CN-122022384-A - Scheduling method and equipment for semiconductor furnace area batch processing equipment based on rule decoding

CN122022384ACN 122022384 ACN122022384 ACN 122022384ACN-122022384-A

Abstract

The invention belongs to the technical field related to semiconductor manufacturing scheduling and intelligent optimization, and discloses a scheduling method and equipment for batch processing equipment of a semiconductor furnace area based on rule decoding, which provide an integrated solving method of coding initialization, rule active decoding, fitness evaluation, elite reservation, genetic evolution and hierarchical neighborhood reinforcement. The invention brings the process sequence decision, the machine allocation decision, the batch forming decision and the time window control decision into a unified solving chain, avoids the conflict accumulation problem caused by the staged isolation decision in the prior method, realizes the comprehensive optimization of the maximum finishing time, the batch rate and the time window violation rate by cooperatively processing the process sequencing, the machine allocation, the batch forming and the time window control in the unified solving chain, and can provide an efficient, stable and executable scheduling scheme for the batch processing equipment of the semiconductor furnace area.

Inventors

ZHANG CHUNJIANG
HU ZISHUN
GAO YIFAN
LI XINYU
GAO LIANG
LIU QIHAO

Assignees

华中科技大学

Dates

Publication Date: 20260512
Application Date: 20260327

Claims (10)

1.A dispatching method of semiconductor furnace area batch processing equipment based on rule decoding is characterized by comprising the following steps: S1, generating an initial population by utilizing a mode of combining random initialization and maximum batch efficiency initialization based on workshop workpieces, working procedures, optional machines, batch processing constraints and time window constraints, and expressing individuals in the initial population by adopting a double-layer chromosome formed by working procedure sequence codes and machine selection codes; S2, executing rule-based active decoding on individuals in the initial population, and converting coding information corresponding to the individuals into an executable scheduling scheme meeting the process sequence relation, batch processing constraint and time window constraint; s3, carrying out fitness evaluation on the obtained executable scheduling scheme, and executing elite retention strategy on individuals according to the fitness evaluation result to obtain dominant individuals participating in the next round of evolution and forming the dominant individuals into a new population; s4, carrying out genetic operation on individuals in the new population to obtain candidate individuals; and S5, carrying out hierarchical neighborhood search on the scheduling schemes corresponding to the candidate individuals, taking the key batches as local search objects according to the key path and the key block information, and sequentially carrying out cross-machine batch movement and same-machine batch movement to obtain an optimal scheduling scheme, wherein the key path, the key block and the key batches are identified based on the scheduling schemes corresponding to the candidate individuals.
2. The method for dispatching semiconductor furnace area batch processing equipment based on rule decoding according to claim 1, wherein a comprehensive fitness function is constructed based on maximum completion time, batch rate and time window violation rate, and the fitness evaluation is performed on the obtained executable dispatching scheme based on the comprehensive fitness function.
3. The method for dispatching semiconductor furnace area batch processing equipment based on regular decoding according to claim 2, wherein the comprehensive fitness function is: Wherein, the Representing individuals Is used for the comprehensive fitness value of the model (1), And Respectively representing the normalization results of the total processing time, the batch rate and the time window violation rate, Is the corresponding weight coefficient and satisfies 。
4. The method for dispatching semiconductor furnace area batch processing equipment based on regular decoding according to claim 1, wherein the initial population is composed of two parts, wherein the first part is a random initialization individual and is obtained by randomly arranging working procedures and randomly selecting a feasible machine, the second part is a maximum batch efficiency initialization individual and is obtained by preferentially selecting a machine which can be matched with the formed batch in a selectable machine set, and if no batch machine exists, the earliest machine which can be started is selected.
5. The method for dispatching semiconductor furnace area batch processing equipment based on rule decoding according to any one of claims 1-4, wherein the current process to be dispatched is sequentially extracted according to the process sequence codes, and the target processing machine of the current process to be dispatched is determined by combining the machine selection codes.
6. The dispatching method of semiconductor furnace area batch processing equipment based on regular decoding according to claim 5, wherein the method is characterized in that type identification is carried out on a current to-be-dispatched process, when the current to-be-dispatched process is a time window middle process, a feasible insertion position is searched firstly based on an active insertion rule, then the finishing time of the process is compared with the latest starting time of a corresponding subsequent process, if the finishing time of the process is not later than the latest starting time of the subsequent process, a current decoding result is maintained, if the finishing time of the process is later than the latest starting time of the subsequent process, a selectable machine set of the process is traversed, the earliest starting machine is reselected for processing, when the current process is not a time window middle process, whether a compatible batch exists in the current insertion position is judged firstly, if the current insertion position exists, the process is incorporated into the compatible batch, if the current insertion position does not exist, a batch is searched forwards and backwards along the current insertion position, the moving direction of the batch is determined according to an additional waiting time minimum principle, if the latest starting time of the batch is not found, a new batch is created at the current insertion position, when the current insertion position is the latest starting time of the corresponding window is later than the latest starting time of the subsequent process, and the latest starting time is calculated according to the latest starting time of the corresponding window, and the latest starting time of the corresponding process is calculated according to the latest starting time of the window middle process.
7. The method for dispatching semiconductor furnace area batch processing equipment based on rule decoding according to any one of claims 1 to 4, wherein in step S4, a roulette selection strategy is adopted to select father individuals from the current population, then crossover and mutation are executed for process sequence codes and machine selection codes respectively, POX operators are adopted to recombine father process sequences for process sequence codes, new process arrangement is formed on the basis of keeping partial priority process relative order, insertion mutation operators are adopted to adjust local process positions, uniform crossover operators are adopted to exchange machine allocation genes of different father individuals for machine selection codes, and single-point mutation operators are adopted to randomly change equipment selection results of target processes.
8. The method for dispatching semiconductor furnace area batch processing equipment based on rule decoding as claimed in any one of claims 1-4, wherein a critical path and a critical block in a dispatching scheme are identified, and the critical batch is divided into a block head batch, a block inner batch and a block tail batch, and a target batch on the critical path is selected Traversing its set of selectable machines to search for idle or batch periods on the target machine in chronological order, for candidate insertion locations to insert a previous batch of locations Is to be completed in the time of completion of (a) And a batch after the insertion site Is to start time of (a) Determining candidate insertion intervals and combining target batches Time window interval of batch structure Performing feasibility judgment when the overlapping length of the candidate insertion interval and the time window interval is longer than the processing time length of the target batch When the target batch is moved to the candidate position, the load between the devices is redistributed; After the movement of the machine-crossing batch is completed, the movement of the same machine batch is further executed, and under the condition that the processing machine where the target batch is located is kept unchanged, the subsequent insertable period is searched when the target batch is the block head batch, the preceding insertable period is searched when the target batch is the block tail batch, and the preceding and following insertable periods on the machine where the target batch is located are searched in time sequence when the target batch is the block inner batch.
9. A semiconductor furnace area batch processing equipment scheduling system based on rule decoding is characterized by comprising a memory and a processor, wherein the memory stores a computer program, and the processor executes the method for scheduling semiconductor furnace area batch processing equipment based on rule decoding according to any one of claims 1-8 when executing the computer program.
10. A computer readable storage medium storing machine executable instructions which when invoked and executed by a processor cause the processor to implement the rule-based decoded semiconductor furnace area batch device scheduling method of any one of claims 1-8.

Description

Scheduling method and equipment for semiconductor furnace area batch processing equipment based on rule decoding Technical Field The invention belongs to the technical field related to semiconductor manufacturing scheduling and intelligent optimization, and particularly relates to a scheduling method and equipment for batch processing equipment of a semiconductor furnace area based on rule decoding. Background The fourth industrial revolution is pushing the global manufacturing industry to accelerate the transition to intelligentization and digitalization, the semiconductor industry is used as the core of the electronic information industry, the market demand for semiconductor products is exponentially increased, and enterprises are urgent to break through the traditional efficiency bottleneck through flexible production modes of multiple varieties and small batches so as to adapt to the core requirements of the production industry. Wafer fabrication is a fundamental step in the semiconductor industry, where the furnace area is a core bottleneck for increased throughput due to the critical processes such as oxidation, thin film deposition, etc. The area has three significant problems that firstly, the processing period is long due to a high-precision process, for example, advanced film deposition single processing can reach several hours and strict quality inspection is needed, secondly, batch division and processing sequence planning are complex in a batch processing mode, the production efficiency and quality are directly influenced, thirdly, the time window constraint is adopted, the wafer is subjected to out-of-date processing, the performance degradation is easily caused, for example, the surface is oxidized due to the fact that the wafer is not passivated in time after etching, the chip yield is influenced, and a great challenge is brought to production management. Currently, the production scheduling of a furnace tube region mainly depends on manual experience, and under the problems of long processing period, complex batch processing, time window constraint, equipment isomerism, reentrant working procedures and the like, the efficiency fluctuation of the traditional scheduling mode is large, and the semiconductor manufacturing requirement is difficult to meet. Disclosure of Invention Aiming at the defects or improvement demands of the prior art, the invention provides a dispatching method and equipment for semiconductor furnace area batch processing equipment based on rule decoding, which aim to solve the problem of low batch efficiency in the dispatching process of the conventional semiconductor furnace area batch processing equipment. In order to achieve the above object, according to one aspect of the present invention, there is provided a dispatching method for semiconductor furnace area batch processing equipment based on rule decoding, comprising the steps of: S1, generating an initial population by utilizing a mode of combining random initialization and maximum batch efficiency initialization based on workshop workpieces, working procedures, optional machines, batch processing constraints and time window constraints, and expressing individuals in the initial population by adopting a double-layer chromosome formed by working procedure sequence codes and machine selection codes; S2, executing rule-based active decoding on individuals in the initial population, and converting coding information corresponding to the individuals into an executable scheduling scheme meeting the process sequence relation, batch processing constraint and time window constraint; s3, carrying out fitness evaluation on the obtained executable scheduling scheme, and executing elite retention strategy on individuals according to the fitness evaluation result to obtain dominant individuals participating in the next round of evolution and forming the dominant individuals into a new population; s4, carrying out genetic operation on individuals in the new population to obtain candidate individuals; and S5, carrying out hierarchical neighborhood search on the scheduling schemes corresponding to the candidate individuals, taking the key batches as local search objects according to the key path and the key block information, and sequentially carrying out cross-machine batch movement and same-machine batch movement to obtain an optimal scheduling scheme, wherein the key path, the key block and the key batches are identified based on the scheduling schemes corresponding to the candidate individuals. Further, a comprehensive fitness function is constructed based on the maximum completion time, the batch rate and the time window violation rate, and fitness evaluation is performed on the obtained executable scheduling scheme based on the comprehensive fitness function. Further, the comprehensive fitness function is: Wherein, the Representing individualsIs used for the comprehensive fitness value of the model (1),AndRespectively repr