Beer production scheduling method under uncertain condition

Abstract

The invention provides a beer production scheduling method under an uncertain condition aiming at the scheduling problem in beer production. According to the method, through an iterative optimization strategy, a target function is adjusted step by step, an alternating mechanism is introduced, an original target is used as a guide in the initial stage of optimization, an evaluation function is alternately adopted in the later stage, exploration and development are balanced, and a trap of a local optimal solution is avoided. According to the algorithm, a non-dominated sorting genetic algorithm (NSGA-II) is adopted, population diversity is maintained through selection, crossover and mutation operation, a Pareto frontier solution set is generated through a non-dominated sorting mechanism, and the multi-objective optimization problem is effectively solved. Finally, the algorithm provides a series of Pareto optimal solutions, supports decision makers to perform tradeoff selection among productivity, inventory and other optimization objectives, and significantly improves the flexibility and efficiency of beer production scheduling.

Description

Technical Field

[0001] The invention relates to the field of process industry and is a beer production scheduling method under uncertain conditions. Background Art

[0002] Beer is the third most consumed beverage in the world after water and tea, and has a broad market base in many countries and regions. In particular, in the Chinese market, beer consumption continues to grow, becoming one of the most popular drinks in summer. This consumption trend reflects the important position of beer in daily life and social culture, and also promotes increasingly fierce competition in the beer production industry. The beer production process is complex and highly systematized, including multiple processes such as saccharification, fermentation, filtration, and filling. There is a strong dependency between the processes and the production cycle is long, which puts extremely high demands on scheduling efficiency and process stability. In this context, optimizing production scheduling has become a key means for enterprises to improve efficiency, reduce costs, and accelerate market response. As a constrained optimization problem, production scheduling aims to maximize efficiency and minimize costs under limited conditions by reasonably allocating resources and arranging tasks.

[0003] Although researchers at home and abroad have proposed a variety of optimization methods for production scheduling problems, there are still many deficiencies in practical applications. On the one hand, single-objective optimization methods (such as those based on heuristic algorithms, genetic algorithms, etc.) usually only focus on a specific goal (such as the shortest completion time or the lowest energy consumption), which is difficult to meet the needs of multi-objective comprehensive optimization in actual production. In addition, although there has been progress in the research on multi-objective dynamic scheduling problems, Chinese patent application CN117910889A discloses a method that uses order advance and delay costs and production costs as objective functions, and converts multiple objectives into a single objective through weight coefficients for solving, but this method often loses the diversity of solutions during the optimization process and cannot provide decision makers with sufficient range of choices. At the same time, these methods are difficult to fully adapt to the long cycle and complex process dependencies unique to beer production.

[0004] On the other hand, the uncertainty factors that are prevalent in production scheduling (such as equipment failure, order changes, logistics delays, etc.) significantly increase the difficulty of scheduling. Although some current studies have introduced robust optimization and dynamic adjustment methods, most of them are aimed at specific scenarios or single objectives, and lack comprehensive support for complex multi-objective requirements.

[0005] Therefore, in view of the complex requirements of multi-objective optimization and dynamic adjustment in beer production scheduling, it is urgent to develop a new scheduling method to address the shortcomings of existing technologies in terms of solution diversity, dynamic adjustment capability and robustness. This method needs to be able to effectively balance the requirements of multi-objective optimization, while having the ability to cope with uncertainties and emergencies, and provide a more efficient, flexible and reliable scheduling solution for actual production. Summary of the invention

[0006] The present invention proposes a beer production scheduling method under uncertain conditions, which realizes multi-objective scheduling optimization and production continuity guarantee through a number of innovative technical measures, and shows high adaptability and flexibility, especially under equipment failure conditions. Specifically, the method includes:

[0007] S01: Schedule the saccharification, filtration, boiling, clarification or fermentation process of beer production;

[0008] S02: In the first iteration, the original objective function is used to optimize the scheduling scheme based on the production process time requirements and equipment capacity constraints;

[0009] S03: In each subsequent iteration, the original objective function or the evaluation function is selected as the optimization target in turn;

[0010] S04: Dynamically adjust the optimization objective function according to the dynamics of equipment failure or changes in the production environment;

[0011] S05: Optimizing Pareto frontier solutions using non-dominated sorting genetic algorithm;

[0012] S06: Output a set of Pareto optimal solutions.

[0013] Preferably, the method monitors key process parameters in real time during the dynamic scheduling process, including saccharification efficiency, filtration loss, boiling time, clarification effect or fermentation status.

[0014] Preferably, the method adjusts the optimization function of the scheduling model according to a preset change rule during each iteration. Preferably, the method selects the original objective function or evaluation function in rotation based on the principle of the shortest scheduling priority time of the saccharification, filtration, boiling and clarification processes. This method can balance exploration and development, avoid the algorithm from falling into a local optimum, and realize a comprehensive search of the solution space.

[0015] Preferably, the method dynamically adjusts the activation state of the fermentation equipment and the raw material allocation plan according to the capacity of the fermentation tank and the limitation of the fermentation time.

[0016] Preferably, the evaluation function evaluates the impact of equipment failure or raw material shortage on overall production efficiency, and constructs a penalty function to adjust the scheduling plan to reduce costs and losses. The evaluation function can cope with the impact of equipment failure on production scheduling.

[0017] Preferably, the non-dominated sorting genetic algorithm maintains the diversity of the population and approaches the optimal solution set by simulating the selection, crossover and mutation operations in the natural evolution process.

[0018] Preferably, the core of the non-dominated sorting genetic algorithm is a non-dominated sorting mechanism, which can sort the solution set and select solutions that perform well on multiple objectives.

[0019] Preferably, the method stops iterating when a predetermined maximum number of iterations is reached or a preset convergence condition is reached, and outputs a series of Pareto graphical optimal solutions representing different trade-off schemes when the iteration is stopped.

[0020] Preferably, the method establishes a dynamic balance model of production capacity and inventory in different process links. The method is suitable for handling complex multi-batch production tasks, and achieves a simultaneous improvement in beer quality and output by optimizing the time and resource allocation of the production process.

[0021] Compared with the prior art, the present invention has the following beneficial effects:

[0022] (1) It can simultaneously meet multiple goals in production scheduling, such as capacity maximization and inventory control;

[0023] (2) In view of the possible equipment failure during the production process, a dynamic optimization target adjustment mechanism is introduced to achieve flexible adjustment through penalty functions, so that the continuity and stability of production under failure conditions can be guaranteed;

[0024] (3) Output a series of Pareto optimal solutions to provide decision makers with flexible trade-off options between multiple objectives;

[0025] (4) The algorithm design comprehensively considers the balance between exploration and development, avoids local optimal problems, and improves the overall optimization effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0027] Figure 1 A flow chart of an algorithm according to an embodiment of the present invention is shown.

[0028] Figure 2 The optimization algorithm framework of the embodiment of the present invention is shown.

[0029] Figure 3 A diagram showing simulation results under a fault condition provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0030] Beer brewing follows a sophisticated and systematic production process that covers all key stages from raw material processing to the final product. Typical processes include but are not limited to gelatinization, saccharification, filtration, boiling, clarification and fermentation. Among them, the saccharification process involves converting starch in malt into fermentable sugars, which is a crucial step in the brewing process; then, the mash passes through the filter tank to remove solid residues and enters the boiling pot, where hops are added to give the beer a unique flavor and aroma, and sterilization and protein precipitation are carried out; the clarification step ensures the purity of the liquid; and fermentation is completed in a separate fermentation facility, where yeast converts fermentable sugars into alcohol and carbon dioxide, ultimately forming the basic flavor structure of beer. The first four of the above processes - gelatinization, saccharification, filtration, boiling and clarification are all performed in the same brewing room, while the fermentation process is carried out in a dedicated fermentation room. To further illustrate the time cycle required for each process, Table 1 provides detailed data support. Analysis of these data shows that except for the fermentation process, which takes a long time due to its biochemical nature, the other steps in the saccharification room take a short time. This time difference reflects the different requirements for time and temperature control in different processes, and also reflects the need for a balance between efficiency and quality in the brewing process.

[0031] Tables 2 and 3 provide detailed quantitative indicators for storage capacity and equipment processing capacity. The former shows the maximum storage capacity of the intermediate products obtained after each step, while the latter indicates the upper limit of the materials that the relevant equipment can handle. This information is crucial for optimizing production plans, because the scheduling goal is to maximize production capacity within 15 days while reducing the inventory of intermediate products, thereby reducing storage costs and increasing the flexibility of the entire production line. In addition, in the face of sudden equipment failures, the robust design of the production system is also particularly important to ensure that basic production activities can be maintained even under adverse conditions, thereby ensuring production continuity and product stability.

[0032] Table 1. Time required for beer production process Technology Saccharification filter boiled Shen Qing Fermentation time 185 minutes 95 minutes 91 minutes 84 minutes 13 days

[0033] Table 2. Maximum capacity of products (unit: tons)

[0034] Table 3. Maximum processing capacity of equipment (unit: tons) equipment Mashing Pot Filter tank Boiling Pot Sedimentation tank Fermentation tank Processing power 60 60 60 60 120

[0035] In order to better solve the beer production scheduling problem, we make the following assumptions:

[0036] 1. The scheduling period is divided into h time intervals, whose length is τ = 1.5h. t is the start time of the tth scheduling period.

[0037] 2. During the production process, the processing time of different batches of products on various processing equipment remains unchanged.

[0038] 3. Any transfer of intermediate products between different devices is considered to be completed instantly.

[0039] Please refer to the flowchart of the algorithm Figure 1 . Beer production scheduling is essentially a constrained optimization problem. The scheduling goal of maximizing production capacity and minimizing intermediate products can be used as the optimization objective function of the model. Therefore, the objective function of the model is shown in formula (1). Where t is the scheduling cycle, l is the number of production lines, h is the number of fermentation equipment on each production line, and g is the g-th process in the saccharification workshop. B lht represents the batch size of the fermentation equipment in the scheduling period t, B lgt It represents the batch size of the tasks processed by the saccharification workshop equipment within the scheduling period t. h and θ g Both represent the output ratio coefficient. 1 Represents the capacity of beer, F 2 Represents the amount of intermediate products of beer.

[0040]

[0041] The constraint function part of the scheduling model is divided into the saccharification workshop part and the fermentation workshop part. The constraint conditions of the saccharification workshop are shown in formulas (2) to (6). Formula (2) means that new materials should not be processed after the (g+1)th equipment is started. gt Indicates the storage capacity of the material produced by the (g+1)th equipment. l(g+1)t Indicates whether the device is started in the scheduling period t, 0 means not started, 1 means started. When the device fails, X l(g+1)t The value of is 0. Formula (3) indicates that the amount of material handled by the equipment should not exceed its capacity. It represents the maximum processing capacity of the g-th equipment. Formula (4) means that the storage volume of the material should not exceed its storage limit. It refers to the storage limit of materials produced by equipment of type g. Formula (5) represents the raw material supply constraint. represents the consumption coefficient. Formula (6) represents the material balance constraint.

[0042] [SI g(t+1) -SI gt ]X l(g+1)t e≥0#(2)

[0043]

[0044]

[0045]

[0046]

[0047] The constraints of the fermentation plant are shown in formulas (7) to (9). Formula (7) indicates that the amount of material processed by the fermentation equipment should not exceed its capacity. Refers to the maximum processing capacity of high temperature and high pressure equipment. lht Indicates whether the equipment is started in the scheduling period t, 0 means not started, and 1 means started. Formula (8) means that the storage volume of the material should not exceed its storage limit. It refers to the storage limit of the material produced by the hth fermentation equipment. Formula (9) represents the material balance constraint.

[0048]

[0049]

[0050]

[0051] Based on the aforementioned beer production process and its scheduling model, a dynamic scheduling algorithm is proposed in this invention to solve the optimization scheduling problem in beer production. The design goal of this algorithm is to enhance the system's ability to cope with sudden equipment failures while ensuring the efficient operation of the production line, thereby ensuring the continuity and stability of the production process to the greatest extent.

[0052] In the production process, the impact of equipment failure on production is certainly related to the processing capacity of the equipment and the required maintenance time. Based on this, the corresponding penalty function (10) is constructed. In the formula, B bro Represents the processing capability of the faulty device, Δt rep represents the repair time required for the faulty equipment. 1 Functions are combined to construct an improved objective function F′ 1 It is used as the optimization objective function in the dynamic scheduling algorithm, as shown in formula (11). This indirectly amplifies the impact of equipment failures, making it easier for the algorithm to make corresponding adjustments to the failures during the optimization process.

[0053] f pusnish =B bro Δt rep #(10)

[0054]

[0055] Some literatures simply replace the objective function in the model with formula (11) without making further improvements. In order to achieve better optimization results, the present invention proposes a three-level optimization framework based on the improved objective function and beer production scheduling model proposed in the previous article. For the specific framework, please refer to Figure 2 The overall algorithm design adopts the idea of ​​parallel optimization in multidisciplinary optimization. The initial level, discipline level and system level are all composed of different objective functions and constraint functions.

[0056] Figure 2 The 15 initial stages in represent the production of the saccharification workshop for 15 days. The objective function of the initial stage is shown in formula (12). t B lgt θ g represents the daily output of the g-th equipment in the saccharification workshop, B exp represents the expected daily output of each equipment obtained from the subject level: saccharification level. i represents the i-th initial level. The constraint functions of the initial level are formula (2) to formula (6), that is, the constraint functions of the saccharification workshop production.

[0057]

[0058] Figure 2 The subject level shown: The saccharification model is shown in formula (13). sac It is subject level: optimization result of saccharification, B sys is the expected result obtained at the system level. i represents the best result at the initial level. ε is the relaxation factor, which will make the best result better and relax the restrictions within the tolerable range. Subject level: The objective function in the fermentation model is shown in formula (14). In the formula, ∑ t B ht θ h Indicates the daily output of the fermentation device. expsys It refers to the expected value of the fermentation unit output obtained from the system level. Discipline level: The constraint functions in the fermentation model are shown in formulas (7) to (9).

[0059]

[0060]

[0061] Figure 2 The constraint function of the system-level model is shown in formula (15). sys represents the optimal result at the system level, while X disRepresents the optimization results of the two subject levels. j represents the sum of the number of independent variables including the two subject levels. The objective function is divided into the original objective function and the improved objective function, as shown in formula (1) and formula (11) respectively. During the execution of the algorithm, the specific objective function used by the system-level objective function is determined by the current number of iterations. The specific execution steps of the algorithm will be described in detail later.

[0062]

[0063] The original objective function is used as the system-level optimization objective function in the first iteration of the algorithm. In the subsequent iterations, the original objective function or the evaluation function is selected alternately as the system-level optimization objective function. In the optimization process, the non-dominated sorting genetic algorithm is used to obtain the Pareto solution of the multi-objective optimization problem.

[0064] In the specific implementation of the dynamic scheduling algorithm, the non-dominated sorting genetic algorithm (NSGA-II) is the core tool of the optimization process, and its application runs through the entire three-level optimization framework. The NSGA-II algorithm can effectively handle the trade-off relationship in complex production scheduling problems through multi-objective optimization. Specifically, NSGA-II uses sorting and crowding distance evaluation mechanisms in the optimization process to ensure the uniformity and diversity of the solution set in the target space. First, the algorithm performs a fast non-dominated sort on the current population and divides the population into different non-dominated levels. Secondly, the density distribution of each solution is determined by crowding distance calculation, thereby retaining representative Pareto frontier solutions in the target space. Finally, the algorithm uses gene recombination and mutation operations to continuously generate new populations, and selects solutions with higher fitness based on the dynamically adjusted optimization objective function. Compared with other optimization algorithms, the advantage of NSGA-II is that it can efficiently handle conflicting requirements in multi-objective optimization problems, especially for complex scenarios such as beer production that need to balance production efficiency and stability at the same time.

[0065] In dynamic scheduling optimization, the specific implementation process of the non-dominated sorting genetic algorithm (NSGA-II) runs through the entire optimization stage and has the characteristics of dynamic update. The algorithm first generates an initial population through random initialization, where each individual in the population represents a possible scheduling solution. These individuals are encoded as genotype data, usually using real number coding or binary coding to describe key parameters such as scheduling order and machine allocation. In practical applications, in order to ensure the quality of the initial population, domain knowledge and heuristic rules can be combined to generate some high-quality solutions to reduce the excessive randomness of the algorithm in the early iterations.

[0066] After the population is generated, NSGA-II evaluates the population based on the objective function. The multi-objective functions in dynamic scheduling generally include key indicators such as maximizing production efficiency, optimizing resource utilization, and minimizing equipment failure recovery time. NSGA-II divides the solution set into different levels through fast non-dominated sorting, and uses crowding distance sorting to retain evenly distributed solutions in the target space. Subsequently, the algorithm uses selection, crossover, and mutation operations to generate the next generation of population. The selection operation usually uses tournament selection or roulette, while the crossover operation generates new individuals by exchanging some gene fragments, further enhancing the diversity of the solution set. The mutation operation introduces small perturbations, thereby improving the ability of the population to jump out of the local optimum.

[0067] In order to adapt to the characteristics of dynamic scheduling, the algorithm needs to adjust the weight of the optimization objective function in real time in each iteration. For example, when equipment failure occurs, the algorithm will give priority to adjusting the production line scheduling plan to minimize loss time; in the case of delays in raw material supply, the optimization goal may shift to maximizing inventory utilization. In addition, to improve adaptability in dynamic environments, NSGA-II can dynamically change the crossover probability and mutation probability through an adaptive parameter adjustment mechanism to ensure that the algorithm responds quickly when the environment changes.

[0068] There are also many practical challenges in dynamic scheduling, such as the uncertainty of equipment failure frequency, sudden changes in order demand, and the complexity of parallel scheduling of multiple production lines. In this context, the advantages of NSGA-II are reflected in the following aspects: first, its multi-objective optimization capability allows balancing multiple conflicting objectives at the same time, such as the trade-off between production efficiency and quality; second, its Pareto frontier solution set provides a variety of options for scheduling decisions and can adapt to different production scenarios; finally, NSGA-II's non-dominated sorting mechanism enables it to perform well in multi-objective optimization problems, outperforming methods such as particle swarm optimization (PSO) and simulated annealing algorithm (SA), which are prone to local optimality or insufficient solution set diversity when dealing with dynamic changes.

[0069] For the application of dynamic scheduling algorithms, the original objective function is used as the optimization target in the initial stage. The main purpose is to provide a stable initial solution set for the entire production process and ensure the reliability of the scheduling model in the initial optimization stage. In subsequent iterations, the original objective function and the improved objective function are used alternately as system-level optimization targets, aiming to gradually enhance the model's adaptability to emergencies (such as equipment failures). In each iteration, a set of Pareto optimal solutions is generated through NSGA-II. These solutions not only reflect the trade-off between production efficiency and quality, but also provide diversified options for further scheduling decisions. During the execution of the algorithm, the weight of the optimization objective function is dynamically adjusted according to different equipment failure conditions, thereby improving the robustness of the model under complex scheduling problems.

[0070] Next, numerical experiments are carried out using relevant data to verify the effectiveness of the algorithm by solving the production scheduling problem under equipment failure environments encountered in four production scheduling. The four failure conditions are shown in Table 4. All simulations in the present invention are carried out in Matlab R2021.

[0071] Table 4.4 Fault Conditions

[0072] In this paper, we show the results of the optimization algorithm proposed in this invention, and compare and analyze the results obtained by directly solving the original problem objective function model and the improved objective function model using genetic algorithm. For specific simulation results, please refer to Figure 3 . By comparing these results, it can be clearly observed that in solving the beer production scheduling problem, the algorithm proposed in the present invention has a better performance than directly using the genetic algorithm to optimize the original objective function or improve the objective function. Specifically, the method of the present invention shows significant advantages in terms of both solution efficiency and the quality of the final solution, which shows the effectiveness and reliability of the new algorithm in dealing with complex production scheduling challenges.

[0073] The effectiveness of the algorithm is further verified by comparing the simulation results with the results obtained by solving the original problem objective function as the objective function model using the genetic algorithm under normal equipment conditions.

[0074] First, the effectiveness of the algorithm is evaluated by the average value of the optimization result population. The solutions under normal conditions and fault conditions are selected to obtain their average values, and the results are shown in Table 5. As can be seen from Table 5, the scheduling effect of the algorithm proposed in the present invention is the smallest compared with that under normal conditions. However, the average value is sometimes affected by the data distribution and cannot reflect the characteristics of the data well. Therefore, the present invention selects the HV indicator of the multi-objective optimization algorithm to further analyze the effectiveness of the algorithm.

[0075] Table 5. Average results

[0076] The H index, whose full name is Hypervolume index, is a method used to evaluate the performance of multi-objective optimization algorithms. It mainly measures the volume covered by a set of non-dominated solutions in the target space, relative to a fixed reference point (usually a point that is worse than any feasible solution in all target dimensions). If all solutions are close to the ideal point, the HV value is larger. Therefore, the larger the HV, the better the optimization effect of the algorithm. The present invention selects (-1,60000) as the reference point, obtains the HV values ​​of the populations of each solution under different conditions and compares them. Regardless of the fault situation, the HV value of the algorithm proposed in the present invention is closer to the HV value of the solution under normal conditions than other methods. In addition, the present invention also uses the IGD index to verify the algorithm's ability to handle faults.

[0077] The IGD indicator, whose full name is Inverted Generational Distance, is an indicator used to evaluate the performance of a multi-objective optimization algorithm. It is mainly used to evaluate the convergence and diversity of the algorithm, and can also be used to compare the closeness between the frontiers found by different algorithms. The present invention calculates the sum of the distances from each point on the Pareto front to the nearest solution on the Pareto front under normal circumstances, and takes the average value to obtain the value of IGD. If the Pareto front obtained by the algorithm is closer to the solution obtained under normal circumstances, the smaller the value of IGD. Therefore, the algorithm proposed in the present invention can better handle beer production scheduling problems and equipment failures encountered in the production process.

[0078] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A beer production scheduling method under uncertainty, characterized in that: The method comprises: S01: Schedule the saccharification, filtration, boiling, clarification or fermentation process of beer production; S02: In the first iteration, the original objective function is used to optimize the scheduling scheme based on the production process time requirements and equipment capacity constraints; S03: In each subsequent iteration, the original objective function or the evaluation function is selected as the optimization target in turn; S04: Dynamically adjust the optimization objective function according to the dynamics of equipment failure or changes in the production environment; S05: Optimizing Pareto frontier solutions using non-dominated sorting genetic algorithm; S06: Output a set of Pareto optimal solutions.

2. The beer production scheduling method under uncertainty according to claim 1 is characterized in that: The method monitors key process parameters in real time during the dynamic scheduling process, including saccharification efficiency, filtration loss, boiling time, clarification effect or fermentation status.

3. The beer production scheduling method under uncertainty according to claim 1 is characterized in that: The method adjusts the optimization function of the scheduling model according to a preset change rule during each iteration.

4. The beer production scheduling method under uncertainty according to claim 3 is characterized in that: The method selects the original objective function or the evaluation function in rotation based on the principle that the scheduling priority time of the saccharification, filtration, boiling and clarification processes is the shortest.

5. The beer production scheduling method under uncertainty according to claim 4 is characterized in that: The method dynamically adjusts the activation state of the fermentation equipment and the raw material allocation plan according to the capacity of the fermentation tank and the limitation of the fermentation time.

6. The beer production scheduling method under uncertainty according to claim 1, characterized in that: The evaluation function evaluates the impact of equipment failure or raw material shortage on overall production efficiency, and constructs a penalty function to adjust the scheduling plan to reduce costs and losses.

7. The beer production scheduling method under uncertainty according to claim 1 is characterized in that: The non-dominated sorting genetic algorithm maintains the diversity of the population and approaches the optimal solution set by simulating the selection, crossover and mutation operations in the natural evolution process.

8. A beer production scheduling method under uncertainty according to claim 1 or 7, characterized in that: The core of the non-dominated sorting genetic algorithm is the non-dominated sorting mechanism, which can sort the solution set and select solutions that perform well on multiple objectives.

9. The beer production scheduling method under uncertainty according to claim 1, characterized in that: The method stops iterating when a predetermined maximum number of iterations is reached or a preset convergence condition is reached, and outputs a series of Pareto graphical optimal solutions representing different trade-off schemes when the iteration is stopped.

10. A beer production scheduling method under uncertain conditions according to claim 1 or 2, characterized in that: The method establishes a dynamic balance model of production capacity and inventory in different process links.

Images