Cross-domain collaborative workshop dynamic scheduling method based on adaptive evolution of balance indicators

Abstract

The present invention discloses a cross-domain collaborative workshop dynamic scheduling method based on adaptive evolution of balance indicators, which relates to the technical field of workshop dynamic scheduling. The specific steps include the following: discrete indirect coding of the scheduling task; after completing the discrete indirect coding, traversing the coding to convert the sequence, and using an operator based on the exchange sequence to generate a new coding individual; using an adaptive evolution strategy based on the balance indicator to evaluate and constrain the coding individual, and the evaluation result is used as an indicator for parent individual selection and local search selection. The present invention avoids the generation of infeasible solutions and thus improves the solution speed of the algorithm by designing a discrete indirect variable coding form; and the new solution generation operator based on the exchange sequence can generate individuals quickly and efficiently. At the same time, combined with the adaptive evolution strategy based on the balance indicator, the algorithm is applied to scheduling problems in different scenarios with stable results and good solution speed.

Description

Technical Field

[0001] The present invention relates to the technical field of workshop dynamic scheduling, and more particularly to a cross-domain collaborative workshop dynamic scheduling method based on adaptive evolution of balance indicators. Background Art

[0002] Because different manufacturing fields share certain similarities, the complementary and efficient nature of each field in a distributed workshop environment enables effective cross-domain collaboration. However, the dynamic nature of tasks and environments, coupled with the complex nature of cross-domain collaboration scenarios, makes it difficult to achieve dynamic scheduling of distributed workshops with cross-domain collaboration using traditional methods.

[0003] In existing technologies, some approaches consider combining improved traditional algorithms with other algorithms for dynamic scheduling. However, because these algorithms simplify scenarios and ignore the heterogeneity of related problems, they struggle to provide detailed solutions for distributed workshop environments and achieve balanced resource utilization. Cross-domain collaborative optimization problems often involve numerous uncertainties, making these algorithms less robust and unable to meet adaptive requirements.

[0004] Some approaches also consider using reinforcement learning and deep learning models to predict task requirements and schedule small batches of independent or process-based tasks. Existing reinforcement learning algorithms combine deep neural networks and utilize industrial big data, such as changes in user demand and the mobility of IoT devices, for model training. These algorithms have been partially designed for dynamic scheduling in distributed workshops. However, existing reinforcement learning models are only applicable to specific scenarios. When tasks, resources, environments, and scheduling objectives change, the models become inapplicable and require retraining for the new scenarios. Therefore, dynamic scheduling algorithms based on reinforcement learning are difficult to adapt to. Furthermore, reinforcement learning does not require prior knowledge and requires a large amount of data to support model training, making model learning inefficient and time-consuming.

[0005] Therefore, for those skilled in the art, how to solve the shortcomings of the dynamic scheduling algorithm, such as poor adaptability and slow solution speed, is an urgent problem to be solved. Summary of the Invention

[0006] In view of this, the present invention provides a cross-domain collaborative workshop dynamic scheduling method based on adaptive evolution of balance indicators to solve the problems existing in the background technology.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: a cross-domain collaborative workshop dynamic scheduling method based on adaptive evolution of balance indicators, the specific steps of which include the following:

[0008] Discrete indirect encoding of the scheduling tasks;

[0009] After completing the discrete indirect coding, the coding is traversed to convert the sequence, and a new coding individual is generated using an operator based on the exchange sequence;

[0010] An adaptive evolutionary strategy based on a balance index is used to evaluate and constrain the coding individuals, and the evaluation results are used as indicators for parent individual selection and local search selection.

[0011] Optionally, the specific process of the discrete indirect coding is: when there are n computing tasks and m manufacturing tasks to be scheduled, the coding bits are divided into two parts: manufacturing tasks and computing tasks. The lengths of the two segments are n and m, respectively, and the values are positive integers between 0 and (n-1) and 0 and (m-1), respectively, and cover all positive integers within the value range.

[0012] Optionally, the operation process of the exchange sequence is: select Pop1 and Pop2 as parent individuals, as the implementation object and reference object respectively, and set the exchange probability to 0.5; traverse the value on each coding position in Pop1, find the location in Pop2, obtain the exchange pair with a probability of 0.5, and finally obtain the exchange pair sequence from Pop2 to Pop1, operate on the individuals according to the exchange pair sequence, and obtain a new coded individual.

[0013] Optionally, the adaptive evolution strategy based on the balance indicator is:

[0014] Calculate the diversity component and convergence component of each individual in the population separately;

[0015] Weighting the diversity component and the convergence component to obtain a balance index;

[0016] Sort the individuals in the population in ascending order of the balance index value, select the first individual with the smallest population balance index, and then select the second individual through the roulette strategy.

[0017] Optionally, the calculation formula of the convergence component is:

[0018] f C (x) = d(r, f(x));

[0019] Among them, f C (x) is the convergence component reflecting the distance between the solution and the POF, r represents the objective function value of the closest point on the POF, d(r,f(x)) represents the distance between r and f(x), and f(x) is the objective function value vector composed of all objective functions under the solution x.

[0020] Optionally, the calculation formula of the diversity component is:

[0021]

[0022] in, is the metric solution x in the target space relative to f m The diversity component of the axis position, v m =(0,...,1 m ,...,0) is an M-dimensional vector, θ(v m ,f(x)-z * ) means v m and f(x)-z * The angle, z * As a reference point.

[0023] Optionally, the balance indicator is expressed as:

[0024] B_fit(x)=η1f C (x)+η2f D (x);

[0025] Among them, f C (x) is the convergence component reflecting the distance between the solution and POF, is the metric solution x in the target space relative to f m The diversity component of the axis position, η1 represents the convergence component f C The weight of (x), η2 represents the diversity component f D The weight of (x).

[0026] On the other hand, a workshop dynamic scheduling system based on adaptive evolution of balance indicators is provided, comprising a variable encoding module, an operator calculation module, and an evolutionary strategy module; wherein,

[0027] The variable encoding module is used to perform discrete indirect encoding on the scheduled tasks;

[0028] The operator calculation module is used to traverse the code to convert the sequence after completing the discrete indirect coding, and generate new code individuals using an operator based on the exchange sequence;

[0029] The evolutionary strategy module is used to evaluate and constrain the coding individuals using an adaptive evolutionary strategy based on a balance indicator, and the evaluation results are used as indicators for parent individual selection and local search selection.

[0030] As can be seen from the above technical solution, compared with the prior art, the present invention discloses a cross-domain collaborative workshop dynamic scheduling method based on adaptive evolution of balance indicators, which has the following beneficial technical effects:

[0031] (1) By designing a discrete indirect variable encoding form, the generation of infeasible solutions is avoided, thereby improving the algorithm's solution speed;

[0032] (2) The new solution generation operator based on the exchange order can generate individuals quickly and efficiently. At the same time, combined with the adaptive evolutionary strategy based on the balance indicator, the algorithm can be applied to scheduling problems in different scenarios with stable results and good solution speed.

[0033] (3) The adaptive search strategy based on the balance index proposed in the present invention can accelerate the convergence of the algorithm and improve the performance of the algorithm, so that the algorithm has a good solution effect for scheduling problems in different scenarios, and can save 30%-50% of the running time in different scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0035] Figure 1 is a flow chart of the method of the present invention;

[0036] Figure 2 An example diagram of discrete integer encoding for sorting of the present invention;

[0037] Figure 3 Schematic diagram of the exchange sequence algorithm of the present invention using Pop1 as the implementation object;

[0038] Figure 4 This is a flow chart of the adaptive strategy based on the balance index of the present invention;

[0039] Figure 5 It is a framework diagram of the dynamic scheduling algorithm of the present invention;

[0040] Figure 6 This is a system structure diagram of the present invention. DETAILED DESCRIPTION

[0041] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0042] Embodiment 1 of the present invention discloses a cross-domain collaborative workshop dynamic scheduling method based on adaptive evolution of balance indicators, such as Figure 1 As shown, the specific steps include:

[0043] S1, discrete indirect encoding of the scheduling task;

[0044] S2. After completing the discrete indirect coding, the coding is traversed to transform the sequence, and a new coding individual is generated using an operator based on the exchange sequence;

[0045] S3. An adaptive evolutionary strategy based on a balance indicator is used to evaluate and constrain the coding individuals, and the evaluation results are used as indicators for parent individual selection and local search selection.

[0046] The adaptive evolutionary strategy based on a balance metric aims to ensure that the scheduling algorithm performs well in solving scheduling problems in different scenarios. This means improving the algorithm's ability to adapt to the dynamic changes of different scenarios, rather than focusing solely on solving a single scenario. S2 describes a method for generating new individuals, similar to the crossover and mutation operations in genetic algorithms. It aims to expand the algorithm's search space to enhance the diversity of the population and improve the algorithm's search performance. Both strategies operate on the solution set, but with different emphases. S3 evaluates individuals using metrics to obtain solutions with improved diversity and convergence, enhancing the algorithm's ability to dynamically adapt to different scenarios. S2 expands the algorithm's search range by generating new solutions.

[0047] In order to avoid infeasible solutions during the solution process and improve the search efficiency of the multi-objective evolutionary algorithm, the present invention adopts a discrete indirect coding method, and the coding bits represent the priority when allocating manufacturing resources or computing resources to tasks. Specifically, when there are n computing tasks and m manufacturing tasks to be scheduled, the coding bits are divided into two parts: manufacturing tasks and computing tasks. The lengths of the two parts are n and m, respectively, and the values are positive integers between 0 and (n-1) and 0 and (m-1), respectively, and cover all positive integers within the value range. Figure 2 Examples of coding for five computational and manufacturing tasks.

[0048] After the discrete integer encoding of the scheduling task is completed, a new encoding individual is generated using an operator based on the exchange order. The pseudo code of the operator in this embodiment is shown in Algorithm 1.

[0049]

[0050] First, we need to select two parent individuals, one as the implementation object and the other as the reference object, to obtain a sequence of conversion processes from the reference object to the implementation object. Here, Pop1 is the implementation object and Pop2 is the reference object. We traverse the code bits in Pop1 and Pop2, and after searching and comparing, we obtain a sequence of swap pairs that convert from Pop2 to Pop1. More specifically, for the value of Pop2[i], we search for the same code bit in Pop1, record the number j of that code bit in Pop1, and then store the swap pair (i, j) in the swap pair sequence V. After completing the traversal of the two selected individuals, we can obtain a swap pair sequence of indeterminate length. In other words, if we apply the obtained swap pair sequence to Pop2, we can obtain Pop1.

[0051] The operation process of the above exchange sequence method is as follows: Figure 3 As shown, corresponding to the pseudo code, Figure 3 In the solution, Pop1 and Pop2 are still selected as parent individuals, serving as the implementation object and reference object respectively, with the exchange probability set to 0.5. The numerical value on each coding bit in Pop1 is traversed, and its position is found in Pop2. The exchange pairs are obtained with a probability of 0.5, such as (2.4), (2.7), etc., and finally the sequence of exchange pairs from Pop2 to Pop1 is obtained. A new individual can be obtained by operating on a certain individual according to the exchange pair sequence; or the exchange pair sequence can be obtained by limiting the length of the exchange sequence without setting the exchange probability. Since the manufacturing task and the computing task in the coding bit are independent, the manufacturing task coding bit and the computing task coding bit are operated separately in the process of generating a new solution.

[0052] Different from static scheduling tasks in a single scenario, this invention targets cross-domain collaborative distributed workshop dynamic scheduling. After completing the coding design and determining the scheduling operator, the consideration is how to maintain good solution capabilities in different scenarios. Accordingly, this invention proposes an adaptive evolutionary strategy based on a balance indicator to adapt to multi-scenario environments.

[0053] In multi-objective evolutionary algorithms, the balance between convergence and diversity is crucial to algorithm performance. To improve algorithm speed, an evaluation mechanism based on convergence and diversity indicators is designed. The evaluation results are used as indicators for parent individual selection and local search selection. That is, the smaller the evaluation value, the greater the probability of the individual being selected. The specific calculation method is as follows:

[0054] Assume that the optimization objective of a dynamic scheduling problem is as shown in formula (1), with a total of M optimization objectives. Define the convergence index and diversity index to solve as shown in formulas (2) and (3), f C (x) is the convergence component reflecting the distance between the solution and POF (Pareto Optimal Frontier), is the metric solution x in the target space relative to fm As shown in formula (4), the average value of the diversity components corresponding to each target is used as the diversity index value of the solution x.

[0055] min f(x)=min(f1(x),f2(x),...,f M (x)) (1)

[0056] f C (x)=d(r,f(x)) (2)

[0057]

[0058]

[0059] Among them, r represents the distance on the POF x The objective function value of the nearest point, d(r,f(x)) represents the distance between them, v m =(0,...,1 m ,...,0) is an M-dimensional vector, θ(v m ,f(x)-z * ) means v m and f(x)-z * The angle, z * As a reference point.

[0060] The balance index between convergence and diversity is defined as shown in formula (5):

[0061] B_fit(x)=η1f C (x)+η2f D (x) (5)

[0062] η1 represents the convergence component f C The weight of (x), η2 represents the diversity component f D The weight of (x).

[0063] The algorithm is as follows: First, the diversity component and convergence component of each individual in the population are calculated using the above formula, and the weighted balance index designed by the present invention is obtained. Then, the individuals in the population are sorted in ascending order according to the balance index value. Finally, individual 1 with the smallest balance index of the current population is selected, and individual 2 is selected through the roulette strategy. The overall process is as follows: Figure 4 shown.

[0064] The specific pseudo code of the adaptive evolution strategy based on the balance indicator is as follows:

[0065]

[0066] Combining the above technical solutions, the overall implementation process of the present invention is as follows: Figure 5 As shown. After the algorithm is initialized, the current distributed workshop environment is dynamically detected. When changes occur, dynamic adjustments are made based on the dynamic corresponding strategy of the uncertain dynamic environment. When the environment has not changed, a combination of global search strategy and local search strategy is adopted. The global search strategy adopts an adaptive evolutionary strategy based on the balance index. After the search is completed, the objective function value is calculated and the population and the Pareto frontier solution set are updated. Then, the iteration is repeated and the output scheduling plan is exited after the termination condition is reached. It should be noted that the focus of the present invention lies in the algorithm initialization coding method and the adaptive evolutionary strategy based on the balance index.

[0067] Embodiment 2 of the present invention provides a workshop dynamic scheduling system based on adaptive evolution of balance indicators, such as Figure 6 As shown, it includes variable encoding module, operator calculation module, and evolution strategy module; among them,

[0068] Variable encoding module, used for discrete indirect encoding of scheduling tasks;

[0069] The operator calculation module is used to traverse the code after completing the discrete indirect coding to transform the sequence and generate new code individuals using operators based on the exchange sequence;

[0070] The evolution strategy module is used to evaluate and constrain the coding individuals using an adaptive evolution strategy based on a balance indicator. The evaluation results are used as indicators for parent individual selection and local search selection.

[0071] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the common and similar parts between the various embodiments. For the devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method description.

[0072] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A cross-domain collaborative workshop dynamic scheduling method based on adaptive evolution of balance indicators, characterized by: The specific steps include the following: Discrete indirect encoding of the scheduling tasks; After completing the discrete indirect coding, the coding is traversed using an operator based on the exchange sequence to transform the sequence and generate a new coding individual; An adaptive evolutionary strategy based on a balance index is used to evaluate and constrain the coding individuals, and the evaluation results are used as indicators for parent individual selection and local search selection; The adaptive evolution strategy based on the balance index is: Calculate the diversity component and convergence component of each individual in the population separately; Weighting the diversity component and the convergence component to obtain a balance index; Sort the individuals in the population in ascending order of the balance index value, select the first individual with the smallest population balance index, and select the second individual through the roulette strategy; The calculation formula of the convergence component is: f C (x)=d(r,f(x)); Among them, f C (x) is the convergence component reflecting the distance between the solution and the POF, r represents the objective function value of the point closest to the POF, d(r,f(x)) represents the distance between r and f(x), and f(x) is the objective function value vector composed of all objective functions under the solution x; The calculation formula of the diversity component is: in, is the metric solution x in the target space relative to f m The diversity component of the axis position, v m =(0,...,1 m ,...,0) is an M-dimensional vector, θ(v m ,f(x)-z * ) means v m and f(x)-z * The angle, z * is the reference point; The expression of the balance index is: B_fit(x)=η1f C (x)+η2f D (x); Among them, f C (x) is the convergence component reflecting the distance between the solution and POF, is the metric solution x in the target space relative to f m The diversity component of the axis position, η1 represents the convergence component f C The weight of (x), η2 represents the diversity component f D The weight of (x).

2. The cross-domain collaborative workshop dynamic scheduling method based on adaptive evolution of balance indicators according to claim 1 is characterized in that: The specific process of the discrete indirect coding is as follows: when there are n computing tasks and m manufacturing tasks to be scheduled, the coding bits are divided into two parts: manufacturing tasks and computing tasks. The lengths of the two segments are n and m, respectively, and the values are positive integers between 0 and (n-1) and 0 and (m-1), respectively, and cover all positive integers within the value range.

3. The cross-domain collaborative workshop dynamic scheduling method based on adaptive evolution of balance indicators according to claim 1 is characterized in that: The operation process of the exchange sequence is as follows: select Pop1 and Pop2 as parent individuals, as the implementation object and reference object respectively, and set the exchange probability to 0.5; traverse the value on each coding position in Pop1, find the location in Pop2, obtain the exchange pair with a probability of 0.5, and finally obtain the exchange pair sequence from Pop2 to Pop1. Operate on the individuals according to the exchange pair sequence to obtain new coded individuals.

Images