磁保持型航天电磁继电器性能提升与质量一致性设计

Magnetic latching aerospace electromagnetic relay (MLAER) has high sensitivity, low heat generation, low power consumption, and small size, used in various aerospace equipment such as vehicles, spacecraft, and satellites. The calculation accuracy of MLAER performance parameters and the iterative search efficiency of the optimal combination of consistency under the constraints of multiple performance objectives are the important factors affecting the optimization of MLAER. This paper proposes a design method for modeling and iterative optimization of MLAER combined with the Stoner-Wahlfarth method and the Preisach method, based on a non-linear permanent magnet local hysteresis model. An approximate model basis function is obtained based on fuzzy chaotic expansion. The parameters of the measured prototype are determined using the fastest curvature descent, and the error between the measured and calculated values is corrected using the kriging method to achieve accurate and efficient calculation of the performance parameters of the relay. The law between the magnetic retention relay design parameter and the shape factor of the suction curve is revealed, and the differential evolution algorithm is improved to form an adaptive optimal population iteration strategy. It improves the optimization efficiency and robustness of the differential evolution algorithm in solving the magnetic retention relay optimization problem. The following results have been achieved. (1) The magnetically maintained aerospace relay with nonlinear anisotropic permanent magnets has strong nonlinearity and the local hysteresis effect of magnetization and demagnetization. Therefore, this paper proposes a virtual prototype modeling method that can be applied to all types of relays. The S-W method is used to calculate the vectorization of the magnetic dipole after magnetization and demagnetization of nonlinear anisotropic permanent magnets. Accordingly, the virtual prototype model of the relay's static and dynamic performance characteristics is established. The electromagnetic suction force, mechanical reaction force, operating voltage, breaking speed are calculated for different voltages and rotation angles. Compared with the measured results, the maximum error is 5.01 %. (2) The calculation accuracy of the existing agent model is poor when the performance characteristics are high dimensional, strongly nonlinear, and output zero crossing. Based on a small number of virtual prototype model calculation values, the basis function of the proxy model is obtained using fuzzy chaos expansion to correct the error between measured and the calculated results. Then, the proxy model is established. The maximum error is 6.72 %, the computational accuracy is similar to that of the virtual prototype, and the computational resource and time consumptions are less than one percent of the virtual prototype model. (3) Based on the agent model, the formation mechanism and improvement direction of relay performance characteristics are determined from the perspectives of suction-reverse force coordination and partition kinetic energy. Meanwhile, the multi-objective differential evolution algorithm is improved to meet the demand for multi-objective optimization. The adaptive population evolution strategy and the optimal solution set screening strategy based on the idea of small habitat are proposed, which can obtain the optimal Pareto frontier solution set combination more uniformly and quickly. (4) The iterative optimization search process avoids maturing prematurely and falling into local optimal solution sets. The extensiveness and uniformity of solution distribution are enhanced. Thus, the iterative optimization of the MLAER multi-correlation performance parameters can be ensured with accuracy and efficiency. (5) The potential zeroing faults, such as armature neutralization and operation voltage overrun, were optimized, and the optimal mean and tolerance combinations of design parameters were given. The performance and quality consistency of the relay were improved by 16.8 % and 40 % compared with the pre-optimization period. The optimization results demonstrate the virtual prototype model considering the nonlinear permanent magnet charging and demagnetizing effects, the agent model introducing the measured results, and the differential evolution algorithm with the adaptive population strategy are effective in solving the MLAER problem.