Task decomposition method for home appliance product crowdsourcing design

Abstract

The application discloses a kind of task decomposition methods of household electrical appliances crowdsourcing design, it is characterized in that, include the following steps: S1, according to product BOM table reference library, based on product structure tree, crowdsourcing design task is decomposed, form a group of meta task set, and determine granularity reference value;S2, the correlation between meta task is expressed as design structure matrix, and the edge between task is measured with the method of edge weight, using K means algorithm to the candidate task set is clustered;S3, adopt the method of module degree to determine clustering cluster number K, obtain design task decomposition set;S4, the granularity of design task decomposition set in step S3 is calculated, if granularity is within granularity reference value, then determine as final task decomposition set;Otherwise, with the determination clustering cluster number K of lower level module degree, obtain design task decomposition set;Repeat step S3 and S4.The application has the advantages that the cohesion in the decomposed subtask is higher, the coupling between subtask is lower and the like.

Description

Technical Field

[0001] This invention relates to the field of crowdsourced design technology, and in particular to a task decomposition method for crowdsourced design of home appliances. Background Technology

[0002] With the rapid development of internet technology, crowdsourcing design has become a key method for promoting product innovation and development. It not only fully utilizes social resources and optimizes their allocation, but also gathers global wisdom and creativity, breaking down geographical and industry barriers to innovation and allowing more people to participate in problem-solving and innovation. However, in the field of product innovation and development, crowdsourcing design is currently only applied to the innovation of simple products. Because the development of complex products involves multiple parties and numerous stages, problems arise in management models and personnel interaction, hindering the application of crowdsourcing design in complex products. Home appliances, as complex products, are particularly vulnerable; traditional manufacturing methods struggle to meet the growing demand for personalized home appliances, thus placing new demands on the innovation capabilities of home appliance products.

[0003] By leveraging internet platforms, crowdsourcing distributes tasks to a large number of distributed online users, fully tapping into social innovation resources and improving the innovation level and market competitiveness of home appliances. However, home appliances are typically highly complex, involving multiple components and technologies, posing challenges for crowdsourcing platforms in managing these tasks. To overcome these challenges, crowdsourcing platforms need to decompose complex tasks related to home appliances into relatively simple, independent sub-tasks, making it easier to assign design tasks to users on the platform. However, task decomposition in crowdsourcing is characterized by multi-domain intersections and multi-participant collaboration; therefore, the decomposed sub-tasks should possess the characteristics of "high cohesion and low coupling," and existing methods are not yet suitable for task decomposition in crowdsourcing. Summary of the Invention

[0004] In view of the shortcomings of the prior art, the technical problem to be solved by the present invention is: how to provide a task decomposition method for crowdsourcing design of home appliances that can achieve high cohesion among the decomposed subtasks and low coupling between the subtasks.

[0005] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0006] A task decomposition method for crowdsourced design of home appliances, characterized by the following steps:

[0007] S1. Based on the product BOM reference library, decompose the crowdsourcing design task according to the product structure tree to form a set of meta-tasks and determine the granularity reference value.

[0008] S2. Represent the correlation between meta-tasks as a design structure matrix, measure the edges between tasks using edge weights, and use the K-means algorithm to cluster the candidate task set.

[0009] S3. Determine the number of clusters K using the modularity method to obtain the design task decomposition set;

[0010] S4. Calculate the granularity of the task decomposition set designed in step S3. If the granularity is within the granularity reference value, it is determined as the final task decomposition set; otherwise, determine the number of clusters K with the next lower level of modularity to obtain the design task decomposition set; repeat steps S3 and S4.

[0011] Furthermore, in step S1, the relationships between tasks in the meta-task set are represented as follows:

[0012] G = (T, R, S)

[0013] Where, T = {t i |i=1,2,···n} is the vertex set of the directed graph of the product; represents the meta-task that makes up the product; R={(t i ,t j )|t i ∈T,t j Let ∈T, i, j = 1, 2, ..., n be the edge set in the directed graph of the products, t i Indicates the starting node, t j Indicates the terminating node; indicates the connection relationship between product meta-tasks; S = {s ij |s ij ,i,j∈N} represents the edge set (t) in the directed graph of the product. i ,t j The weight set corresponding to ).

[0014] Furthermore, in step S2, the correlation between meta-tasks is defined using a [0,1] average distribution and represented as a fuzzy correlation value, as shown in the table below:

[0015]

[0016] Based on the above correlation metrics, an evaluation team composed of experts and technicians determined the functional and structural correlations among the various tasks, and constructed functional design matrices and structural design matrices:

[0017]

[0018] In the formula, RM x R represents the submatrix between tasks i and j. Sij This represents the relevance value between tasks i and j.

[0019] Furthermore, the correlation between tasks is measured using an edge weight metric, specifically:

[0020] S ij =S ji =(s i s j ) α (α>0)

[0021] In the formula: s i and s j Let represent the degree of task i and task j respectively, and α be the edge weight coefficient, which describes the relationship between edge weight and node degree.

[0022] Furthermore, in step S2, Euclidean distance is used to calculate the design task T. i and design task T j Distance D between ij Specifically:

[0023]

[0024] Furthermore, in step S3, the formula for calculating the modularity is:

[0025]

[0026] In the formula L W To sum the weights of all edges in the weighted task network, w ij The value in the i-th row and j-th column of the adjacency matrix W represents the task association degree between meta-tasks i and j; n represents the number of meta-task nodes; k i The sum of the weights of all connected edges of the meta-task node i; δ(v i ,v j The value is 1 when two nodes belong to the same community and 0 when they belong to different communities.

[0027] Furthermore, the task granularity is measured by the average of all subtask granularity metrics:

[0028]

[0029] in, For task t i Particle size, A i Let c be the task coupling coefficient. i c is the task cohesion coefficient. i =α i ·β i α i β is the task correlation coefficient. i The task reuse factor is as follows:

[0030]

[0031]

[0032] In the formula, |t| represents the number of meta-tasks, and p o q o For output meta-task; e i d i This is the input meta-task.

[0033] In summary, the present invention has the advantages of enabling high cohesion among the decomposed subtasks and low coupling between subtasks. Attached Figure Description

[0034] Figure 1 Flowchart for crowdsourcing the design of home appliance products.

[0035] Figure 2 This is a product structure tree diagram.

[0036] Figure 3 This is a directed graph representing the product structure.

[0037] Figure 4 Flowchart of crowdsourced design task decomposition algorithm.

[0038] Figure 5 This is a product structure tree diagram for 3D printers.

[0039] Figure 6 This is a directed graph representing the structure of a 3D printer product.

[0040] Figure 7 This is a graph showing the changes in the modularity of task aggregation.

[0041] Figure 8 This is a meta-task aggregation graph.

[0042] Figure 9 This is a graph showing the aggregated results of the tasks.

[0043] Figure 10 This is a schematic diagram showing the density change.

[0044] Figure 11 This is a schematic diagram illustrating the change in entropy.

[0045] Figure 12 This is a diagram illustrating the changes in topicality. Detailed Implementation

[0046] The present invention will be further described in detail below with reference to the embodiments.

[0047] 1. Crowdsourced Design Workflow for Home Appliances

[0048] Product design crowdsourcing is typically divided into two forms: public participation and direct public involvement in product design. Direct public participation is the most common form on current domestic crowdsourcing platforms. This embodiment establishes a workflow for a home appliance product crowdsourcing design task based on this form, as follows: Figure 1 As shown, in crowdsourced design tasks for home appliances, the crowdsourcing platform breaks down the design task into multiple design requirements based on the client's needs and task specifications, and publishes these requirements on the platform for contractors to choose from. Simultaneously, the platform categorizes contractors' skills and abilities, creating design resources to provide more comprehensive design services to the client. During this process, contractors can choose tasks that interest them and that they possess the relevant skills to submit their design proposals. Collaboration among contractors is also crucial during the completion of the design task, reflecting the characteristics of collaborative crowdsourcing. Crowdsourced design tasks may be broken down into multiple smaller tasks, completed collaboratively by multiple contractors. Contractors can exchange ideas and learn from each other, working together to improve efficiency and quality. After the design is completed, the crowdsourcing platform integrates the design proposals and provides feedback to the client for selection and adoption. Upon adoption, the platform rewards the contractors to incentivize them to provide better design services. This model can help the client obtain high-quality design services quickly, while also providing the contractor with more design opportunities and room for development, and promoting communication and learning among contractors.

[0049] 2. Task decomposition principles for crowdsourced design of home appliances

[0050] Not all crowdsourced design tasks for home appliances are complex. Complex crowdsourced tasks involve a large amount of content and complexity, which cannot be completed independently by a single entity and requires collaboration among multiple entities. Complex crowdsourced design tasks for home appliances typically have the following characteristics:

[0051] (1) Multi-level refinement: The crowdsourcing design task for home appliances should be refined in multiple levels, from the whole to the part, and from large to small.

[0052] (2) Multi-field crossover: Crowdsourced design tasks for home appliances need to involve multiple fields, such as mechanical design, electronic design, industrial design, user research, etc., involving knowledge and skills from different fields.

[0053] (3) Consideration from multiple dimensions: Crowdsourced design of home appliances requires consideration of issues from multiple dimensions, such as product performance, cost-effectiveness, user needs, market competition, etc.

[0054] (4) Multi-party collaboration: The crowdsourcing design of home appliances requires the collaboration of multiple parties, including designers, engineers, market researchers, user representatives, and other roles. The task needs to be reasonably allocated and coordinated to achieve high quality and high efficiency.

[0055] Based on the characteristics of the crowdsourced design task for home appliances, the following principles are proposed:

[0056] (1) Hierarchical decomposition principle: Divide the design tasks of home appliances into different levels, including product functions, structure, appearance, etc. In this way, the structural integrity and feasibility of each sub-task can be guaranteed, so as to better meet the needs of the client.

[0057] (2) Matching principle: In order to ensure the efficiency and quality of the completion of the sub-tasks of home appliance product design, the task needs to be broken down into multiple sub-tasks and published on the crowdsourcing platform so as to match with the corresponding contractors.

[0058] (3) High cohesion principle: In the process of decomposing the design tasks of home appliances, it is necessary to consider the correlation and relevance between these sub-tasks. This can promote communication and collaboration among the contractors, and also facilitate the parallel execution of sub-tasks and the efficient completion of the overall task.

[0059] (4) Low coupling principle: In the process of decomposing the design tasks of home appliances, it is necessary to minimize the logistics and information interaction between sub-tasks in order to reduce mutual interference and improve the independence of tasks and overall efficiency.

[0060] (5) Scale control principle: Control the number and size of sub-tasks in the design of home appliances to ensure that each sub-task is of appropriate size. This ensures that the overall task execution efficiency will not be affected by individual tasks that are too large or too difficult, and at the same time, it can also avoid the increase in the number of tasks due to the task granularity being too small, which would affect the task allocation efficiency.

[0061] 3. Product Crowdsourcing Design Task Decomposition Method

[0062] Crowdsourcing, as an internet-based collaborative model, has been widely applied in product design. Through crowdsourcing, companies can access more design resources and improve the efficiency and quality of product design. Task decomposition is a crucial step in crowdsourced design. For complex product design tasks in home appliances, it is necessary to break them down into sub-tasks for easier allocation and management. This embodiment analyzes the characteristics of home appliance product design tasks and proposes using a product structure tree to decompose tasks into meta-tasks. Then, a fuzzy method is used to measure the relevance of sub-tasks and to measure the edge weights between sub-tasks to determine the degree of influence between them. Next, the K-means clustering algorithm is used to cluster the sub-tasks, and the modularity method is used to determine the value of K. Finally, the task granularity is analyzed and controlled to obtain a reasonable task decomposition result. To evaluate the aggregation results of the task decomposition, this embodiment introduces indicators such as density, entropy, and topic difference coefficient for measurement and evaluation.

[0063] 3.1 Preliminary decomposition based on the product structure tree

[0064] In the crowdsourced design task decomposition of home appliances, the Work Structure Tree (WBS) is a crucial tool. It helps companies break down the overall product into several subsystems and components, and further into smaller parts and components, facilitating better task allocation and monitoring. By initially decomposing the work structure tree of a home appliance, the product can be hierarchically divided according to its function and composition, resulting in more specific and detailed design requirements and tasks, providing the contractor with clearer task objectives and scope. Through the work structure tree, the home appliance task is initially divided into multiple meta-task sets. The process of constructing the work structure tree for a crowdsourced design task is as follows: Figure 2 As shown.

[0065] Based on the establishment of the product structure tree, we obtain the following: Figure 3 The product structure is shown in a directed graph. The relationships between tasks are represented by an ordered triple G = (T, R, S), where T = {t...} i |i=1,2,···n} is the vertex set of the directed graph of the product; it represents the meta-tasks that make up the product. R={(t i ,t j )|t i ∈T,t j Let ∈T, i, j = 1, 2, ..., n be the edge set in the directed graph of the products, t i Indicates the starting node, t j Indicates the terminating node; indicates the connection relationship between product meta-tasks. S = {s} ij |s ij ,i,j∈N} represents the edge set (t) in the directed graph of the product.i ,t j The weight set corresponding to ).

[0066] 3.2 Task Aggregation Based on K-means Clustering Algorithm

[0067] Task Relevance Measurement: After the crowdsourcing design task for home appliances is decomposed, a set of meta-tasks is formed. The degree of interaction between these meta-tasks often varies, so it is necessary to measure their task relevance to more accurately express the degree of information dependence between them. This embodiment uses fuzzy theory to measure the relevance between meta-tasks. To better describe and fuzzily quantify the relevance between each meta-task, this embodiment uses a [0,1] average distribution to define the relevance between meta-tasks and represents it as a fuzzy correlation value, as shown in Table 1. This allows for a more accurate description and analysis of the crowdsourcing design task for home appliances.

[0068] Table 1. Ambiguity Correlation Values

[0069]

[0070] Based on the correlation measurement among the aforementioned meta-tasks, an evaluation team composed of relevant experts and technicians determines the functional and structural correlations among the meta-tasks, and then constructs the functional design matrix and structural design matrix, as shown in the following formula:

[0071]

[0072] In the above formula, RM x R represents the submatrix between tasks i and j. Sij This represents the relevance value between tasks i and j.

[0073] Edge weight metric: In product crowdsourcing design tasks, the relevance between meta-tasks can be assessed by the influence of the connecting edges between them. By analyzing the edge weights between meta-tasks, the degree of influence between them can be evaluated. The magnitude of the edge weight represents the degree of influence between meta-tasks; a larger edge weight indicates a stronger influence. In product crowdsourcing design tasks, the following edge weight metric formula can be used to measure the relevance between tasks:

[0074] S ij =S ji =(s i s j ) α (α>0) (2)

[0075] In the formula: s i and s jLet represent the degree of task i and task j respectively, and α be the edge weight coefficient, which describes the relationship between edge weight and node degree.

[0076] Task aggregation: Connections between tasks are established by calculating the correlations between sub-tasks, and a clustering algorithm is used to aggregate the task set. Specifically, the correlations between sub-tasks are represented as a design structure matrix, and the edges between tasks are measured using edge weights. Then, the K-means algorithm is used to cluster the candidate task set, and finally, the modularity method is used to determine the number of clusters K, thus obtaining the final design task decomposition set. When calculating the distances between tasks, Euclidean distance is used to calculate task T. i And Task T j Distance D between ij The calculation formula is equation (3). To determine the number of clusters K in the clustering algorithm, this embodiment uses the modularity proposed by Newman as the metric. The maximum value of the modularity is calculated using the following equation (4), and the number of clusters K in the clustering algorithm is determined. Where L W To sum the weights of all edges in the weighted task network, w ij The value in the i-th row and j-th column of the adjacency matrix W represents the task association degree between meta-tasks i and j; n represents the number of meta-task nodes; k i The sum of the weights of all connected edges of the meta-task node i; δ(v i ,v j The value is 1 when two nodes belong to the same community and 0 when they belong to different communities.

[0077]

[0078]

[0079] Task Granularity Analysis and Control: Task granularity analysis and control is a crucial step in the task decomposition process for crowdsourced design of home appliances. During task decomposition, it's essential to analyze and control the task granularity appropriately to avoid either overly fine-grained tasks (leading to too many tasks and difficulty in coordination) or overly coarse-grained tasks (resulting in unclear tasks and difficulty in management). Therefore, the rationality analysis and control of task granularity is an indispensable part of the task decomposition process for crowdsourced design of home appliances. In task granularity analysis, the appropriate task decomposition granularity needs to be determined based on the complexity and requirements of the specific task. For simple, clear, and well-structured tasks, a more refined task decomposition can be performed, breaking down the task granularity to a finer level for better management and coordination.

[0080] (1) Task correlation coefficient

[0081]

[0082] Where |t| is the number of meta-tasks, p o q o For output meta-task; e i d i This is the input meta-task.

[0083] (2) Task reuse coefficient

[0084]

[0085] (3) Task cohesion coefficient

[0086] The task cohesion coefficient is a measure of the degree of interrelation between meta-tasks within a subtask, reflecting whether the internal structure of a task is tight and independent. Specifically, when the meta-tasks within a subtask are closely connected and interact frequently, the task has high cohesion. Conversely, when the meta-tasks are less connected and more independent, the task has low cohesion. The task cohesion coefficient can be calculated using formula (7).

[0087] c i =α i ·β i (7)

[0088] The granularity of a task is determined by both the number of tasks and the task granularity coefficient. As the number of tasks increases, the granularity becomes smaller. Conversely, a larger task granularity coefficient indicates looser internal connections within the task, leading to a larger granularity. The task cohesion coefficient c is used as an example. i Coupling coefficient A of the task i The ratio τ is used to describe task t i The particle size is shown in the following formula.

[0089]

[0090] The overall task granularity of a crowdsourced design task for home appliances can be measured by calculating the average of all subtask granularity metrics. That is:

[0091]

[0092] τ N The smaller the value, the more refined the task granularity. If the subtasks decomposed from a crowdsourced design task have high cohesion, the correlation between the meta-tasks will be strong, the coupling between the subtasks will be low, and the relative independence between the subtasks will be strong.

[0093] Crowdsourced Design Task Decomposition Process: As shown above, task decomposition breaks down complex tasks into relatively independent, appropriately granular subtasks, making the implementation of these subtasks easier and enabling efficient collaboration across the entire task. In crowdsourced design of home appliances, task decomposition is a crucial step. It requires decomposing tasks based on product structure trees, functional analysis, and other methods. Furthermore, it involves controlling the task granularity using indicators such as task granularity coefficients and task cohesion coefficients to ultimately obtain a reasonable set of subtasks, enabling efficient execution of the design task. The basic steps of the task decomposition method for crowdsourced design of home appliances are as follows: Figure 4 As shown.

[0094] 4. Case Analysis

[0095] 3D printers can print objects of various shapes and sizes, enabling personalized customization of household goods, decorations, models, and other items to meet people's needs. Therefore, this embodiment takes the crowdsourced design task of a home 3D printer as the research object and uses the task decomposition model proposed in this embodiment for case verification. In the 3D printer design phase, using the BOM (Bill of Materials) provided by the task provider and a reference library of similar historical products, the WBS (Work Breakdown Structure) method is used to decompose the system's tasks and determine relevant design parameters, ultimately obtaining the functional decomposition diagram of the 3D printer, as shown below. Figure 5 As shown. And the directed graph of the product's structure is obtained, as shown. Figure 6 As shown.

[0096] By analyzing the directed graph of the product structure, the product can be decomposed into multiple meta-tasks, and a structural design matrix between meta-tasks can be further constructed. As shown in Table 2, each element in the matrix represents the relationship between two meta-tasks. Based on the meta-task relevance matrix, the K-means algorithm is used to cluster the meta-task set. First, the distance between meta-tasks needs to be calculated to obtain the meta-task distance matrix. Here, Euclidean distance is used to calculate the distance between meta-tasks, and the resulting meta-task distance matrix is ​​shown in Table 4 below.

[0097] Table 2 Metatask Structure Design Matrix

[0098]

[0099] Table 3 Metatask Relevance Matrix

[0100]

[0101] Table 4. Distance Matrix Between Metatasks

[0102]

[0103] Following the steps of the crowdsourced design task decomposition method, the modularity M is first calculated for the original directed graph. Then, each meta-task is treated as an individual, and the modularity value of the clustering under different values ​​of K is calculated. The K with the largest modularity value is taken as the optimal clustering result. The above clustering algorithm is implemented using MATLAB programming, and the changes in modularity are recorded during the clustering process, finally constructing a modularity change graph. Figure 8 As shown in the figure, the modularity reaches its maximum value of 0.512 when K is 4. This is achieved through the meta-task clustering tree. Figure 7 This shows the aggregation of various meta-tasks.

[0104] Based on the meta-task aggregation tree, we can see that when K equals 4, the meta-tasks are successfully aggregated into four sub-tasks. The partitioning of each sub-task can be seen through the task-weighted undirected graph. The specific partitioning results are as follows: Figure 9 As shown.

[0105] By analyzing the results of task aggregation, the cohesion coefficient between metatasks within each subtask and the coupling coefficient between subtasks are calculated, thereby determining the granularity metric for each subtask, as shown in Table 5.

[0106] Table 5 Task Granularity Measurement Table

[0107]

[0108] To measure whether the task granularity meets the requirements, this embodiment uses the decomposition granularity of tasks of similar size in the reference product resource library as the task granularity measurement threshold, and calculates the granularity measurement value of each subtask. By averaging the granularity measurement values ​​of each subtask, the overall granularity measurement value of the task is obtained. Finally, this granularity measurement value is compared with the set threshold.

[0109]

[0110] The task decomposition and aggregation method proposed in this embodiment can effectively decompose complex crowdsourced design tasks for home appliances into a set of reasonable sub-tasks. The effectiveness of the method is demonstrated through task granularity metrics and threshold comparisons. When analyzing task correlation, the low correlation at the split points further proves the rationality of the decomposition and aggregation results. Finally, the method proposed in this embodiment successfully yielded an effective decomposition and aggregation result for the crowdsourced design task of home appliances.

[0111] This embodiment employs three metrics to evaluate the final clustering results: density, entropy, and topic dissimilarity coefficient (TDiff). Density refers to the ratio of edges within a community to the total number of edges in the community, measuring the density of edges within different communities. Entropy, a thermodynamic concept describing the degree of disorder in matter, is used in this embodiment to measure the degree of disorder within a community; a lower entropy value indicates a more ordered community. The topic dissimilarity coefficient is an evaluation metric proposed to measure the themes of different communities; a higher coefficient indicates less similarity in themes between communities. By comprehensively evaluating these three metrics, the final clustering results can be objectively assessed and compared to determine the optimal clustering scheme.

[0112] (1) Density

[0113]

[0114] Where |E| represents the number of edges in the network, v i Represents the task, represents C i The value represents the number of communities, and k represents the number of communities. The higher the "density" value, the better the community detection algorithm.

[0115] (2) Entropy

[0116]

[0117] Where |V| represents the total number of nodes, p ij c represents a community with attribute values. j The proportion of nodes in the data.

[0118] (3) Subject Difference Coefficient

[0119]

[0120] Where N is the total number of communities, u and v are community indices, and p(x(t),u) and p(x(t),v) are the community indices. i ) are the attribute probability vectors of communities u and v, respectively; d(p(x(t),u) i ,p(x(t),v) i p(x(t),u) is calculated using the cosine distance. i and p(x(t),v) i The similarity is denoted by x, which represents the attribute vector; p(x(t),u) represents the proportion of t-dimensional attributes in community u.

[0121] This experiment compares the algorithm in this embodiment with the hierarchical clustering algorithm (H-cluster) and the community detection algorithm (Louvain). In the algorithm comparison, because the hierarchical clustering algorithm and the community detection algorithm cannot automatically determine the optimal number of communities, the number of communities needs to be given in advance. Therefore, the algorithm is compared with different community numbers (2, 3, 4, 5, 6), and the changes in different indicators are observed as follows. Figure 10 , Figure 11 , Figure 12 As shown.

[0122] As shown in the figure above, when the number of communities is 4, the method in this embodiment is superior to the other two methods. Figure 10 As can be seen, the density of the method in this embodiment is approximately 0.87, which is optimal. From the perspective of node attributes, the entropy value obtained by the method in this embodiment is... Figure 11 The optimal value is 0.16. Figure 12 In this embodiment, the topic degree of the method is 0.65, which is significantly higher than the other two methods. Therefore, the topic degree of the community obtained by the method of this embodiment is superior to that of other communities. In summary, the metric of the method of this embodiment is superior, indicating that the method proposed in this embodiment can obtain better community results compared to the other two methods.

[0123] In summary, this embodiment proposes a task decomposition method for crowdsourced design of home appliances. This method, through in-depth analysis of the task's characteristics, decomposes the task into more detailed, clear, and reasonable sub-tasks. Experimental results show that the method in this embodiment can better meet task requirements and improve the efficiency and quality of task completion. Task decomposition is a crucial process in crowdsourced design. The method proposed in this embodiment not only provides a new approach for home appliance design but also offers a reference for crowdsourced design in other fields.

[0124] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A task decomposition method for crowdsourced design of home appliances, characterized in that, Includes the following steps: S1. Based on the product BOM reference library, decompose the crowdsourcing design task according to the product structure tree to form a set of meta-tasks and determine the granularity reference value. S2. Represent the correlation between meta-tasks as a design structure matrix, measure the edges between tasks using edge weights, and use the K-means algorithm to cluster the candidate task set. S3. Determine the number of clusters K using the modularity method to obtain the design task decomposition set; S4. Calculate the granularity of the task decomposition set designed in step S3. If the granularity is within the granularity reference value, it is determined as the final task decomposition set; otherwise, determine the number of clusters K with the next lower level of modularity to obtain the design task decomposition set; repeat steps S3 and S4. In step S2, the correlation between meta-tasks is defined by the [0,1] average distribution and expressed as a fuzzy correlation value. The fuzzy correlation value increases as the correlation between tasks increases. Based on the fuzzy correlation values, the functional and structural correlations between each meta-task are determined, and a functional design matrix and a structural design matrix are constructed: In the formula, Indicates task and Submatrices between Indicates task and The correlation value between them; The correlation between tasks is measured using an edge weight metric, specifically: In the formula: and Representing tasks and tasks The degree, This is the edge weight coefficient, used to describe the relationship between edge weight and node degree; In step S2, Euclidean distance is used to calculate the design task. and design tasks Distance between Specifically: 。 2. The task decomposition method for crowdsourced design of home appliances as described in claim 1, characterized in that, In step S1, the relationships between tasks in the meta-task set are represented as follows: in, The vertex set of the directed graph of the product; representing the meta-tasks that make up the product; Let the set of edges be the product's directed graph. Indicates the starting node, Indicates the termination node; indicates the connection relationship between product meta-tasks; Represents the edge set in the directed graph of products. The corresponding weight set.

3. The task decomposition method for crowdsourced design of home appliances as described in claim 1, characterized in that, In step S3, the formula for calculating modularity is: In the formula The sum of the weights of all edges in the weighted task network, Adjacency matrix The Middle Okay, number The values ​​in the column represent the meta-tasks. and Task relevance; Indicates the number of meta-task nodes; Meta-task node The sum of the weights of all connecting edges; The value is 1 when two nodes belong to the same community and 0 when they belong to different communities.

4. The task decomposition method for crowdsourced design of home appliances as described in claim 3, characterized in that, Task granularity is measured by the average of all subtask granularity metrics: in, For the task Particle size, The task coupling coefficient is... The task cohesion coefficient. , The task correlation coefficient. The task reuse factor is as follows: In the formula, The number of meta-tasks. , To output the meta-task; , This is the input meta-task.

Images