Adaptive Knowledge-based Multi-Objective Evolutionary Algorithm for Hybrid Flow Shop Scheduling Problems with Multiple Parallel Batch Processing Stages

Feige Liu,Xin Li,Chao Lu,Wenying Gong

2024-09-27

Abstract:Parallel batch processing machines have extensive applications in the semiconductor manufacturing process. However, the problem models in previous studies regard parallel batch processing as a fixed processing stage in the machining process. This study generalizes the problem model, in which users can arbitrarily set certain stages as parallel batch processing stages according to their needs. A Hybrid Flow Shop Scheduling Problem with Parallel Batch Processing Machines (PBHFSP) is solved in this paper. Furthermore, an Adaptive Knowledge-based Multi-Objective Evolutionary Algorithm (AMOEA/D) is designed to simultaneously optimize both makespan and Total Energy Consumption (TEC). Firstly, a hybrid initialization strategy with heuristic rules based on knowledge of PBHFSP is proposed to generate promising solutions. Secondly, the disjunctive graph model has been established based on the knowledge to find the critical-path of PBHFS. Then, a critical-path based neighborhood search is proposed to enhance the exploitation ability of AMOEA/D. Moreover, the search time is adaptively adjusted based on learning experience from Q-learning and Decay Law. Afterward, to enhance the exploration capability of the algorithm, AMOEA/D designs an improved population updating strategy with a weight vector updating strategy. These strategies rematch individuals with weight vectors, thereby maintaining the diversity of the population. Finally, the proposed algorithm is compared with state-of-the-art algorithms. The experimental results show that the AMOEA/D is superior to the comparison algorithms in solving the PBHFSP.

Neural and Evolutionary Computing,Systems and Control

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the scheduling problem of multi - stage parallel batch - processing machines in the Hybrid Flow Shop Scheduling Problem (HFSP), namely Parallel Batch Processing Machines in Hybrid Flow Shop Scheduling Problem (PBHFSP). Specifically, the author focuses on how to simultaneously optimize two objectives while considering multiple parallel batch - processing stages: minimizing the makespan and the Total Energy Consumption (TEC).

### Problem Background

1. **Applications of Parallel Batch Processing**
- Parallel batch - processing machines are widely used in the semiconductor manufacturing process. However, most previous studies regarded parallel batch - processing as a fixed stage in the processing process.
- This paper proposes a more general model, and users can arbitrarily set certain stages as parallel batch - processing stages as needed.

2. **Limitations of Existing Research**
- Existing literature mainly focuses on a single parallel batch - processing stage or treats it as a special stage, lacking a comprehensive study of multi - stage parallel batch - processing.
- Although some studies consider that any stage can be a batch - processing stage, they assume that all job sizes are the same, which limits the generality of the model.

3. **Optimization Objectives**
- PBHFSP is an NP - hard problem, and it is difficult to find the optimal solution within a limited time. Therefore, it is crucial to design an algorithm that can effectively solve the multi - batch - processing - stage scheduling problem.
- Simultaneously optimize the makespan and the total energy consumption to increase corporate profits and reduce environmental burdens.

### Main Contributions

1. **Expand the Problem Model**
- A PBHFSP model that allows any stage to be a parallel batch - processing stage is proposed, and the machine capacity and batch - processing size can be different.

2. **Propose a Discrete - Graph - Based Model**
- Based on the knowledge of PBHFSP, a discrete - graph model is constructed to correlate batch - processing stages and discrete - processing stages.

3. **Design an Adaptive Knowledge - Based Multi - Objective Evolutionary Algorithm (AMOEA/D)**
- Combine the coding method and the knowledge of PBHFSP, and integrate the neighborhood structure based on the critical path into AMOEA/D.
- Adaptively adjust the search range through Q - learning and the attenuation law to balance the exploration and exploitation capabilities.
- Introduce an improved population update strategy and a weight - vector update operation to maintain population diversity.

4. **Experimental Verification**
- Comprehensive experiments are carried out to verify the effectiveness and superior performance of AMOEA/D.

### Mathematical Model

To describe the problem more clearly, the following is the mathematical model of PBHFSP:

#### Symbol Definitions

- \( i \): job index, \( i = 1,2,\cdots,n \)
- \( j \): stage index, \( j = 1,2,\cdots,m \)
- \( k \): machine index, \( k = 1,2,\cdots,p \)
- \( t \): position index on the machine, \( t = 1,2,\cdots,q \)

#### Parameters

- \( n \): the number of jobs
- \( m \): the number of stages
- \( B_{j} \): the type of stage \( j \). If stage \( j \) is a parallel batch - processing stage, then \( B_{j}=1 \), otherwise \( B_{j} = 0 \)
- \( p \): the number of machines
- \( p_{ijk} \): the processing time of job \( i \) on machine \( k \) at stage \( j \)
- \( s_i \): the size of job \( i \)
- \( C_k \): the capacity of machine \( k \)
- \( E_L \): power consumption when the machine is loaded
- \( E_I \): power consumption when the machine is idle

#### Variables