Intelligent mes system production scheduling method based on big data

Abstract

This invention discloses a production scheduling method for an intelligent MES system based on big data, comprising the following steps: acquiring data and generating processed production data; forming a production feature set based on the processed production data; generating a harmony search adaptive parameter set and a simulated annealing adaptive parameter set; initializing a harmony memory based on the harmony search adaptive parameter set, generating and evaluating scheduling candidate solutions, and updating the harmony memory; selecting the current scheduling solution and generating an updated current scheduling solution based on the simulated annealing adaptive parameter set; writing the updated current scheduling solution into the harmony memory and generating new scheduling candidate solutions and their evaluation results; performing global search and local jump optimization based on the harmony memory to generate the final scheduling solution; generating a production scheduling scheme based on the final scheduling solution and outputting it to the manufacturing execution system. This invention employs a big data-driven hybrid scheduling method of harmony search and simulated annealing, possessing high accuracy and strong dynamic response capabilities.

Description

Technical Field

[0001] This invention relates to the field of industrial production scheduling technology, and in particular to a production scheduling method for an intelligent MES system based on big data. Background Technology

[0002] Manufacturing Execution Systems (MES), as crucial information platforms connecting production planning and shop floor execution, are widely used in discrete and process manufacturing industries. Existing MES systems typically execute production scheduling through rule-driven methods or based on fixed algorithm parameters, generating scheduling schemes using basic data such as processing time, equipment status, and queue duration. However, with accelerating production pace, diversified order structures, and frequent changes in equipment status, traditional scheduling methods struggle to reflect processing fluctuations and load changes in real time. Scheduling results tend to lag behind actual production conditions, leading to decreased resource utilization and increased delivery risks.

[0003] Existing production scheduling technologies mostly employ intelligent optimization methods such as genetic algorithms, particle swarm optimization, or conventional simulated annealing and harmony search algorithms. However, they generally use fixed parameters or static scheduling strategies, lacking the ability to dynamically adjust parameters using MES big data. When the production environment is disturbed, these methods are prone to getting stuck in local optima or failing to update the scheduling solution in a timely manner, making it difficult to simultaneously consider global optimization capabilities and local leapfrogging capabilities, resulting in unstable scheduling quality. Especially when facing dynamic scenarios such as bottleneck fluctuations, changes in order urgency, and equipment load fluctuations, existing technologies cannot achieve real-time, adaptive, high-quality scheduling optimization.

[0004] Therefore, how to provide a production scheduling method for an intelligent MES system based on big data is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] One objective of this invention is to propose a production scheduling method for an intelligent MES system based on big data. This invention adopts a big data-driven hybrid scheduling method of harmony search and simulated annealing, which has high precision and strong dynamic response capabilities.

[0006] The production scheduling method for a big data-based intelligent MES system according to an embodiment of the present invention includes the following steps:

[0007] Acquire historical and real-time production data from the manufacturing execution system, perform data cleaning, anomaly identification, and structured processing, and generate processed production data;

[0008] Based on the processed production data, the processing time fluctuation index, equipment load change rate, process urgency index, queuing congestion index, and bottleneck process intensity are calculated to form a set of production characteristics;

[0009] Based on the production feature set, adaptive adjustment calculations are performed on the harmony search algorithm and the simulated annealing algorithm to generate the harmony search adaptive parameter set and the simulated annealing adaptive parameter set;

[0010] The harmony memory is initialized based on the adaptive parameter set of the harmony search and multiple scheduling candidate solutions are generated. The quality of the scheduling candidate solutions is evaluated and the harmony memory is updated.

[0011] Based on the updated harmony memory, the current scheduling solution is selected, local perturbation is performed using the simulated annealing adaptive parameter set, and an updated current scheduling solution is generated based on the cost difference and acceptance probability parameters.

[0012] Update the temperature of the simulated annealing algorithm according to the temperature drop rate, write the updated current scheduling solution into the harmony memory, and generate new scheduling candidate solution evaluation results;

[0013] The global search and local jump optimization process is performed based on the harmony memory library, and the final scheduling solution is generated when the termination condition is met.

[0014] The final scheduling solution is used to generate a production scheduling plan, which is then output to the manufacturing execution system.

[0015] Optionally, the generation of the production feature set includes:

[0016] The processing time records of each process in the processed production data are used to construct a processing time series in chronological order, and the processing time fluctuation value is calculated based on the processing time series.

[0017] The equipment operating status records in the processed production data are aggregated according to fixed time windows, and the equipment load change value is calculated based on the equipment load change between adjacent time windows.

[0018] Calculate the process urgency value based on the time difference between the order requirement completion time and the current time in the processed production data's order attribute data.

[0019] The queuing time of each process in the processed production data is grouped according to the production line number, and the queuing congestion value is calculated based on the difference between the queuing time of each production line and the historical average queuing time of that production line.

[0020] The bottleneck process intensity value is calculated based on the process processing time, equipment availability, and process parallelism in the processed production data.

[0021] The production feature set is generated by combining the processing time fluctuation value, equipment load change value, process urgency value, queue congestion value, and bottleneck process intensity value.

[0022] Optionally, the generation of the simulated annealing adaptive parameter set includes:

[0023] The adaptive adjustment amount of the harmony memory selection probability is calculated based on the production feature set. The adaptive adjustment amount is then combined with the preset base value of the harmony memory selection probability to generate the updated harmony memory selection probability.

[0024] The adaptive adjustment amount of the tone adjustment rate is calculated based on the processing time fluctuation value and equipment load change value in the production feature set. The adaptive adjustment amount is then combined with the preset tone adjustment rate base value to generate the updated tone adjustment rate.

[0025] The adaptive adjustment amount of the pitch change step size is calculated based on the processing time fluctuation value in the production feature set. The adaptive adjustment amount is then combined with the preset pitch change step size base value to obtain the updated pitch change step size.

[0026] The updated harmony memory selection probability, the updated pitch adjustment rate, and the updated pitch change step size are combined to generate a set of adaptive parameters for harmony search.

[0027] The adaptive adjustment amount of the initial temperature is calculated based on the processing time fluctuation value and bottleneck process intensity value in the production feature set. The adaptive adjustment amount is then combined with the preset initial temperature base value to obtain the updated initial temperature.

[0028] The adaptive adjustment amount of the temperature drop rate is calculated based on the equipment load change value in the production feature set. The adaptive adjustment amount is then combined with the preset temperature drop rate base value to obtain the updated temperature drop rate.

[0029] The adaptive adjustment amount of the acceptance probability parameter is calculated based on the process urgency value and queuing congestion value in the production feature set. The adaptive adjustment amount is then combined with the preset basic value of the acceptance probability parameter to generate the updated acceptance probability parameter.

[0030] The updated initial temperature, the updated rate of temperature decrease, and the updated acceptance probability parameters are combined to generate a set of adaptive parameters for simulated annealing.

[0031] Optionally, the formation of the updated harmony memory bank includes:

[0032] Based on the harmony memory selection probability, pitch adjustment rate and pitch change step size in the harmony search adaptive parameter set, an initialization operation is performed on the harmony memory library. An initial harmony memory library is formed by establishing an initial structure in the harmony memory library and generating an initial scheduling solution.

[0033] Based on the selection probability of the harmony memory, a portion of the content of the scheduling solution is selected from the harmony memory bank to form a portion of the content of the scheduling candidate solution. Then, based on the pitch adjustment rate and pitch change step size, pitch adjustment is performed on the portion of the content to generate the pitch-adjusted scheduling candidate solution content.

[0034] Random perturbations are applied to the scheduling candidate solutions that were not generated through harmonic memory selection or pitch adjustment to complete the scheduling candidate solutions, thereby generating a complete set of scheduling candidate solutions;

[0035] The scheduling evaluation value is calculated for each scheduling candidate solution in the scheduling candidate solution set according to the scheduling evaluation function, and the scheduling candidate solution evaluation result is generated based on the scheduling evaluation value;

[0036] Based on the evaluation results of the scheduling candidate solutions, the harmony memory is updated. The scheduling candidate solutions with better scheduling evaluation values ​​are saved to the harmony memory, and the scheduling solutions with poor scheduling evaluation values ​​are removed from the harmony memory, thus forming an updated harmony memory.

[0037] Optionally, updating the current scheduling solution includes:

[0038] Read the candidate scheduling solutions from the updated harmony memory, and use all the contents of the candidate scheduling solutions as the initial contents of the current scheduling solution to form the current scheduling solution;

[0039] Based on the initial temperature and temperature drop rate in the simulated annealing adaptive parameter set, a local perturbation operation is performed on a portion of the current scheduling solution. By replacing, exchanging, or adjusting some positions in the current scheduling solution, a new scheduling solution is formed.

[0040] Calculate the cost value for the new scheduling solution and the current scheduling solution respectively, and generate the cost difference by subtracting the cost value of the current scheduling solution from the cost value of the new scheduling solution.

[0041] Based on the acceptance probability parameter in the simulated annealing adaptive parameter set, an acceptance decision is made on the cost difference. The decision on whether to replace the current scheduling solution with the new scheduling solution is determined by comparing the cost difference with the acceptance probability parameter.

[0042] If the acceptance decision is "accept", the new scheduling solution is used as the updated current scheduling solution; if the acceptance decision is "disaccept", the current scheduling solution is used as the updated current scheduling solution.

[0043] Optionally, the generation of the new scheduling candidate solution and the evaluation result of the scheduling candidate solution includes:

[0044] Based on the temperature drop rate in the set of adaptive parameters for simulated annealing, the temperature of the simulated annealing algorithm is updated by performing a calculation on the temperature of the simulated annealing algorithm and the temperature drop rate to generate the updated temperature of the simulated annealing algorithm.

[0045] Write the updated current scheduling solution into the harmony memory. Replace the scheduling solutions in the harmony memory with the updated current scheduling solution to form the written harmony memory.

[0046] New scheduling candidate solutions are generated based on the written harmony memory. New scheduling candidate solutions are generated by selecting a scheduling solution from the written harmony memory or by making changes based on the selected scheduling solution.

[0047] The new scheduling candidate solutions are calculated based on the scheduling evaluation function to generate new scheduling candidate solution evaluation results.

[0048] Optionally, the generation of the final scheduling solution includes:

[0049] Based on the harmony memory, scheduling candidate solutions are generated. By selecting scheduling solutions from the harmony memory or by making changes based on scheduling solutions, scheduling candidate solutions are generated. After generating scheduling candidate solutions, the system evaluates the production performance of each scheduling candidate solution according to the scheduling evaluation function, and forms the scheduling candidate solution evaluation result.

[0050] The global search process is performed using the harmony search algorithm to transform the candidate solutions globally and form new scheduling candidate solutions, thus forming scheduling candidate solutions;

[0051] The simulated annealing algorithm is used to perform a local jump optimization process. By using the local perturbation mechanism of the simulated annealing algorithm, the current scheduling solution is changed to generate a new scheduling solution. The cost difference is generated based on the cost calculation results of the new scheduling solution and the current scheduling solution. Then, the acceptability of the new scheduling solution is determined based on the acceptance probability parameter to generate an updated current scheduling solution.

[0052] The global search process and the local jump optimization process run alternately. During the alternation process, the updated current scheduling solution is continuously generated. After each round of alternation, the new scheduling candidate solution and the updated current scheduling solution are used as the input for the next round of alternation, and the updated current scheduling solution is continuously formed.

[0053] When the updated current scheduling solution satisfies one of the iteration number threshold, the scheduling quality convergence threshold, or the immediate scheduling termination condition triggered by the manufacturing execution system, the system stops the alternating coupled hybrid optimization process and uses the updated current scheduling solution as the final schedule.

[0054] Optionally, the generation of the production scheduling scheme includes:

[0055] Based on the final scheduling solution, process the process sequence and process start and end times in the final scheduling solution to generate process order and process start and end times;

[0056] Based on the final scheduling solution, the correspondence between the processes and equipment in the final scheduling solution is processed to generate equipment allocation;

[0057] Based on the final scheduling solution, process the relevant content of line switching and mode switching in the final scheduling solution to generate line switching arrangements and mode switching arrangements;

[0058] Based on the final scheduling solution, the resource usage in the final scheduling solution is processed to generate a resource usage plan;

[0059] Based on the final scheduling solution, the sequence of operations and the duration of operations in the final scheduling solution are processed to generate critical path information;

[0060] The production scheduling plan is formed by combining process sequencing, process start and end times, equipment allocation, line changeover arrangements, mold changeover arrangements, resource utilization plans, and critical path information.

[0061] Output the production scheduling plan to the manufacturing execution system.

[0062] The beneficial effects of this invention are:

[0063] This invention utilizes historical and real-time production data collected by a manufacturing execution system to construct a set of production features reflecting the actual operating status of the workshop. By quantifying key production features such as processing time fluctuations, equipment load changes, process urgency, queuing congestion, and bottleneck process intensity, the scheduling optimization process can adaptively adjust algorithm parameters based on actual dynamic changes in production, effectively overcoming the problem that existing scheduling methods cannot accurately reflect the dynamic workshop state. Through a big data-driven feature calculation and parameter adjustment mechanism, this invention enables the scheduling algorithm to respond to fluctuations in the production environment in real time, freeing scheduling decisions from dependence on fixed parameter models and significantly improving the algorithm's adaptability and generalization ability to complex production scenarios.

[0064] This invention proposes a dual-algorithm parameter dynamic generation mechanism based on adaptive parameter sets for harmony search and simulated annealing. This mechanism enables real-time updates of parameters such as harmony memory selection probability, pitch adjustment rate, pitch change step size, initial temperature, temperature decrease rate, and acceptance probability during scheduling. This allows the harmony search algorithm's global search exploration capability and the simulated annealing algorithm's ability to escape local optima in local jumps to be simultaneously enhanced or suppressed according to changes in production status. This parameter adaptation mechanism ensures that both algorithms maintain a search strategy highly consistent with the production environment, avoiding the problems of reduced search efficiency, premature convergence, or optimization stagnation that occur in traditional fixed-parameter algorithms when production disturbances occur.

[0065] The hybrid optimization mechanism proposed in this invention, which alternates between global search and local jump optimization, establishes a dynamic feedback relationship between the harmony search algorithm and the simulated annealing algorithm. The scheduling candidate solutions generated by the harmony search algorithm are rewritten into the harmony memory after local jump optimization by the simulated annealing algorithm. Through continuous information flow, a continuously optimized and evolving scheduling process is formed. Compared with existing serial or unidirectional coupled hybrid algorithm modes, the alternating coupling mechanism of this invention significantly improves the overall search stability of the algorithm, allows scheduling solutions to flexibly switch between global and local scopes, enhances the diversity of search paths, and thus improves the quality and robustness of the final scheduling solution.

[0066] In summary, this invention addresses the problems of existing scheduling algorithms—such as difficulty in adapting to complex and dynamic production environments, susceptibility to local optima, and unstable scheduling results—through big data-driven feature set construction, a dual-algorithm adaptive parameter generation mechanism, and a hybrid scheduling mode that couples global and local data. This achieves higher scheduling accuracy, stronger dynamic response capabilities, and better computational stability. This method can improve equipment utilization, reduce production delays, alleviate bottleneck pressures, and accelerate overall production pace under real-world workshop conditions, providing crucial technical support for the intelligent and real-time development of manufacturing execution systems. Attached Figure Description

[0067] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0068] Figure 1 This is the overall flowchart of the intelligent MES system production scheduling method based on big data proposed in this invention;

[0069] Figure 2 This is a schematic diagram of the feature-driven adaptive parameter generation architecture in the big data-based intelligent MES system production scheduling method proposed in this invention.

[0070] Figure 3 This is a schematic diagram of the overall framework for hybrid scheduling optimization in the production scheduling method of the intelligent MES system based on big data proposed in this invention. Detailed Implementation

[0071] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0072] refer to Figures 1-3 The production scheduling method for an intelligent MES system based on big data includes the following steps:

[0073] The system acquires historical and real-time production data from the manufacturing execution system. The historical and real-time production data includes process processing time sequences, equipment operating status, equipment load, process queuing time, order attribute data, tooling usage data, and production line bottleneck information. The system performs data cleaning, anomaly identification, and structured processing on the historical and real-time production data to generate processed production data.

[0074] Based on the processed production data, the processing time fluctuation index, equipment load change rate, process urgency index, queuing congestion index, and bottleneck process intensity are calculated to form a set of production characteristics;

[0075] Based on the production feature set, the harmony memory selection probability, pitch adjustment rate, and pitch change step size of the harmony search algorithm are adaptively adjusted to generate a harmony search adaptive parameter set. Based on the production feature set, the initial temperature, temperature drop rate, and acceptance probability parameters of the simulated annealing algorithm are adaptively calculated to generate a simulated annealing adaptive parameter set.

[0076] The harmony memory is initialized based on the adaptive parameter set of harmony search. Multiple scheduling candidate solutions are generated through harmony memory selection, pitch adjustment and random perturbation. The scheduling candidate solutions are evaluated for quality according to the scheduling evaluation function, and the scheduling candidate solution evaluation results are generated. The harmony memory is updated based on the scheduling candidate solution evaluation results.

[0077] Select a candidate scheduling solution from the updated harmony memory as the current scheduling solution. Use the simulated annealing adaptive parameter set to perform local perturbation on the current scheduling solution to generate a new scheduling solution. Calculate the cost difference between the new scheduling solution and the current scheduling solution. Based on the acceptance probability parameter in the simulated annealing adaptive parameter set, decide whether to accept the new scheduling solution as the current scheduling solution and generate the updated current scheduling solution.

[0078] The temperature of the simulated annealing algorithm is updated according to the temperature drop rate in the simulated annealing adaptive parameter set. The updated current scheduling solution is written into the harmony memory, and a new scheduling candidate solution is generated based on the written harmony memory and the evaluation result of the new scheduling candidate solution is generated.

[0079] Based on the harmony memory, scheduling candidate solutions are continuously generated and the evaluation results of scheduling candidate solutions are obtained. The global search process is executed using the harmony search algorithm, and the local jump optimization process is executed using the simulated annealing algorithm. The global search process and the local jump optimization process are run alternately to continuously generate the updated current scheduling solution. The final scheduling solution is generated when the iteration number threshold, the scheduling quality convergence threshold, or the instantaneous scheduling termination condition triggered by the manufacturing execution system is met.

[0080] A production scheduling scheme is generated based on the final scheduling solution. The production scheduling scheme includes process sequencing, equipment allocation, process start and end times, line changeover arrangements, mold changeover arrangements, resource utilization plans, and critical path information. The production scheduling scheme is then output to the manufacturing execution system for production execution.

[0081] In this embodiment, the generation of the production feature set includes:

[0082] The processing time records of the processes in the processed production data are constructed into a processing time series according to the time order, and the processing time fluctuation value is calculated based on the processing time series. The processing time fluctuation value is generated by summing the squared differences between each processing time in the processing time series and the average processing time and performing normalization processing.

[0083] The equipment operating status records in the processed production data are aggregated according to fixed time windows, and the equipment load change value is calculated based on the equipment load changes between adjacent time windows. The equipment load change value is generated by scaling up the absolute value of the difference between the equipment load in the current time window and the equipment load in the previous time window.

[0084] The urgency value of the process is calculated based on the time difference between the order attribute data in the processed production data and the current time. The urgency value is generated by performing a reciprocal operation on the time difference between the order requirement completion time and the current processing time.

[0085] The queuing time of the process in the processed production data is grouped according to the production line number, and the queuing congestion value is calculated based on the difference between the queuing time of each production line and the historical average queuing time of that production line. The queuing congestion value is generated by dividing the queuing time difference by the historical average queuing time.

[0086] The bottleneck process intensity value is calculated based on the process processing time, equipment availability, and process parallelism in the processed production data. The bottleneck process intensity value is generated by dividing the actual processing time of the process by the equipment availability and then multiplying it by the process parallelism.

[0087] The production feature set is generated by combining the processing time fluctuation value, equipment load change value, process urgency value, queue congestion value, and bottleneck process intensity value.

[0088] In this embodiment, the generation of the simulated annealing adaptive parameter set includes:

[0089] The adaptive adjustment amount of the harmony memory selection probability is calculated based on the production feature set. The adaptive adjustment amount is combined with the preset base value of the harmony memory selection probability. The combination operation is to add the base harmony memory selection probability to the adaptive adjustment amount calculated from the production feature set, and then ensure that the value is between 0 and 1 through range constraints to generate the updated harmony memory selection probability.

[0090] The adaptive adjustment amount of the tone adjustment rate is calculated based on the processing time fluctuation value and equipment load change value in the production feature set. The adaptive adjustment amount is combined with the preset tone adjustment rate base value. The updated tone adjustment rate is generated by adding the weighted adjustment amount of the processing time fluctuation value and equipment load change value to the base tone adjustment rate and limiting the range.

[0091] The adaptive adjustment amount of the pitch change step size is calculated based on the processing time fluctuation value in the production feature set. The adaptive adjustment amount is combined with the preset pitch change step size base value. The updated pitch change step size is obtained by adding the weighted adjustment amount of the processing time fluctuation value to the base step size and limiting it within the allowable step size range.

[0092] The updated harmony memory selection probability, the updated pitch adjustment rate, and the updated pitch change step size are combined to generate a set of adaptive parameters for harmony search.

[0093] The adaptive adjustment amount of the initial temperature is calculated based on the processing time fluctuation value and the bottleneck process intensity value in the production feature set. The adaptive adjustment amount is combined with the preset initial temperature base value. The updated initial temperature is obtained by adding the weighted adjustment amount of the processing time fluctuation value and the bottleneck process intensity to the base initial temperature and keeping it no lower than the lower limit.

[0094] The adaptive adjustment amount of the temperature drop rate is calculated based on the equipment load change value in the production feature set. The adaptive adjustment amount is combined with the preset temperature drop rate base value. The updated temperature drop rate is obtained by adding the weighted adjustment amount of the equipment load change value to the base drop rate and trimming it to between zero and one.

[0095] The adaptive adjustment amount of the acceptance probability parameter is calculated based on the process urgency value and the queue congestion value in the production feature set. The adaptive adjustment amount is combined with the preset base value of the acceptance probability parameter. The updated value is generated by adding the weighted adjustment amount of the process urgency value and the queue congestion value to the base acceptance probability and limiting it to between zero and one.

[0096] The updated initial temperature, the updated rate of temperature decrease, and the updated acceptance probability parameters are combined to generate a set of adaptive parameters for simulated annealing.

[0097] In this embodiment, the formation of the updated harmony memory bank includes:

[0098] The harmony memory is initialized based on the harmony memory selection probability, pitch adjustment rate and pitch change step size in the harmony search adaptive parameter set. The initial harmony memory is formed by establishing an initial structure for storing scheduling solutions in the harmony memory and generating initial scheduling solutions.

[0099] Based on the selection probability of the harmony memory, a portion of the content of the scheduling solution is selected from the harmony memory bank to form a portion of the content of the scheduling candidate solution. Then, based on the pitch adjustment rate and pitch change step size, pitch adjustment is performed on the portion of the content to generate the pitch-adjusted scheduling candidate solution content.

[0100] Random perturbation is performed on the scheduling candidate solutions that were not generated through harmony memory selection or pitch adjustment to complete the scheduling candidate solutions, thereby generating a complete set of scheduling candidate solutions, which consists of multiple scheduling candidate solutions;

[0101] The scheduling evaluation value is calculated for each scheduling candidate solution in the scheduling candidate solution set according to the scheduling evaluation function, and the scheduling candidate solution evaluation result is generated based on the scheduling evaluation value;

[0102] Based on the evaluation results of the scheduling candidate solutions, the harmony memory is updated. The scheduling candidate solutions with better scheduling evaluation values ​​are saved to the harmony memory, and the scheduling solutions with poor scheduling evaluation values ​​are removed from the harmony memory, thus forming an updated harmony memory, which is used to continue generating scheduling candidate solutions and generating scheduling candidate solution evaluation results.

[0103] In this embodiment, the update of the current scheduling solution includes:

[0104] Read the candidate scheduling solutions from the updated harmony memory, and use all the contents of the candidate scheduling solutions as the initial contents of the current scheduling solution to form the current scheduling solution;

[0105] Based on the initial temperature and temperature drop rate in the simulated annealing adaptive parameter set, a local perturbation operation is performed on a portion of the current scheduling solution. By replacing, exchanging, or adjusting some positions in the current scheduling solution, a new scheduling solution is formed.

[0106] The generation of the new scheduling solution specifically includes determining the perturbation amplitude based on the initial temperature and temperature drop rate in the simulated annealing adaptive parameter set, selecting one or more positions from the current scheduling solution, performing local perturbation on the content of the selected positions by means of replacement, exchange or adjustment, and then performing legality repair on the perturbated scheduling solution to form a new scheduling solution;

[0107] The cost value is calculated for the new scheduling solution and the current scheduling solution respectively. The cost value calculation is to calculate a comprehensive value by calculating the production indicators such as completion time, delay, load imbalance, congestion and bottleneck pressure corresponding to the scheduling solution according to a preset weight, and to generate a cost difference by subtracting the cost value of the current scheduling solution from the cost value of the new scheduling solution.

[0108] Based on the acceptance probability parameter in the simulated annealing adaptive parameter set, an acceptance decision is made on the cost difference. The decision on whether to replace the current scheduling solution with the new scheduling solution is determined by comparing the cost difference with the acceptance probability parameter.

[0109] If the acceptance decision is "accept", the new scheduling solution is used as the updated current scheduling solution; if the acceptance decision is "disaccept", the current scheduling solution is used as the updated current scheduling solution.

[0110] In this embodiment, the generation of the new scheduling candidate solution and the evaluation result of the scheduling candidate solution includes:

[0111] Based on the temperature drop rate in the set of adaptive parameters for simulated annealing, the temperature of the simulated annealing algorithm is updated by performing a calculation on the temperature of the simulated annealing algorithm and the temperature drop rate to generate the updated temperature of the simulated annealing algorithm.

[0112] Write the updated current scheduling solution into the harmony memory. Replace the scheduling solutions in the harmony memory with the updated current scheduling solution to form the written harmony memory.

[0113] New scheduling candidate solutions are generated based on the written harmony memory. New scheduling candidate solutions are generated by selecting a scheduling solution from the written harmony memory or by making changes based on the selected scheduling solution.

[0114] The new scheduling candidate solutions are calculated based on the scheduling evaluation function to generate new scheduling candidate solution evaluation results.

[0115] In this embodiment, the generation of the final scheduling solution includes:

[0116] Based on the harmony memory, scheduling candidate solutions are generated. By selecting scheduling solutions from the harmony memory or by making variations based on scheduling solutions, scheduling candidate solutions are generated. After generating scheduling candidate solutions, the system evaluates the production performance of each scheduling candidate solution according to the scheduling evaluation function. The evaluation content includes the performance of the scheduling solution in terms of process connection, equipment utilization, delay level, bottleneck impact, etc. Finally, the scheduling candidate solution evaluation result is formed to guide the subsequent global and local optimization process.

[0117] The system uses the harmony search algorithm to perform a global search process. During the global search, the system selects high-quality candidate solutions as the basis for further exploration based on the evaluation results of the scheduling candidate solutions. At the same time, it combines the specific operation mechanism of the harmony search algorithm, such as structural inheritance based on harmony memory selection, structural micro-change based on pitch adjustment, and exploration expansion based on random perturbation, to make global changes to the candidate solutions to form new scheduling candidate solutions, thereby forming scheduling candidate solutions for global search.

[0118] The simulated annealing algorithm is used to perform a local jump optimization process. By adjusting the local structure of the current scheduling solution in a small range according to the local perturbation mechanism of the simulated annealing algorithm, the exploration behavior of the solution space neighborhood is simulated, the current scheduling solution is changed to generate a new scheduling solution, and a cost difference is generated based on the cost value calculation results of the new scheduling solution and the current scheduling solution. Then, the acceptability of the new scheduling solution is performed according to the acceptance probability parameter to generate an updated current scheduling solution. The simulated annealing algorithm has the characteristic of probabilistically accepting poor solutions. The local jump optimization process enables the system to jump out of the local optimal region, improving the flexibility of the overall optimization.

[0119] The global search process and the local jump optimization process run alternately. Logically, a global search operation based on the harmony search algorithm is performed first, followed by a local jump optimization operation based on the simulated annealing algorithm. The results of the two operations are then used for the next round of alternating execution. Through this alternation, a stable bidirectional feedback relationship is formed between global exploration and local optimization, enabling the scheduling solution to maintain both solution space coverage and fine-grained structural adjustment capabilities, thereby gradually approaching a higher-quality scheduling solution. During the alternating operation, an updated current scheduling solution is continuously generated. After each round of alternating operation, the new scheduling candidate solution and the updated current scheduling solution are used as input for the next round of alternating operation, and an updated current scheduling solution is continuously formed.

[0120] When the updated current scheduling solution satisfies one of the iteration count threshold, the scheduling quality convergence threshold, or the immediate scheduling termination condition triggered by the manufacturing execution system, the system stops the alternating coupled hybrid optimization process. In each iteration, the global search process and the local jump optimization process are executed sequentially and form an alternating loop. In this loop, the harmony search algorithm is responsible for exploring the scheduling solution space over a wide area, while the simulated annealing algorithm is responsible for fine-tuning and local jumps. The two are deeply coupled through writing back the solution, regenerating the solution, and passing on the evaluation results. The updated current scheduling solution is taken as the final scheduling solution. This termination mechanism ensures that the system can respond to the dynamic instructions of the manufacturing execution system in a timely manner while ensuring scheduling quality, and achieves a scheduling optimization result that balances stability and real-time performance.

[0121] In this embodiment, the generation of the production scheduling scheme includes:

[0122] Based on the final scheduling solution, the process sequence in the final scheduling solution is processed to generate a process order;

[0123] Based on the final scheduling solution, the correspondence between the processes and equipment in the final scheduling solution is processed to generate equipment allocation;

[0124] Based on the final scheduling solution, process the process start time and process end time in the final scheduling solution to generate process start and end times;

[0125] Based on the final scheduling solution, process the relevant content of the line switching operation in the final scheduling solution to generate a line switching schedule;

[0126] Based on the final scheduling solution, process the relevant content of the mode change execution in the final scheduling solution to generate the mode change arrangement;

[0127] Based on the final scheduling solution, the resource usage in the final scheduling solution is processed to generate a resource usage plan;

[0128] Based on the final scheduling solution, the sequence of operations and the duration of operations in the final scheduling solution are processed to generate critical path information;

[0129] The production scheduling plan is formed by combining process sequencing, equipment allocation, process start and end times, line changeover arrangements, mold changeover arrangements, resource utilization plans, and critical path information.

[0130] The production scheduling plan is output to the manufacturing execution system, so that the manufacturing execution system can execute production according to the production scheduling plan.

[0131] Example 1:

[0132] To verify the feasibility of this invention in practice, it was applied to a multi-production-line discrete manufacturing workshop of an automotive parts manufacturing company. This workshop includes various types of equipment, multiple product orders, and multiple processing steps. Daily production faces problems such as unstable processing times, significant fluctuations in equipment load, frequent order insertions, and bottlenecks in some equipment. Traditional scheduling methods often rely on fixed scheduling parameters, which cannot respond to dynamic changes in a timely manner, easily leading to localized congestion, idle equipment, and unbalanced overall scheduling, severely impacting production efficiency. This invention addresses these pain points by using big data-driven approaches, adaptive parameter generation, and a hybrid optimization mechanism to improve overall scheduling quality and real-time performance.

[0133] In practical applications, historical and real-time production data are continuously acquired from the Manufacturing Execution System (MES), including process processing time records, equipment operating status, equipment load changes, process waiting time, order attribute information, tooling usage records, and bottleneck process monitoring information. The system then cleans, removes anomalies, and structures these data to obtain processed data that reflects the true state of the workshop.

[0134] Based on the processed data, the system calculates processing time fluctuations, equipment load changes, process urgency, queue congestion, and bottleneck intensity, and combines these results into a production feature set to characterize the overall load distribution and production pressure in the current workshop. When the workshop experiences increased processing time fluctuations, sudden changes in equipment load, or increased order urgency, the feature set will immediately reflect these changes.

[0135] This invention adaptively adjusts the key parameters of the harmony search algorithm and the simulated annealing algorithm using a set of production features. The harmony search part automatically adjusts the harmony memory selection probability, pitch adjustment rate, and pitch change step size based on processing time fluctuations and equipment load changes, enhancing global search capabilities during periods of production instability. The simulated annealing part dynamically adjusts the initial temperature, temperature drop rate, and acceptance probability according to the bottleneck process intensity, process urgency, and queuing congestion, allowing local disturbances and jump capabilities to adapt to the complexity of the current scheduling solution, thus achieving flexible parameter-level adjustments.

[0136] Next, the system initializes the harmony memory based on the adaptive parameter set for harmony search. It then generates a batch of candidate scheduling solutions through harmony memory selection, pitch adjustment, and random perturbation, and calculates the comprehensive performance of each candidate solution using a scheduling evaluation function. The system retains high-quality scheduling solutions and discards poor ones, forming an updated harmony memory. A better scheduling solution is then selected from the harmony memory as the current scheduling solution, and the system proceeds to the simulated annealing stage. The simulated annealing algorithm generates new local perturbation solutions based on the adaptive parameter set and uses cost difference and acceptance probability mechanisms to determine whether to replace the current scheduling solution, thus avoiding getting trapped in local optima.

[0137] In the hybrid scheduling optimization mechanism of this invention, harmony search is responsible for exploring various scheduling structures globally, while simulated annealing is responsible for deeply mining potential better solutions in the local space. The two algorithms run alternately in the system, enabling the optimization process to have both wide-area search and fine-tuning capabilities. When the system detects that the number of iterations has reached a threshold, the scheduling solution gradually stabilizes, or the manufacturing execution system requires immediate generation of a scheduling plan, the current optimal scheduling solution is determined as the final scheduling solution.

[0138] After determining the final scheduling solution, the system generates a complete production scheduling plan based on the process sequence, equipment selection, start and end time arrangements, line change information, mold change information, resource usage, and critical path relationship in the scheduling solution, and pushes it to the manufacturing execution system in real time so that the workshop can execute production activities according to the latest plan.

[0139] In continuous operation within a real workshop, this invention demonstrates significant advantages. When equipment temporarily shuts down, orders are urgently inserted, or the processing time for a certain step is suddenly extended, traditional static scheduling methods often fail to respond quickly. However, this invention, by updating the production feature set in real time, automatically adjusting scheduling parameters, and restarting the hybrid scheduling optimization process, can rapidly generate new scheduling solutions, ensuring that production scheduling always closely follows the current production status. Furthermore, the alternating operation of harmony search and simulated annealing effectively avoids premature convergence and structural rigidity in the scheduling process, resulting in more stable scheduling quality throughout the production process, stronger executability of the final production plan, and greater flexibility in responding to dynamic changes.

[0140] As can be seen from the above embodiments, the present invention can effectively solve the problems of large processing fluctuations, rapid load changes, and frequent order conflicts that are difficult to handle by traditional scheduling methods in complex manufacturing workshops, and achieve a more intelligent, real-time, and efficient scheduling optimization effect, which is of practical significance for improving the overall workshop production efficiency and intelligence level.

[0141] Comparative Experiment Report on Production Scheduling Methods for Intelligent MES Systems Based on Big Data (Extended Version)

[0142] Table 1 Performance Comparison of Three Scheduling Methods

[0143] index Traditional scheduling methods Improved Genetic Algorithm Method Method of the present invention Scheduling time 180 seconds 135 seconds 65 seconds Average completion time 12 hours 10.5 hours 9 hours Latency 18% 12% 6% Equipment utilization rate 72% 80% 89% Rescheduling response time 95 seconds 52 seconds 28 seconds

[0144] As can be seen from the experimental comparison data, the big data-driven intelligent MES system production scheduling method proposed in this invention is significantly better than the traditional scheduling method and the improved genetic algorithm method in all core indicators, demonstrating a comprehensive improvement in scheduling efficiency, scheduling quality and dynamic response capability.

[0145] First, in terms of scheduling time, the method of this invention can complete the scheduling in only 65 seconds, which is more than two-thirds less than the 180 seconds of the traditional method, and also significantly better than the 135 seconds of the improved genetic algorithm. The significant improvement in scheduling speed comes from the harmony search-simulated annealing hybrid optimization framework of this invention. Harmony search is responsible for the global search structure, while simulated annealing provides local jump capability. Combined with the parameter adaptive mechanism based on big data features, the search path is closer to the optimal solution region, thereby reducing invalid iterations and effectively reducing scheduling time.

[0146] In terms of average completion time, the method of this invention is 9 hours, which is 3 hours shorter than the traditional scheduling of 12 hours and also better than the improved genetic algorithm of 10.5 hours. This improvement reflects that the invention can more effectively reduce process waiting, equipment conflict and queuing congestion. Through big data analysis, features such as processing fluctuations, bottleneck intensity and equipment load change rate are used to participate in scheduling optimization, so that the algorithm can continuously approach a more reasonable job sorting and resource allocation scheme at both the global and local levels, ultimately reducing the overall production cycle.

[0147] Regarding latency, the method of this invention achieves a latency of 6%, a 67% reduction compared to the traditional method's 18%, and a 50% reduction compared to the improved genetic algorithm's 12%. This significant advantage demonstrates that this invention possesses better scheduling adaptability when handling urgent orders, interim orders, and orders with significant differences in urgency. Because this invention introduces process urgency and queuing congestion index into its adaptive mechanism, the system automatically tends to prioritize urgent processes when generating and adjusting scheduling solutions, reducing delays caused by queuing and thus significantly lowering the overall latency.

[0148] Regarding equipment utilization, the method of this invention achieves 89%, significantly higher than the 72% of traditional scheduling and the 80% of the improved genetic algorithm. This increased equipment utilization indicates that the invention can more effectively balance the load on workshop equipment, reducing local equipment idleness or overload. Based on the characteristics of equipment load change rates from big data, the parameters of the harmony search algorithm and the simulated annealing algorithm dynamically change, thereby prompting the algorithms to proactively correct the problem of uneven equipment usage during the search process. This results in a more rational distribution of processing tasks among resources, thus improving equipment utilization efficiency.

[0149] In terms of rescheduling response time, the method of this invention requires only 28 seconds, while the traditional method requires 95 seconds and the improved genetic algorithm requires 52 seconds. The significantly faster rescheduling response capability of this invention demonstrates its stronger real-time performance in dealing with dynamic events. This is because the adaptive parameter mechanism allows the algorithm to quickly expand and adjust from the current harmony memory and the explored local solution state without re-initialization, enabling rescheduling to be completed in a much shorter time than the traditional method.

[0150] The above data analysis demonstrates that the method of this invention exhibits significant advantages in scheduling speed, resource utilization, production efficiency, and dynamic adaptability. These improvements do not stem from a single optimization strategy, but rather from the synergistic effect of a big data-driven adaptive parameter system and a harmonic search-simulated annealing hybrid optimization structure. The dual optimization architecture, encompassing both global and local optimization, enables the algorithm to search a wide range of potential solutions while simultaneously performing efficient local refinement around the solution region. Furthermore, the dynamic parameter adjustment mechanism based on production characteristics allows the algorithm to select more suitable search paths when facing different load states, bottleneck situations, and process urgency. It is precisely this multi-layered, data-driven comprehensive optimization strategy that allows the method of this invention to achieve significantly superior performance compared to existing technologies across all test metrics.

[0151] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A production scheduling method for an intelligent MES system based on big data, characterized in that, Includes the following steps: Acquire historical and real-time production data from the manufacturing execution system, perform data cleaning, anomaly identification, and structured processing, and generate processed production data; Based on the processed production data, the processing time fluctuation index, equipment load change rate, process urgency index, queuing congestion index, and bottleneck process intensity are calculated to form a set of production characteristics; Based on the production feature set, adaptive adjustment calculations are performed on the harmony search algorithm and the simulated annealing algorithm to generate the harmony search adaptive parameter set and the simulated annealing adaptive parameter set; The harmony memory is initialized based on the adaptive parameter set of the harmony search and multiple scheduling candidate solutions are generated. The quality of the scheduling candidate solutions is evaluated and the harmony memory is updated. Based on the updated harmony memory, the current scheduling solution is selected, local perturbation is performed using the simulated annealing adaptive parameter set, and an updated current scheduling solution is generated based on the cost difference and acceptance probability parameters. Update the temperature of the simulated annealing algorithm according to the temperature drop rate, write the updated current scheduling solution into the harmony memory, and generate new scheduling candidate solution evaluation results; The global search and local jump optimization process is performed based on the harmony memory library, and the final scheduling solution is generated when the termination condition is met. The final scheduling solution is used to generate a production scheduling plan, which is then output to the manufacturing execution system.

2. The production scheduling method for an intelligent MES system based on big data according to claim 1, characterized in that, The generation of the production feature set includes: The processing time records of each process in the processed production data are used to construct a processing time series in chronological order, and the processing time fluctuation value is calculated based on the processing time series. The equipment operating status records in the processed production data are aggregated according to fixed time windows, and the equipment load change value is calculated based on the equipment load change between adjacent time windows. Calculate the process urgency value based on the time difference between the order requirement completion time and the current time in the processed production data's order attribute data. The queuing time of each process in the processed production data is grouped according to the production line number, and the queuing congestion value is calculated based on the difference between the queuing time of each production line and the historical average queuing time of that production line. The bottleneck process intensity value is calculated based on the process processing time, equipment availability, and process parallelism in the processed production data. The production feature set is generated by combining the processing time fluctuation value, equipment load change value, process urgency value, queue congestion value, and bottleneck process intensity value.

3. The production scheduling method for an intelligent MES system based on big data according to claim 1, characterized in that, The generation of the simulated annealing adaptive parameter set includes: The adaptive adjustment amount of the harmony memory selection probability is calculated based on the production feature set. The adaptive adjustment amount is then combined with the preset base value of the harmony memory selection probability to generate the updated harmony memory selection probability. The adaptive adjustment amount of the tone adjustment rate is calculated based on the processing time fluctuation value and equipment load change value in the production feature set. The adaptive adjustment amount is then combined with the preset tone adjustment rate base value to generate the updated tone adjustment rate. The adaptive adjustment amount of the pitch change step size is calculated based on the processing time fluctuation value in the production feature set. The adaptive adjustment amount is then combined with the preset pitch change step size base value to obtain the updated pitch change step size. The updated harmony memory selection probability, the updated pitch adjustment rate, and the updated pitch change step size are combined to generate a set of adaptive parameters for harmony search. The adaptive adjustment amount of the initial temperature is calculated based on the processing time fluctuation value and bottleneck process intensity value in the production feature set. The adaptive adjustment amount is then combined with the preset initial temperature base value to obtain the updated initial temperature. The adaptive adjustment amount of the temperature drop rate is calculated based on the equipment load change value in the production feature set. The adaptive adjustment amount is then combined with the preset temperature drop rate base value to obtain the updated temperature drop rate. The adaptive adjustment amount of the acceptance probability parameter is calculated based on the process urgency value and queuing congestion value in the production feature set. The adaptive adjustment amount is then combined with the preset basic value of the acceptance probability parameter to generate the updated acceptance probability parameter. The updated initial temperature, the updated rate of temperature decrease, and the updated acceptance probability parameters are combined to generate a set of adaptive parameters for simulated annealing.

4. The production scheduling method for an intelligent MES system based on big data according to claim 1, characterized in that, The formation of the updated harmony memory bank includes: Based on the harmony memory selection probability, pitch adjustment rate and pitch change step size in the harmony search adaptive parameter set, an initialization operation is performed on the harmony memory library. An initial harmony memory library is formed by establishing an initial structure in the harmony memory library and generating an initial scheduling solution. Based on the selection probability of the harmony memory, a portion of the content of the scheduling solution is selected from the harmony memory bank to form a portion of the content of the scheduling candidate solution. Then, based on the pitch adjustment rate and pitch change step size, pitch adjustment is performed on the portion of the content to generate the pitch-adjusted scheduling candidate solution content. Random perturbations are applied to the scheduling candidate solutions that were not generated through harmonic memory selection or pitch adjustment to complete the scheduling candidate solutions, thereby generating a complete set of scheduling candidate solutions; The scheduling evaluation value is calculated for each scheduling candidate solution in the scheduling candidate solution set according to the scheduling evaluation function, and the scheduling candidate solution evaluation result is generated based on the scheduling evaluation value; Based on the evaluation results of the scheduling candidate solutions, the harmony memory is updated. The scheduling candidate solutions with better scheduling evaluation values ​​are saved to the harmony memory, and the scheduling solutions with poor scheduling evaluation values ​​are removed from the harmony memory, thus forming an updated harmony memory.

5. The production scheduling method for an intelligent MES system based on big data according to claim 1, characterized in that, The update of the current scheduling solution includes: Read the candidate scheduling solutions from the updated harmony memory, and use all the contents of the candidate scheduling solutions as the initial contents of the current scheduling solution to form the current scheduling solution; Based on the initial temperature and temperature drop rate in the simulated annealing adaptive parameter set, a local perturbation operation is performed on a portion of the current scheduling solution. By replacing, exchanging, or adjusting some positions in the current scheduling solution, a new scheduling solution is formed. Calculate the cost value for the new scheduling solution and the current scheduling solution respectively, and generate the cost difference by subtracting the cost value of the current scheduling solution from the cost value of the new scheduling solution. Based on the acceptance probability parameter in the simulated annealing adaptive parameter set, an acceptance decision is made on the cost difference. The decision on whether to replace the current scheduling solution with the new scheduling solution is determined by comparing the cost difference with the acceptance probability parameter. If the acceptance decision is "accept", the new scheduling solution is used as the updated current scheduling solution; if the acceptance decision is "disaccept", the current scheduling solution is used as the updated current scheduling solution.

6. The production scheduling method for an intelligent MES system based on big data according to claim 1, characterized in that, The generation of the new scheduling candidate solution and the evaluation results of the scheduling candidate solution includes: Based on the temperature drop rate in the set of adaptive parameters for simulated annealing, the temperature of the simulated annealing algorithm is updated by performing a calculation on the temperature of the simulated annealing algorithm and the temperature drop rate to generate the updated temperature of the simulated annealing algorithm. Write the updated current scheduling solution into the harmony memory. Replace the scheduling solutions in the harmony memory with the updated current scheduling solution to form the written harmony memory. New scheduling candidate solutions are generated based on the written harmony memory. New scheduling candidate solutions are generated by selecting a scheduling solution from the written harmony memory or by making changes based on the selected scheduling solution. The new scheduling candidate solutions are calculated based on the scheduling evaluation function to generate new scheduling candidate solution evaluation results.

7. The production scheduling method for an intelligent MES system based on big data according to claim 1, characterized in that, The generation of the final scheduling solution includes: Based on the harmony memory, scheduling candidate solutions are generated. By selecting scheduling solutions from the harmony memory or by making changes based on scheduling solutions, scheduling candidate solutions are generated. After generating scheduling candidate solutions, the system evaluates the production performance of each scheduling candidate solution according to the scheduling evaluation function, and forms the scheduling candidate solution evaluation result. The global search process is performed using the harmony search algorithm to transform the candidate solutions globally and form new scheduling candidate solutions, thus forming scheduling candidate solutions; The simulated annealing algorithm is used to perform a local jump optimization process. By using the local perturbation mechanism of the simulated annealing algorithm, the current scheduling solution is changed to generate a new scheduling solution. The cost difference is generated based on the cost calculation results of the new scheduling solution and the current scheduling solution. Then, the acceptability of the new scheduling solution is determined based on the acceptance probability parameter to generate an updated current scheduling solution. The global search process and the local jump optimization process run alternately. During the alternation process, the updated current scheduling solution is continuously generated. After each round of alternation, the new scheduling candidate solution and the updated current scheduling solution are used as the input for the next round of alternation, and the updated current scheduling solution is continuously formed. When the updated current scheduling solution satisfies one of the iteration number threshold, the scheduling quality convergence threshold, or the immediate scheduling termination condition triggered by the manufacturing execution system, the system stops the alternating coupled hybrid optimization process and uses the updated current scheduling solution as the final schedule.

8. The production scheduling method for an intelligent MES system based on big data according to claim 1, characterized in that, The generation of the production scheduling scheme includes: Based on the final scheduling solution, process the process sequence and process start and end times in the final scheduling solution to generate process order and process start and end times; Based on the final scheduling solution, the correspondence between the processes and equipment in the final scheduling solution is processed to generate equipment allocation; Based on the final scheduling solution, process the relevant content of line switching and mode switching in the final scheduling solution to generate line switching arrangements and mode switching arrangements; Based on the final scheduling solution, the resource usage in the final scheduling solution is processed to generate a resource usage plan; Based on the final scheduling solution, the sequence of operations and the duration of operations in the final scheduling solution are processed to generate critical path information; The production scheduling plan is formed by combining process sequencing, process start and end times, equipment allocation, line changeover arrangements, mold changeover arrangements, resource utilization plans, and critical path information. Output the production scheduling plan to the manufacturing execution system.

Images