characteristics. Through the Pontryagin extremum principle, it is found and proved that, under the simplified model, the optimal flow rates are a set of time-independent variables, which significantly reduces the difficulty of the optimization problem, according to which a hybrid algorithm is designed that uses the optimal result of the simplified model as the initial value predictor. Compared with the average of using the random initial value, using the estimated initial value decreased the optimization index by more than 2% and the computation time by about 60%, under the same SQP algorithm.