IPC분류정보
국가/구분 |
United States(US) Patent
등록
|
국제특허분류(IPC7판) |
|
출원번호 |
US-0545050
(2009-08-20)
|
등록번호 |
US-8271311
(2012-09-18)
|
우선권정보 |
TW-98101787 A (2009-01-17) |
발명자
/ 주소 |
- Wang, Kung-Jeng
- Wang, Shih-Min
|
출원인 / 주소 |
- National Taiwan University of Science and Technology
|
대리인 / 주소 |
|
인용정보 |
피인용 횟수 :
0 인용 특허 :
15 |
초록
▼
A system and a method for resource allocation in the semiconductor testing industry are provided. In the system, an industry characteristic conversion module is used to transform the industry characteristic obtained from an input module into a chromosome structure. Next, an artificial intelligence e
A system and a method for resource allocation in the semiconductor testing industry are provided. In the system, an industry characteristic conversion module is used to transform the industry characteristic obtained from an input module into a chromosome structure. Next, an artificial intelligence evolution module proceeds to find an optimal solution by handling the chromosome structure until a candidate solution having a maximum final total profit converges to a value.
대표청구항
▼
1. A method for resource allocation in a semiconductor testing industry, comprising: receiving order information of a product and production information of a manufacturing department in the semiconductor testing industry performed by a computer, wherein the production information comprises multiple
1. A method for resource allocation in a semiconductor testing industry, comprising: receiving order information of a product and production information of a manufacturing department in the semiconductor testing industry performed by a computer, wherein the production information comprises multiple resources, the order information comprises a testing quantity corresponding to a product testing function requirement, and the product testing function requirement is fulfilled in cooperation with a primary resource and at least an auxiliary resource;converting the order information and the production information into an industry characteristic performed by the computer;according to the industry characteristic, establishing a production requirement constraint between the required testing quantity and a production quantity corresponding to each of the resources, and establishing a resource configuration constraint of the resources performed by the computer;converting a resource planning and capacity allocation plan into a chromosome structure performed by the computer;executing a genetic algorithm to obtain a set of candidate solutions in the chromosome structure and calculating total profits of the candidate solutions performed by the computer, so as to evaluate a maximum final total profit under consideration of time value, wherein the candidate solutions comply with the production requirement constraint and the resource configuration constraint; andobtaining the maximum final total profit of all the candidate solutions performed by the computer, so as to obtain the best candidate solution corresponding to the maximum final total profit. 2. The method for resource allocation in the semiconductor testing industry of claim 1, before the step of establishing the production requirement constraint and the resource configuration constraint, further comprising: according to a first feasibility parameter, evaluating whether the resources can cooperate with each other auxiliary resources under the product testing function requirement; andaccording to a second feasibility parameter, evaluating whether each of the resources can be used to obtain the product under the product testing function requirement. 3. The method for resource allocation in the semiconductor testing industry of claim 1, wherein the step of establishing the production requirement constraint comprises: under the product testing function requirement, limiting the production quantity corresponding to each of the resources to be smaller than or equal to the required testing quantity. 4. The method for resource allocation in the semiconductor testing industry of claim 3, wherein the production requirement constraint comprises: ∑mcm,tQp,m,t≤op,t;wherein p is an index of planning periods, cm,t represents whether the mth type resource can be used to obtain the product under a product testing function requirement t, Qp,m,t is a production quantity corresponding to the product testing function requirement t used in cooperation with the mth type resource in the pth period, and op,t is the testing quantity corresponding to the product testing function requirement t in the pth period. 5. The method for resource allocation in the semiconductor testing industry of claim 1, wherein the step of establishing the resource configuration constraint comprises: according to throughput, working hours, a target utilization of each of the resources, determining a production requirement quantity; andunder the product testing function requirement, limiting a sum of the on-hand quantity and the transferred quantity to be greater than or equal to the production requirement quantity. 6. The method for resource allocation in the semiconductor testing industry of claim 5, wherein the resource configuration constraint comprises: Kp,m+∑zXp,m,z≥∑tcm,tQp,m,trm,twp,myp,m;wherein p is an index of planning periods, is an on-hand quantity of the mth type resource in the pth period, Xp,m,z is a transferred quantity of the mth type of resource obtained from a channel z in the pth period, cm,t represents whether the mth type resource can be used to obtain the product under the product testing function requirement t, Qp,m,t is a production quantity corresponding to the product testing function requirement t used in cooperation with the mth type resource in the pth period, rm,t is the throughput of the mth type resource in the pth period under the product testing function requirement t, wp,m is the working hours of the mth type resource in the pth period, and yp,m is the target utilization rate of the mth type of resource in the pth period. 7. The method for resource allocation in the semiconductor testing industry of claim 1, wherein the step of calculating the total profits of the candidate solutions comprises: in a multiple planning periods under the product testing function requirement, sequentially adding a profit from the previous planning period and a production profit of the present planning period, subtracting a transferring cost of transferring each of the resources through the channel, and subtracting a resource variation cost of the present planning period, so as to obtain a total profit of the present planning period, wherein the resource variation cost is obtained by calculating the on-hand quantity in the present planning period and the quantity in the previous planning period of each of the resources. 8. The method for resource allocation in the semiconductor testing industry of claim 7, wherein the total profits of the candidate solutions are calculated according to a following equation: Fp=Fp-1(1+Ip-1)-∑m,zup,m,zXp,m,z-∑mGp,m+∑m,tBm,tQp,m,t;wherein p is an index of planning periods, Fp is the total profit in the pth period, F0 is a profit in the initial planning period, Ip is an interest rate in the pth period, up,m,z is the transferring cost of obtaining the mth type resource through the channel z in the pth period, Xp,m,z is the transferred quantity of obtaining the mth type resource through the channel z in the pth period, Gp,m is the resource variation cost of the mth type resource in the pth period, Bm,t is a profit obtained under the product testing function requirement t in cooperation with the mth type resource, and Qp,m,t is the production quantity corresponding to the product testing function requirement t used in cooperation with the mth type resource in the pth period. 9. The method for resource allocation in the semiconductor testing industry of claim 7, wherein the step of calculating the total profits of the candidate solutions to evaluate the maximum final total profit under consideration of time value comprises: according to the on-hand quantity of each of the resources in the last planning period, calculating a declined residual value of the resources; andaccording to the total profit in the last planning period and the declined residual value, obtaining the maximum final profit at the end of the planning horizons. 10. The method for resource allocation in the semiconductor testing industry of claim 9, wherein the maximum final total profit is evaluated according to a following equation: MAX1∏p(1+Ip)[∑m(dpend,mKpend,m)+Fpend];wherein p is an index of planning periods, Fpend is a total profit in the last planning period, Ip is an interest rate in the pth period, Kpend,m is the on-hand quantity of the mth type resource in the last planning period, and dpend,m is a declined residual value of the mth type resource in the last planning period. 11. The method for resource allocation of the semiconductor industry of claim 1, wherein the step of executing the genetic algorithm to obtain the candidate solutions in the chromosome structure comprises: randomly generating initial candidate solutions in the chromosome structure;individually evaluating whether the candidate solutions comply with the production requirement constraint and the resource configuration constraint;when one of the candidate solutions does not comply with at least one of the production requirement constraint and the resource configuration constraint, randomly selecting another candidate solution to be the candidate solution in the feasible range of both the production requirement constraint and the resource configuration constraint; andperforming a regeneration evolution, a crossover evolution and a mutation evolution on the candidate solutions, then re-evaluating whether the candidate solutions comply with the production requirement constraint and the resource configuration constraint, and repeatedly performing the above regeneration evolution, crossover evolution and mutation evolution until the maximum final total profit corresponding to the best candidate solutions converges to a value. 12. The method for resource allocation in the semiconductor testing industry of claim 11, wherein the crossover evolution comprises one of a random selecting one-point crossover, a two-point crossover, an uniform crossover, a arithmetical crossover, an uniform arithmetical crossover and a blind crossover operator. 13. A system for resource allocation of a semiconductor testing industry, comprising: a storage device, storing a plurality of programs written in a computer language; anda central processing unit, executing the programs;wherein the programs comprises: an input module for receiving order information of a product and production information of a manufacturing department in the semiconductor testing industry, wherein the production information comprises multiple resources, the order information comprises a testing quantity corresponding to the product under a product testing function requirement, and the product testing function requirement is fulfilled by a primary resource in cooperation with at least an auxiliary resource;an industry characteristic conversion module, comprising: a problem extraction module for converting the order information and the production information into an industry characteristic, establishing a production requirement constraint between the required testing quantity and a production quantity corresponding to each of the resources, and establishing a resource configuration constraint of the resources; anda chromosome transformation module for converting a resource planning and capacity allocation plan into a chromosome structure;an artificial intelligence evolution module for executing a genetic algorithm to obtain a set of candidate solutions in the chromosome structure and calculating total profits of the candidate solutions to obtain a maximum final total profit under consideration of time value, wherein the candidate solutions comply with the production requirement constraint and the resource configuration constraint; andan output module for outputting the candidate solution corresponding to the maximum final total profit. 14. The system for resource allocation in the semiconductor testing industry of claim 13, wherein the problem extraction module comprises: a production requirement evaluation module for limiting the production quantity corresponding to each of the resources to be smaller than or equal to the testing quantity under the product testing function requirement, so as to establish the production requirement constraint;a resource requirement evaluation module for limiting a sum of the on-hand quantity and the transferred quantity to be greater than or equal to a production requirement quantity under the product testing function requirement, so as to establish the resource configuration constraint, wherein the production requirement quantity is determined according to throughput, working hours and a target utilization rate of each of the resources. 15. The system for resource allocation in the semiconductor testing industry of claim 13, wherein the problem extraction module further comprises: a resource configuration evaluation module for evaluating whether the resources can be used in cooperation under the product testing function requirement according to a first feasibility parameter, and evaluating whether each of the resources can be used to obtain the product under the product testing function requirement according to a second feasibility parameter. 16. The system for resource allocation in the semiconductor testing industry of claim 13, wherein the artificial intelligence evolution module comprises: an initializing module for randomly generating the candidate solutions in the chromosome structure;an infeasible planning repair module for respectively evaluating whether the candidate solutions comply with production requirement constraint and the resource configuration constraint, so that when one of the candidate solutions does not comply with at least one of the production requirement constraint and the resource configuration constraint, regenerating another candidate solution as the candidate solution in a feasible range of both the production requirement constraint and the resource configuration constraint;an evaluation module for calculating the total profits of the candidate solutions, so as to evaluate the maximum final total profit corresponding to each of the candidate solutions;a regeneration module for executing a regeneration evolution using the candidate solution repaired by the infeasible planning repair module;a crossover module randomly selecting two candidate solutions in the candidate solutions generated in the regeneration evolution, so as to execute the crossover evolution;a mutation module for executing a mutation evolution using the candidate solutions generated by the crossover evolution,wherein the infeasible planning repair module is used to re-evaluated the candidate solutions generated by the mutation module, so as to ensure that the candidate solutions generated by the mutation module comply with the production requirement constraint and the resource configuration constraint, and the above evolutions are repeatedly performed until the maximum final total profit obtained from the evaluation module converges to a value. 17. The system for resource allocation in the semiconductor testing industry of claim 16, wherein the evaluation module further comprises: a resource variation evaluation module for calculating a cost variation from the on-hand quantity of the present planning period and the on-hand quantity of the previous planning period;a present planning period profit evaluation module for sequentially adding a profit from the previous planning period and a production profit of the present planning period under the product testing function requirement, subtracting a transferring cost of transferring each of the resources through the channel, and subtracting the resource variation cost in the present planning period; anda maximum profit evaluation module for calculating a declined residual value of the on-hand resources according to the on-hand quantity of each of the resources in a last planning period, and evaluating the maximum final total profit according to the total profit and the declined residual value in the last planning period. 18. The system for resource allocation in the semiconductor testing industry of claim 16, wherein the crossover evolution comprises randomly selecting one of a one-point crossover, two-point crossover, uniform crossover, arithmetical crossover, uniform arithmetical crossover and blind crossover operator.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.