Abstract:

This article takes the dual color injection molding of a precision instrument panel for a certain sedan as the research object. By optimizing the dual color injection molding process parameters, the product warpage and deformation are reduced, thereby improving product quality. In view of the high dimensionality, nonlinearity, volatility and other characteristics between the two-color injection process parameters and the product warpage deformation, and the serious coupling of multiple processes, it is very easy to cause the traditional optimization method to fall into the local optimum, resulting in optimization difficulties and other problems. Based on temporal evolution particle swarm optimization algorithm(TEPSO) is proposed to address the above issues. The algorithm utilizes the advantage of balanced dispersion in orthogonal expansion space to improve the search ability and efficiency of particle swarm optimization, and adopts the Q-learning concept to develop a learning strategy based on temporal evolution through continuous interaction and exploration between particles and the environment to determine the expansion factor of particle orthogonal space. Firstly, each particle generates a Q-table, and the corresponding expansion factor is determined based on the Q-table and state. The fitness value change of each particle is obtained as an immediate reward, and the Q-table is updated using social learning behavior strategies. Secondly, the expansion factor is obtained through Q-learning, and orthogonal design is used for expansion to obtain more environmental information. Finally, shrink each orthogonal expansion space, select the particle position with the smallest fitness value to update the current particle position, and complete the iteration. Optimization design of injection molding process parameters for precision instrument panel of a certain sedan, compared with the initial experimental plan, the use of TEPSO algorithm for optimization reduced the Z-direction warping of the outer cover plate from 4.698 mm to 2.194 mm, and the optimization efficiency reached 53.3%, confirming the effectiveness and practicality of TEPSO algorithm.

Key words: dual-color injection molding, particle swarm optimization, reinforcement learning, optimal design, warping and deformation, simulation, prediction

摘要：

以某轿车精密仪表板双色注射成型为研究对象，通过优化双色注射成型工艺参数，降低产品翘曲变形，从而提高产品质量。鉴于双色注射工艺参数与产品翘曲变形之间呈现高维度、非线性、波动性等特征且多工序耦合严重，极易导致传统优化方法陷入局部最优，造成优化困难等问题，提出了一种基于时序演变的粒子群优化算法(TEPSO)，利用正交膨胀空间均衡散布的优点提高粒子群的搜索能力和效率，并采用Q-Learning思想，通过粒子与环境的不断交互探索，开发基于时序演变的学习策略以确定粒子正交空间的膨胀因子。在某轿车仪表板优化设计中，与初始试验方案相比，采用TEPSO算法优化后仪表板Z向翘曲从4.698 mm降低到2.194 mm，优化效果达到53.3%，证实了TEPSO算法的有效性和实用性。

关键词: 双色注射成型, 粒子群算法, 强化学习, 优化设计, 翘曲变形, 模拟, 预测