The innovation activities of new generation of employees have the characteristics of double network embeddedness, and the degree of psychological contract fulfilment is an important factor that affects their innovation performance. Based on the attributes of internal network embeddedness and external network embeddedness, this paper builds a hypothesis model of the relationship between network embeddedness, psychological contract and innovation performance. It explores the impact and mechanism of network embeddedness on the innovation performance of new generation of employees and the mediating role of the psychological contract. Empirical research shows that network embeddedness has a positive effect on the innovation performance of new generation of employees. The psychological contract has a mediating role in network embeddedness on innovation performance of new generation of employees. These conclusions continue and deepen the research on network embeddedness and innovation performance and further enrich and expand the application of social networks in the research of individual innovation performance of new generation of employees.

A number of suggestions are made based on the improved evolutionary algorithm and using the polishing parameter optimization of an industrial robot as an example to optimize the industrial process parameter control. By fitting a cubic B-spline curve, the trajectory curve of each joint is determined. The kinematic constraint is replaced with the control point constraint of a B-spline curve, and the time optimal time node is solved using an enhanced evolutionary algorithm. This foundation allows for the creation of the nonlinear trajectory curve that satisfies the time optimization. The research shows that based on the improved genetic algorithm (GA), the “degradation” phenomenon of the traditional GA can be avoided, and the optimal solution can be obtained faster, that is, the polishing working time of the polishing industrial robot reaches the optimal level. An enhanced GA that incorporates simulated annealing is suggested to address the mathematical model of robot deburring process parameter optimization. Population selection is accomplished by the use of metropolis sampling, which successfully addresses the issue of the GA’s simple local convergence. The process parameter optimization verification is done while a robot deburring test platform is being constructed. The test results demonstrate a considerable reduction in burr removal time per unit length and an increase in efficiency when compared with the empirical method.