Ford Motor Company (United States)

Intelligent manufacturing involves scheduling methods that maximize efficiency, use of energy, and cost. This paper introduces a hybrid metaheuristic framework which combines the exploration improvement capability of Genetic Algorithms (GA) and the exploitation power of Particle Swarm Optimization (PSO). The suggested merge GA-PSO model develops Job Shop Scheduling Problem (JSSP) as a multiobjective optimization task with the function of minimum makespan, energy consumed, and operational cost according to the induction function. The framework is validated by means of both standard JSSP benchmarks as well as Industry 4.0 inspired datasets. Experimental results show the significant performance improvements, i.e., 18.6 % makespan reduction, 12.4 % energy consumption reduction and 15.2 % operational cost reduction based on standalone GA, PSO and baseline methods. In addition, the Pareto-front diversity and convergence speed of GA--PSO are better, and hence, more effective trade-offs can be achieved among conflicting objectives. These outcomes have established GA-PSO as a strong and efficient scheduling approach for realtime, energy awareness and cost-optimized operations in the Industry 4.0 environment to pave the way for sustainable and intelligent manufacturing systems.

Abstract The straightness of a carburized pinion gear shaft is directly linked to the heat treatment process. When the distortion induced by the heat treatment process cannot be controlled, bending straightening is a common manufacturing practice used to correct the distortion without removing the hardened surface layer. The straightening process can be effectively simulated using computer-aided engineering (CAE) analysis, but its accuracy will be limited if information about the case depth and material composition is unavailable. Moreover, both the case depth and material composition are highly dependent on the specifics of the heat treatment process. Since distortion is caused by biased quenching, assuming uniform case depth and homogeneous material properties is unrealistic. To accurately simulate the straightening process, detailed information about the heat treatment history must first be obtained. The primary objective of this research is to develop a CAE methodology to extract heat treatment process parameters from distortion measurements using an inverse modeling approach. Given that distortion is a direct result of the heat treatment process, and assuming the quenching heat transfer coefficient is a function of surface temperature only, it can be shown that a one-to-one mapping exists between a biased quenching profile and the resulting distortion. Based on this finding, the inverse heat conduction method can be used to reconstruct the quenching profile from the distortion data. After determining the quenching profile, the case depth and material composition then are calculated, which enable a precise CAE simulation of the bending straightening process. The availability of both the quenching profile and distortion metrics also supports a secondary objective of this research: to reverse-engineer the flow conditions inside the quench tank. This information can be used to improve quench tank design and flow pattern, thereby minimizing quenching-induced distortion in future production.

The three-phase single-stage differential boost inverter (DBI) has drawn significant attention in power conversion systems when feeding permanent magnet synchronous motors (PMSM). It offers advantages such as single-stage boosting, high efficiency, cost-effectiveness, and a compact structure. However, its nonlinear and non-minimum phase characteristics pose challenges for control. In this paper, the application of model predictive control (MPC) for PMSM fed by the three-phase DBI is focused on. Firstly, the modeling method of the three-phase DBI is presented. Subsequently, an MPC approach for PMSM fed by three-phase DBI is proposed. Based on the established predictive models of circuit currents and voltages, the appropriate duty cycle for the next sampling period can be predicted, which simplifies the control process and parameter design compared to traditional exclusively PI-based control. Simulation experiments are carried out to verify the effectiveness of the proposed MPC strategy for PMSM driven by the three-phase DBI.

This letter presents a data-driven safe motion planning approach, <monospace xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">DaSP-RRT</monospace>, designed to generate collision-free paths with guaranteed optimality through the use of invariant sets. The proposed planner constructs a sequence of performance-aware invariant sets using available data and a new control design approach. These sets are centered around randomly generated waypoints, which are then connected to form a continuous path from the initial to the target point. For each waypoint, an optimization problem determines the largest performance-aware invariant set and learns its corresponding controller. A key feature of the algorithm is its incorporation of performance-reachability between connected waypoints, leveraging available resources and system information to minimize the need for frequent re-planning. The effectiveness of <monospace xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">DaSP-RRT</monospace> is demonstrated through a real-world implementation on an omnidirectional wheeled robot and simulations on spacecraft motion planning. These scenarios, which include obstacle avoidance, highlight the algorithm's potential for practical, real-world applications.