Journal of Applied Engineering and Technological Science Vol 6.
2025: 1280-1296
FLEXIBLE JOB SHOP SCHEDULING OPTIMIZATION USING GENETIC
ALGORITHM FOR HANDLING DYNAMIC FACTORS
Masmur Tarigan1*.
Ford Lumban Gaol2.
Alexander AS Gunawan3.
Widodo Budiharto4 Computer Science Department.
Esa Unggul University.
Jakarta.
Indonesia 1 Department of Doctor of Computer Science.
BINUS - Graduate Program.
Bina Nusantara University.
Jakarta.
Indonesia2 Computer Science Department.
School of Computer Science.
Bina Nusantara University.
Jakarta.
Indonesia34 tarigan@esaunggul.
Received: 30 July 2024.
Revised: 24 February 2025.
Accepted: 18 March 2025
*Corresponding Author
ABSTRACT
This research introduces the Genetic Adaptive Scheduling System (GASS), a novel framework designed to optimize scheduling in Flexible Job Shop Scheduling Problems (FJSP).
Due to its complexity.
FJSP presents significant challenges stemming from machine flexibility, dynamic routing, and operation precedence constraints.
GASS addresses these challenges by incorporating real-time, dynamic data, enabling the system to adapt to machine downtimes, fluctuating job priorities, and process variability.
Leveraging advanced genetic algorithm techniques.
GASS integrates enhanced mutation and selection processes that dynamically adjust setup times, prioritize urgent tasks, and balance machine workloads to minimize makespan effectively.
Empirical results demonstrate that GASS achieves up to a 45.
reduction in makespan within the flexible packaging industry, showcasing its ability to enhance scheduling efficiency and adaptability.
The research highlights the systemAos scalability and potential applicability across diverse industries, including printing, electronics, pharmaceuticals, and food manufacturing, where operational flexibility and efficiency are critical.
By bridging existing gaps and integrating real-time constraints into scheduling models.
GASS provides practical solutions for modern manufacturing environments.
The findings contribute to the advancement of optimization techniques in FJSP, offering valuable insights for researchers and practitioners seeking efficient, scalable, and adaptive scheduling systems.
Keywords: Flexible Job Shop Scheduling.
Genetic Adaptive Scheduling System.
Dynamic Scheduling Optimization.
Manufacturing Process Efficiency.
Real-Time Production Scheduling.
Introduction Due to its relevance to efficient resource and production time allocation.
Flexible Job Shop Scheduling (FJSS) has come to the forefront of many modern manufacturing research The FJSS problem belongs to optimization problems and is an NP-hard problem that becomes more complex due to the consideration of dynamic factors in real production environments (Psarommatis, 2020.
Shao, 2021.
Tyrkylmaz, 2020.
Zhang, 2021, 2023.
Zhuang et al.
, 2.
As a powerful metaheuristic approach, genetic algorithms (GA) have been successfully used to address this issue.
The VSH developed a mathematical model for FJSS, whose objective function is the maximization of total profit, which accounts for the cost of raw material and selling price, as well as the changing demand throughout each period.
This methodology effectively identifies the economical production quantity on different machines to fulfill customer demand at each period (Awad, 2021.
Luo, 2020.
Xu, 2.
Moreover, a comprehensive scheduling model would incorporate dynamic factorsAimost importantly, setup times, job priorities, and machine downtimes that can change throughout an Consequently, the scheduling system has the flexibility to accommodate fluctuating production conditions in real time, resulting in enhanced operational efficiency (Y.
Li et al.
In this context.
GA-based approaches have demonstrated superior schedule quality, taking into account the actual dynamics of the production environment.
The application of GA for FJSS scheduling taking dynamic factors into account has been investigated in few industrial case studies under the heading of practical implementation.
It has been shown that it enhances the efficiency of production and makes it easier to adapt to Tarigan et al A Vol 6.
2025: 1280-1296 changes in demand and operating conditions.
Hence, the use of GA in FJSS scheduling is a useful approach to solve complex problems in dynamic manufacturing environment.
There is a great variety of problems, ranging from simple to very complex ones but the Flexible Job Shop Problem (FJSP) in particular - or more general solutions named General manufacturing scheduling problem GSTP.
FJSP entails more complex task scheduling because of the necessity for processing different products using various machines and with distinct operating sequences.
Efficient scheduling is also very important to make the most of your resources and minimize lead times and production overall.
Fig.
In this example given in Fig.
the FJSP scheduling problem is a real-world challenge of a flexible packaging manufacturing environment where each job presents an intricate structure and demands several operations to be conducted by undistinguishable machines available at the production shop floor.
Fig.
FJSP in Flexible Packaging Manufacture(Tarigan et al.
, 2.
In particular, genetic algorithms .
field of evolutionary optimization technique.
have been widely adopted for complex scheduling problems.
It uses selection, crossover, and mutation genetic processes to emulate natural evolutionary processes by searching for the best A search is an acceptable approach for FJSP as the problem of finding a scheduling solution involves a large and complex search space.
The main aim of this thesis is to formulate a mathematical model for scheduling the FJSP (Flexible Job Shop Proble.
by using Genetic Algorithms.
This model is our attempt to put in place an automated queuing mechanism implemented with real data picked up from the manufacturing industry.
Automated queuing ensures that jobs will be processed efficiently, taking into account priority and the product category or product length in running meters.
The approach aims to minimize the total processing time (TP), which is referred to using makespan, to achieve optimal schedules and improve efficiency in manufacturing.
Challenge: The most significant challenge of this research is to design such a scheduling model that could assimilate the real-time manufacturing data into scheduling models and can automate queue maintenance by prioritizing them on given parameters like priority level, product type, or runtimes.
The approach uses a genetic algorithm to identify the best plan, which minimizes makespan of the maximum task.
We are coming up with good solutions for navigating the complexities of FJSP.
The goal of this query is to develop a concrete scheduling model for the Flexible Job Shop Problem (FJSP) that can well incorporate trusted data coming from practice in We use evolutionary algorithms to Optimize the Scheduling of Flexible Job Shop problems (FJSP).
Utilize an extremely successful scheduling strategy for impeccable time management of the entire process.
The best practice is to establish a rigorous testing and validation procedure based on real data in order to evaluate the utility of your models for working in practical production scenarios.
This is an area of production scheduling and optimization that we anticipate will be materially affected by the research, in particular, building a new mathematical model for FJSP scheduling, including real data from the manufacturing industry.
The research literature, to this end, seeks primarily to illustrate the potential of genetic algorithms in modeling complex and dynamic scheduling problems.
The industrial space is one sector that can implement realistic strategies to increase efficiency and productivity.
So with that, the researchers have reviewed in Tarigan et al A Vol 6.
2025: 1280-1296 this study how to use artificial intelligence technology for better scheduling across the Industry 0 era (Ghaleb, 2020.
Li, 2020.
Zhang, 2021.
Genetic Algorithms (GA.
have emerged as a powerful optimization tool due to their ability to navigate large and complex search spaces effectively.
Compared to other techniques like tabu search or simulated annealing.
GAs offer enhanced adaptability and scalability, making them suitable for addressing the unique challenges of FJSP.
However, existing literature reveals gaps in real-time adaptability and integration of dynamic data in FJSP solutions.
This study aims to bridge these gaps by developing a GASS model that incorporates real-world constraints and dynamic scheduling requirements.
Literature Review Flexible Job Shop Scheduling (FJSP) The Flexible Job Shop Scheduling Problem (FJSP) is a modified version of the traditional job shop scheduling problem.
It allows for more flexibility in choosing machines for each operation, making it more suitable for modern production conditions.
FJSP can handle a variety of tasks and machines with varying capacities.
(Fan et al.
, 2.
described FJSP as one of the most intricate scheduling problems due to the numerous variables and constraints involved.
(Huang & Yang, 2.
highlighted that FJSP presents significant optimization challenges because of its extensive machine selection options (Lei, 2024.
Tian, 2023.
Zhu, 2.
Utilizing Genetic Algorithms for Optimizing Scheduling The genetic algorithm process depicted in Fig.
2 begins with the creation of an initial The chromosomes in this population represent a set of people or potential solutions.
We assess each participant using a fitness function, based on either makespan or total completion time to judge the quality of the resulting solution (F.
Zhang, 2022.
Zhang, 2021b.
Zhang et al.
, 2.
Next, we conduct a selection process to identify the most exceptional individuals who will act as parents in the reproductive process.
Techniques like tournament selection achieve this by increasing the likelihood of choosing individuals with higher levels of Crossover, also known as recombination, is a genetic process that involves combining the genetic material of two parental individuals to create a new individual.
The process involves the exchange of specific genes between the parents.
On the other hand, mutation is a process that alters one or more genes in an individual's chromosomes.
The former brings new variables in play which prevent the user from going directly to a local maximum known solution.
After crossover and mutation, the newly generated individuals undergo an evaluation using a fitness Hence, the selection process for a population of tomorrow means keeping those who are best.
This process repeats or creates new solutions until the termination criterion is satisfied .
, a maximum number of iterations, an acceptable best fitness value, and no improvement in fitness over fewer consecutive generation.
This finally gives us the best solution to the Job Scheduling Problem (Tarigan et al.
, 2.
Fig.
Flow Diagram of Genetic Algorithm Tarigan et al A Vol 6.
2025: 1280-1296 Genetic Algorithms (GA) are commonly used evolution optimization technique in the area of Scheduling.
A genetic algorithm (GA) applies the mechanism of natural selection to evolve solutions through generations, utilizing operations like, but limited to - selection, crossover, and mutation.
(Zhuang et al.
, 2.
indicate that in the case of complex scheduling problems like FJSP, the Genetic Algorithm(GA) is an effective approach due to its ability for a wide exploration search space and handle multiple constraints, etc.
(Singh & Sundar, 2.
have discussed that the GA parameters such as population size, mutation rate, and crossover rate play an important role in better performance of genetic algorithm to solve scheduling problems.
Manufacturing systems automate queuing.
It is a system that organizes and regulates the order in which jobs are processed, which is influenced by factors such as priority, product category, and the required time to produce a running meter of products (Abderrahim, 2020.
Qin, 2021.
Samsonov, 2021.
Zhou, 2.
Automated queuing systems see to it that jobs are done efficiently and within expected timeframes, hence, minimizing the time client waits and maximizing production.
For instance, their research, (Hong & Chien, 2.
found that automatic queuing integrated system into the production system improves production operation and reduces machine downtimes.
Automated queuing is ideal when one is managing a dynamic production environment, facing consistent changes in the types of jobs to be done depending on market demands.
(Rossit et al.
, 2.
, for example, assert that automated queuing has been useful in dynamic environments.
Authentic data from the manufacturing sector The validation of such scheduling models needs to use authentic data taken from the Data includes specifics such as job type, processing time for the jobs, and level of cuproproteins to name a few.
It is stressed by (Baykasolu et al.
, 2.
that to increase the accuracy and reliability of solutions in real production cases, hence realistic data-based scheduling models should be developed.
To this end, (K.
Li et al.
, 2.
claim that the inclusion of realistic data increases model realism and practical applicability as it correctly describes all impediments and lead times suffered in daily production.
State of The Art Genetic algorithms and other techniques have been used by researchers, many studies were performed to enhance the efficiency of FJSP scheduling.
Designed genetic algorithms integrated with real-time data for job shop scheduling (Y.
Li et al.
, 2022.
Meng, 2023.
Soares.
Xie, 2.
, and reported an advance in the performance of gene expression programmers by addressing various medical imaging problems more quickly.
In (Sana et al.
, 2.
, genetic algorithms are used for dynamic job shop scheduling concerning the algorithmized adaptation to an alteration, which enables it to change on production conditions.
An improved genetic algorithm for optimizing the makespan in job shop scheduling was successfully applied to reduce the total processing time (Umam et al.
, 2.
Another study was conducted by (Zheng et al.
, illustrating a remarkable improvement in the quality of scheduling using a data-driven genetic algorithm that exploits real-time information from the production environment.
Open access: H.
Zhu et al.
highlight that it is essential to integrate real-time data into genetic algorithm methodologies for the final schedule to be accurate and efficient.
In their research.
Luo et al.
presented a workshop scheduling problem and solved it by using an improved genetic algorithm.
The results indicated that their method outperformed prior works.
A hybrid method that combines the genetic algorithm with other optimization techniques to improve production schedules has been proposed (Wang & Zhu, 2.
Their method exhibited promising results across all experimental settings.
The paper of (Zhao & Zhang, 2.
extensively focuses on the application of artificial intelligence in flexible job scheduling.
Their study provides insights into present trends and challenges within this area.
The present paper provides an efficient and feasible approach for the manufacturing sector, the considered genetic algorithms technique serves as one of the advanced technology applications in scheduling too.
Tarigan et al A Vol 6.
2025: 1280-1296 Research Methods In this work, we use a quantitative research method to develop and validate an algorithmic framework for solving Flexible Job-Shop Scheduling Problems (FJSP) based on genetic algorithms.
Considering the nature of this study, we used a quantitative approach because it can provide accurate measurements and allow us to statistically analyze those data so that reliable conclusions will be capable of being drawn.
Data Source This study used secondary data from Internal Company Records: This is a type of past production operations records.
This information consists of how long the process takes, the productivity of machines, and output.
Real-time information - e.
, processing times, downtimes, or machinery maintenance Interviews - We will conduct interviews with plant managers and machine operators so we can truly understand the day to day challenges that onsite personnel experience, in addition to what preferences they have for scheduling.
Feedback Surveys can help collect feedback from employees on the work shift they want to perform or whether they are happy with it.
The firm's information system collects data in electronic format using a data gathering method depicted in Fig.
We then merge this data into a centralized database for analysis.
subject the raw data to processing and verification procedures to ensure its accuracy and comprehensiveness before using it in simulations or scheduling models.
Fig.
Entity-Relationship Diagram Data Source The provided diagram depicts the database structure of a production scheduling system, showcasing multiple primary tables and the links that exist between them.
The MachineType table contains data regarding the precise type of machine utilized in the production process.
The machine table, which provides more detailed information about each individual machine, links to this table.
The variable MaterialWidth represents the width of the material utilized and is associated with ProductType, which represents the product's type, including both the material width and the number of colors.
The routing table contains the manufacturing process's sequential order and is linked to machine types and orders.
The Orders table stores customer order details, such as product, routing, and machine used.
The customer entity maintains customer data and their priorities, which are associated with orders.
ProductType is associated with Product, which contains precise product information.
These linkages facilitate streamlined data integration and administration in the production scheduling system, ensuring effective control of all aspects of production, including machines, materials, and client orders.
Tarigan et al A Vol 6.
2025: 1280-1296 Prior to utilization in a genetic algorithm, the data must undergo preprocessing to transform a non-uniform format into a standardized format (Fig.
This encompasses the process of normalizing data, imputing missing values, and converting categorical data into a numerical format suitable for mathematical computations.
Fig.
Standard Format Data for FJSP Scheduling Mathematical model development The problemAos formulation Flexible job shop scheduling (FJSP) challenges arise when a manufacturing company needs to schedule different types of work on many machines efficiently.
The primary objective is to decrease the makespan which refers to the overall duration needed to finish all the tasks.
The following mathematical model provides a concise representation of the fundamental framework of the FJSP problem:
Decision Variable:
1 if process of job is assigned to machine , and Start time of process Process completion time Maximum completion time for all processes 1 if job Due date of job of job on machine of job on machine for all jobs processed after job on machine , and 0 otherwise Objective Function:
Minimum Makespan (Find the most optimal completion time for all Constraints:
where every process of every job must be processed exactly once on the appropriate machine.
OcOc where each machine can only process one job for a particular process until it is finished.
where process completion time of job on machine .
where process completion time of job on machine completion time for all processes for all jobs .
smaller equals to maximum Tarigan et al A Vol 6.
2025: 1280-1296 where decision variable .
is binary numbers.
where actual setup time is reduced when working on the same job type as the previous job on the same machine.
where start time takes into account setup time.
where work completed before the due date.
where each process in one job must complete before the next process can start.
where all previously completed processes are not reassigned to any machine.
Job Definition with Dynamic Characteristics This study classifies occupations as dynamic by considering three primary factors:
product type, running meter, and priority.
The term "product type" describes the classification of the item under manufacture, which includes a variety of items with unique characteristics and production needs.
Running Meter: This is a quantity of measurement that calculates how much material is required to manufacture one product.
It has a significant contribution to production planning and scheduling.
Priority: When you define the list of things to do and their hierarchy based on either market demand or production needs.
By combining these 3 factors the system can monitor and manage tasks on a case-by-case basis allowing for more efficient production that can be adjusted according to changes in demand, work schedules, deadlines, etc.
Genetic Algorithm Implementation Execution A genetic algorithm execution contains three initial population individuals or potential solutions, commonly referred to as chromosomes, is a members of an initial went on calmly enacted by its slime balls.
The fitness function evaluates each generation to assess the final solution quality, usually in terms of makespan or total completion time for production Techniques such as roulette wheel selection select the most exceptional individuals as parents, giving those with superior fitness a higher probability of selection.
The crossover, or recombination, process involves merging two parent individuals to generate new individuals through the exchange of their genes.
Mutation follows, altering one or more genes in the individual chromosomes to introduce fresh variation and prevent premature convergence to a local solution.
After the crossover and mutation process, a fitness function reevaluates the newly created individuals.
Then, we carry out a new generation selection to determine the population of the next generation, retaining only the best individuals.
We continue this iterative process for multiple generations until we meet a specific termination condition, such as a predetermined number of generations or a lack of substantial improvement in fitness values.
The final objective is to discover the most optimal solution to an intricate scheduling problem.
Individual Representation Each entity within the genetic algorithm corresponds to a single prospective resolution for the scheduling predicament.
Chromosomes, which contain data regarding job sequences and machine assignments, represent these individuals.
Here is the proposed pseudocode that Tarigan et al A Vol 6.
2025: 1280-1296 describes the procedure for creating individuals using the FJSP model.
Pseudocode to create individual:
1 : FUNCTION create_individual INITIALIZE machine_end_times TO dictionary with all machines set to 0 INITIALIZE job_end_times TO dictionary with all jobs set to 0 INITIALIZE last_job_type_on_machine TO dictionary with all machines set to None INITIALIZE job_process_allocation TO dictionary of dictionaries for each job and its processes set to None SHUFFLE job_list FOR EACH job IN job_list FOR EACH proc_index, proc IN enumerate.
) INITIALIZE best_end_time TO infinity INITIALIZE best_machine TO None INITIALIZE best_start_time TO None INITIALIZE best_process_time TO None INITIALIZE best_setup_time TO None FOR EACH machine IN available_machines[.
ob, pro.
] SET current_job_type TO job_types.
IF last_job_type_on_machine.
EQUALS current_job_type SET setup_time_current TO setup_time[.
achine, current_job_typ.
] * ELSE SET setup_time_current TO setup_time[.
achine, current_job_typ.
] CALCULATE process_time_adjusted AS processing_time[.
ob, pro.
] * machine_speed.
CALCULATE start_time AS MAX of machine_end_times.
and job_end_times.
CALCULATE end_time AS start_time process_time_adjusted setup_time_current IF end_time < best_end_time UPDATE best_end_time TO end_time UPDATE best_machine TO machine UPDATE best_start_time TO start_time UPDATE best_process_time TO process_time_adjusted UPDATE best_setup_time TO setup_time_current APPEND .
ob, proc, best_machine, best_start_time, best_process_time, best_setup_tim.
TO ind UPDATE machine_end_times.
est_machin.
TO best_end_time UPDATE job_end_times.
TO best_end_time UPDATE last_job_type_on_machine.
est_machin.
TO job_types.
UPDATE job_process_allocation.
TO best_machine RETURN creator.
Individual.
END FUNCTION Fitness Evaluation Tarigan et al A Vol 6.
2025: 1280-1296 We use the fitness function to evaluate each person's excellence based on the resulting make-up.
Individuals with a shorter makespan are believed to have higher fitness.
The implementation uses pseudocode, as depicted below.
Pseudocode to evaluate the individual:
1 : Function evaluate.
Initialize:
job_process_done as dictionary of jobs with lists of boolean .
job_end_times as dictionary with job keys and zero values machine_end_times as dictionary with machine keys and zero values penalties set to 0 For each gene in individual Extract job, proc, machine, start_time, process_time, setup_time from gene proc_index = index of proc in job's process list Check if process already done or out of sequence:
If yes, add 1000 to penalties If machine is None:
Add 10000 to penalties Continue to next gene Calculate actual start time:
actual_start = maximum of machine_end_times.
and job_end_times.
setup_time end_time actual_start process_time If due date of current job equals the minimum of all due dates:
Find jobs with the same due date Select job with lowest priority value from these jobs If current job is not the selected job:
Continue to next iteration Manage job and machine timings:
If actual_start < machine_end_times.
Add 1000 to penalties Else:
Update machine_end_times.
to end_time Update job end times:
job_end_times.
= end_time Mark process as done:
Set job_process_done.
roc_inde.
to true Check for due date violations:
If job_end_times.
> due_dates.
Calculate delay = job_end_times.
- due_dates.
Add delay * late_penalty_factor to penalties Compute total fitness:
max_end_time = find maximum value in job_end_times total_fitness = max_end_time penalties Tarigan et al A Vol 6.
2025: 1280-1296 Return total_fitness 51 : End Function Genetic Operations The execution phases of a genetic algorithm consist of three primary stages: selection, crossover, and mutation.
The selection process uses a tournament selection method to determine the reproductive capability of individuals based on their fitness value.
During the crossover stage, the crossover operator chooses two individuals as parents and merges them to generate new progeny.
The mutation stage occurs when people experience genetic mutations, which include altering small parts of their chromosomes and bringing new variants into the population.
Algorithm Parameters The genetic algorithm is executed with the following parameters: population size, which represents the total count of individuals in a population.
crossover probability, which denotes the likelihood of selecting two individuals for the crossover process.
mutation probability, which indicates the likelihood of an individual undergoing a genetic mutation.
and the number of generations, which signifies the total number of iterations performed by the algorithm.
Model Validation and Verification Authentic data from the industrial industry serves as the validation for the model.
executed multiple test scenarios to assess the model's performance across different production We compare the findings of the genetic algorithm with those of classical scheduling algorithms to assess its superiority in minimizing makespan.
We conducted experimentation using the parameters and data specified in Table 1.
Table 1 - Parameters of Genetic Algorithm Parameter Value Population Size Iteration Cycles Crossover Rate Crossover Two Point Crossover Mutation Rate We conduct an experiment using real data and specified parameters to verify the successful execution of the genetic algorithm, which includes the individual creation function and fitness evaluation function tailored to the FJSP mathematical model.
The experiment utilizes data that conforms to the format depicted in Fig.
4 above.
Executing the genetic algorithm model with the parameters listed in Table 1 yields the results shown in Table 2 below:
Table 2 - Output Genetic Algorithm Processing Generation Evaluations Min Max
Avg
Std
403924,39
124022,435
119737,604
457124,287
1746894,45
8902560,03
1560520,09
4821242,06
178435,374
599782,226
228200,947
1102360,05
6280934,81
277160,279
1007041,46
6759818,39
209885,271
793325,421
7297970,96
432877,257
1324457,04
3671933,85
106150,453
374423,028
1491297,77
91870,0958
201631,431
4108470,42
117675,655
429386,364
792138,24
69627,981
135709,056
107960,393
490773,183
Tarigan et al A Vol 6.
2025: 1280-1296
789138,24
44958,3339
107861,82
3200881,77
75981,9919
359201,668
90011,0427
277372,581
504981,746
36281,7654
63415,7611
1991480,79
57484,2642
206436,592
434502,028
32314,63
57911,751
1097112,06
37206,7141
117043,817
997610,046
39334,4485
114680,941
714664,569
29662,0678
70173,043
2583451,78
51230,3994
255970,937
2333563,52
74713,094
275503,494
630696,934
35100,9773
72482,4448
1841130,48
49721,6673
207417,777
405827,593
28828,1265
46524,5268
87815,8148
320329,138
1847543,01
53817,2176
203993,441
1173889,99
39816,6459
122979,469
2154731,48
53816,8303
220529,887
1620819,64
43411,0812
169051,725
122629,185
476905,137
515750,378
26813,2249
49372,1196
1689160,35
54977,9955
203349,922
2261273,75
56956,7813
241489,41
5446138,41
89076,9857
546259,038
262361,544
26852,0144
34772,7626
1151732,63
43009,2897
123874,808
1224703,61
55867,9959
165951,322
2229909,45
61490,2475
241300,353
1384854,33
37479,5284
139011,81
1232839,93
54346,6454
158726,525
61960,7218
214574,825
1252527,84
58964,0339
200669,681
69176,9255
295462,49
244130,857
24886,4946
24049,265
1337193,73
55828,0065
189262,955
1477005,83
44464,4103
152787,525
786102,844
36261,2875
103377,165
1352691,44
52086,5605
187421,39
Analysis of the Results The experiment yielded results that demonstrate the capabilities of the constructed model when using the genetic algorithm refer to Table 2 above, as shown in Fig.
5 below.
The diagram illustrates the fitness statistics throughout multiple generations of the genetic algorithm.
The graph illustrates three fundamental metrics: the minimal fitness, the average fitness, and the standard deviation of fitness for each generation.
The red line represents the minimal fitness Tarigan et al A Vol 6.
2025: 1280-1296 value, which has a tendency to fall and thereafter stabilize as generations advance, suggesting the algorithm's ability to discover superior solutions.
The blue line depicts the mean fitness, which exhibits a declining and stabilizing pattern, indicating an enhancement in the overall quality of solutions within the population.
The grey region represents the standard deviation of fitness, which indicates the degree of variety in the population.
This variation tends to diminish as time progresses, suggesting that the algorithm is moving towards the ideal solution.
general, the graph demonstrates that the genetic algorithm is effectively optimizing the answer, achieving convergence within a specific number of generations.
Fig.
Fitness Statistic Over Generations Results and Discussions The extraction stage is the initial phase, where we collect and validate data from multiple sources to ensure precision.
Subsequently, in the transform phase, the acquired data undergoes processing and organization, ensuring its readiness for utilization.
After conversion, the load phase sends the processed data to the repository.
During the execution stage, the system applies the genetic algorithm to the data repository to identify the most optimal solution.
e system presents the optimal schedule derived from the optimization process.
In these stages, the study ensures a systematic execution of each step, from extraction to algorithm execution, to achieve the best possible scheduling results.
Table 3 - Data Processing for Flexible Packaging Manufacture Job
Type
Running Meter
Colour
Process
Code
Min Speed
Ratio
Speed
Setup
Time
J57
J57
J57
0,75
J57
J57
J57
J57
J57
M10
0,75
J57
M11
0,75
J57
M12
Tarigan et al A Vol 6.
2025: 1280-1296
J57
M13
0,125
J57
M14
J57
M15
0,75
Table 3 displays the outcomes of data processing for a production task involving product type J57.
This particular product has a production length of 162223 meters and utilizes a total of 9 distinct colors.
The provided data contains comprehensive information for each production process, including a Process List with the sequence number (N.
, the name and code of the machine used, the number of colors processed at each stage, the minimum machine speed (MinSpee.
, the machine speed ratio (Ratio Spee.
, and the machine setup time (Setup Tim.
This means during the first procedure we will print at a B1 printing machine (M.
with a minimum velocity of 200, velocity ratio of.
5, and setup duration =150 if, for example, you would like to produce tickets.
Its existence allows us to schedule merged production schedules and plan accordingly how each operation would be executed at full efficiency, about the machine capacity that is available for use.
This access provides more accurate scheduling and lets you pull insights on areas that can be improved operationally.
Results We conducted a series of experiments using a Genetic Adaptive Scheduling System (GASS) model to improve the production scheduling process before creating a Gantt chart to display the most efficient scheduling.
The goal of this experiment is to reduce the longest job duration and optimize task distribution among the available computers.
We anticipate that the GA model, by leveraging authentic industry data and powerful hardware, will deliver a scheduling solution that not only minimizes machine idle time but also enhances overall We will display the outcomes of this model through a Gantt chart, which will illustrate the order and duration of each task on the respective machines.
This will provide a clear and comprehensive depiction of the optimal scheduling that emerges.
Fig.
Gantt Chart Output by Genetic Algorithm Model Fig.
6 Gantt Chart Output by Genetic Algorithm Model.
The Order and Machine Efficiency Information is the most capable plan of production for order work to machines Taskwise.
Every colored block on this plot is a unique task that some machine needs to do within its allotted time.
Taking into account variables such as product type, linear meters to run, work importance, and machine capacity the model determines a schedule in an automated manner.
The outcomes show that the proposed model is capable of assigning jobs to machines accurately, solving allocation problems without idle state, and zero downtime in order to satisfy continuous goods stream.
As a result, industries are now in perfect condition to go with the schedule made for them and improve both operational efficiency as well as effectiveness while Tarigan et al A Vol 6.
2025: 1280-1296 meeting their specific dynamic production requirements.
This automation plan allows enterprises to address market demands quickly, utilize resources efficiently, and enhance We conducted the Genetic Adaptive Scheduling System (GASS) model experiment on an Apple M2 Pro laptop with 16GB of memory.
Table 4 below displays the results of this This hardware guarantees efficient and rapid execution of the scheduling process and makespan computation by harnessing the laptop's high processing power and extensive memory capacity.
The experimental results demonstrate the superiority of the Genetic Adaptive Scheduling System (GASS) approach in generating a more efficient schedule when compared to traditional approaches, in terms of both scheduling time and overall work completion time .
Table 4 - Comparison of Conventional Method vs GASS Method Conventional Method GASS Method Time Makespan Time Makespan 15,19 0,49 22,785 0,49 30,38 0,49 37,975 0,49 45,57 0,49 53,165 0,49 60,76 0,49 68,355 0,49 75,95 0,49 Job Qty In Table 4 shown above, performances of traditional scheduling methods were compared with GASS-based techniques for different job quantities.
In both timetabling, the "Time" column shows the time taken to generate a schedule, and one side of "Makespan' row demonstrates the total used for execution .
by each method.
This table also shows that the scheduling time is frequently reduced .
49 unit-tim.
in the GA technique compared to the conventional method.
The larger the number of jobs .
rom 50 to more than say around 250 units a yea.
this advantage becomes even more pronounced.
addition, the GASS model results with far better makespan than a typical approach.
As an example, the results of 10 jobs should be a makespan of respectively 5.
4 (GASS methodolog.
and a lengthened to approximately three times as long if using traditional methods ending in 19 As the number of jobs increases, this trend is confirmed as again .
GASS results in lower makespans.
For example, for 50 jobs the makespan is equal to 27 when obtained by the GASS method and the conventional approach yields a value of makespan as high as Conclusion This study presents a comprehensive dataset designed to support research and development in Flexible Job Shop Scheduling Problems (FJSS), particularly for dynamic production environments.
The dataset incorporates critical real-world factors such as machine flexibility, varying processing times, job priorities, setup times, and machine downtimes, offering a robust foundation for advanced optimization techniques.
The dataset is specifically tailored for the implementation and evaluation of the Genetic Adaptive Scheduling System (GASS), a modified genetic algorithm framework.
By integrating enhanced mutation and selection processes.
GASS provides significant improvements in scheduling efficiency, achieving up to a 45.
3% reduction in makespan within the flexible packaging industry.
This performance underscores its potential as a scalable and adaptable solution for dynamic scheduling challenges.
The inclusion of real-world constraints ensures the dataset's applicability across various industries, including printing, electronics, and pharmaceuticals.
Researchers can leverage this Tarigan et al A Vol 6.
2025: 1280-1296 dataset to explore new methods for addressing complex scheduling problems, including multiobjective optimization, dynamic adaptability, and real-time decision-making.
Future work should focus on expanding the dataset to include additional dynamic factors, such as variable demand patterns and energy constraints, and testing its applicability in different industrial scenarios.
Moreover, integrating the dataset with machine learning models could provide further insights into predictive and adaptive scheduling strategies.
Acknowledgment We would like to thank PT.
Intikemas Putra Makmur for the great help and cooperation This observation facility, and the data sources that help enable it are things we thank for every our work in the Lapp Constituent Survey would not be possible without a steady flow of this kind of data.
Observation in a real industry was given and a detail understanding of the system along with practical data that played a significant role indirectly in making development more accurate as well as an activity based scheduling model.
We would like to thank PT.
Intikemas Putra Makmur, for their invaluable assistance in supporting us throughout the research of development process!! References