Chip real-time data modeling system and method for digital twinning

Abstract

The present application relates to a chip real-time data modeling system and method for digital twinning, by pre-establishing digital twinning models of multiple same type chips, during the manufacturing process of the target chip, firstly, similarity retrieval is performed to see whether there is a prior same type chip within the similarity threshold, if there is, the next time control instruction of the prior same type chip is directly used to control the next time of the target chip, if not, the deviation value between the second state data of the current time and the average value of the first state data of the prior same type chip at the same time is calculated, and the deviation value is used as the control instruction of the next time state data, the corresponding control instruction is generated through the large model, and the related equipment at the next time is controlled, the method proposed in the present application can cope with complex scenes of multi-parameter coupling and nonlinear disturbance, and can also ensure the stable manufacturing of the target chip under extreme working conditions.

Description

Technical Field

[0001] This invention relates to the field of chip data processing technology, and in particular to a chip real-time data modeling system and method for digital twins. Background Technology

[0002] Digital twins refer to a technology that uses digital means to construct a digital model in virtual space that is highly similar to a physical entity, and uses real-time data exchange and analysis to monitor, analyze and optimize the physical entity.

[0003] In the chip manufacturing process, it is necessary to strictly control the relevant status data of the manufacturing process. This relevant status data includes environmental parameters and electrical signal parameters, which are specifically controlled through relevant equipment.

[0004] Existing methods often involve precise control of each relevant state data item through corresponding control algorithms. However, this approach cannot handle complex scenarios involving multi-parameter coupling and nonlinear disturbances, and is prone to control oscillations and overshoot under extreme conditions. Summary of the Invention

[0005] The purpose of this invention is to at least address one of the shortcomings of the prior art and provide a chip real-time data modeling system and method for digital twins.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: Specifically, a real-time chip data modeling method for digital twins is proposed, including the following: Step 110: Obtain digital twin models of multiple chips of the same type that were previously established for the target chip, and read the first state data of any chip of the same type at any moment in the manufacturing process through the digital twin models; Step 120: Obtain the real-time status data of the target chip during the manufacturing process as the second status data; Step 130: Calculate the similarity between the second state data and the first state data at the same time, and determine whether there is any first state data with a similarity greater than the first threshold; Step 140: If it exists, use the digital twin model corresponding to the first state data with the highest similarity to guide the adjustment of the state data at the next moment; if it does not exist, calculate the deviation value between the second state data at the current moment and the first state data at the same moment, and use the deviation value as the guidance for the adjustment of the state data at the next moment. Step 150: Based on the control guidance, call the large model to calculate the control instructions for the state data at the next moment, and control the corresponding operation of the relevant equipment based on the control instructions.

[0007] Furthermore, specifically, acquiring pre-established digital twin models of multiple similar chips for the target chip, including, The first state data of the same type of chip during the manufacturing process is obtained by collecting sensor data from pre-deployed related sensors. Based on this, a digital twin model is generated using a digital twin algorithm. A database is then built based on this data. By reading the data in the database, a digital twin model of any chip can be obtained.

[0008] Further, specifically, calculating the similarity between the second state data and the first state data at the same time includes, The corresponding weights of all data types in the first state data and the second state data are predetermined. After normalizing all data types, the similarity between the second state data and the first state data at the same time is calculated by weighted summation.

[0009] Furthermore, specifically, the corresponding weights of all data types in the first-state data and the second-state data are pre-determined using the entropy weight method. The specific calculation process is as follows: Construct the original data matrix. ; in, This represents the value of the j-th type of state data under the i-th sample, where m is the total number of samples and n is the total number of state data types. The original data matrix is ​​standardized to obtain a standardized matrix. ; in The data is dimensionless and standardized. For the j-th type of state data, calculate the proportion of the value of the i-th sample in the total sample sum of the index. , , And on Superimposed Minimum Then re-normalize to ensure computational stability; Then the information entropy of the j-th type of state data The calculation formula is as follows: , in, The entropy normalization coefficient is used to ensure... ∈[0,1]; Calculate information utility value ; The information utility values ​​are normalized to obtain the final weights of the j-th class of state data. , , And it satisfies the following constraints: , .

[0010] Furthermore, specifically, calculating the deviation between the second state data at the current moment and the first state data at the same moment includes, ={ }; Where D is the deviation value. This represents the normalized value of the j-th type of state data in the second state data at the current time t. Let be the normalized average value of the j-th class of state data in the first state data at the current time t, where j∈[1, n].

[0011] Furthermore, specifically, the control instructions based on the aforementioned control guidance, which invoke the large model to calculate the state data for the next moment, include: A structured Prompt is constructed based on the aforementioned control guidance, the second state data at the current moment, device constraints, and output format requirements; The control instructions based on the structured Prompt call to the large model output a standardized JSON structure. The standardized JSON structure includes fields such as device identifier, control parameter name, target adjustment value, adjustment step size, and expected control effect, which can be directly used for subsequent device control.

[0012] Furthermore, the method also includes, after the manufacturing of the target chip is completed, generating a digital twin model of the target chip using a digital twin algorithm and storing it in the database to update the database.

[0013] This invention also proposes a real-time chip data modeling system for digital twins, comprising the following: The first data acquisition module is used to acquire digital twin models of multiple chips of the same type that have been pre-established for the target chip, and to read the first state data of any chip of the same type at any time during the manufacturing process through the digital twin model. The second data acquisition module is used to acquire real-time status data of the target chip during the manufacturing process as second status data; The similarity calculation module is used to calculate the similarity between the second state data and the first state data at the same time, and to determine whether there is a first state data with a similarity greater than a first threshold. The judgment module is used to make the following judgments: if they exist, the digital twin model corresponding to the first state data with the highest similarity is used to guide the adjustment of the state data at the next moment; if they do not exist, the deviation value between the second state data at the current moment and the first state data at the same moment is calculated, and the deviation value is used as the adjustment guide for the state data at the next moment. The equipment control module is used to call the control instructions based on the control guidance to calculate the state data of the next moment from the large model, and to control the operation of the relevant equipment based on the control instructions.

[0014] The beneficial effects of this invention are as follows: This invention proposes a real-time data modeling system and method for chips based on digital twins. By pre-establishing multiple digital twin models of the same type of chip, during the manufacturing process of the target chip, a similarity search is first performed to check if there is a prior chip of the same type within the similarity threshold. If so, the control command of the prior chip of the same type at the next moment is directly used to control the target chip at the next moment. If not, the deviation between the second state data at the current moment and the average value of the first state data of the prior chip of the same type at the same moment is calculated. The deviation value is then used as the control guide for the state data at the next moment. The large model generates corresponding control commands to control the relevant equipment at the next moment. The method proposed in this invention can cope with complex scenarios of multi-parameter coupling and nonlinear disturbances, and can ensure the stable manufacturing of the target chip even under extreme conditions. Attached Figure Description

[0015] The above and other features of this disclosure will become more apparent from the detailed description of the embodiments illustrated in conjunction with the accompanying drawings. In the accompanying drawings, the same reference numerals denote the same or similar elements. Obviously, the drawings described below are merely some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained from these drawings without any creative effort. In the drawings: Figure 1 The flowchart shown is a real-time chip data modeling method for digital twins according to the present invention. Detailed Implementation

[0016] The following will provide a clear and complete description of the concept, specific structure, and technical effects of the present invention in conjunction with embodiments and accompanying drawings, so as to fully understand the purpose, solution, and effects of the present invention. It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The same reference numerals used throughout the accompanying drawings indicate the same or similar parts.

[0017] Example 1, referring to Figure 1This invention proposes a real-time chip data modeling method for digital twins, including the following: Step 110: Obtain digital twin models of multiple chips of the same type that were previously established for the target chip, and read the first state data of any chip of the same type at any moment in the manufacturing process through the digital twin models; Step 120: Obtain the real-time status data of the target chip during the manufacturing process as the second status data; Step 130: Calculate the similarity between the second state data and the first state data at the same time, and determine whether there is any first state data with a similarity greater than the first threshold; Step 140: If it exists, use the digital twin model corresponding to the first state data with the highest similarity to guide the adjustment of the state data at the next moment; if it does not exist, calculate the deviation value between the second state data at the current moment and the first state data at the same moment, and use the deviation value as the guidance for the adjustment of the state data at the next moment. Step 150: Based on the control guidance, call the large model to calculate the control instructions for the state data at the next moment, and control the corresponding operation of the relevant equipment based on the control instructions.

[0018] In this embodiment 1, by pre-establishing multiple digital twin models of the same type of chip, during the manufacturing process of the target chip, a similarity search is first performed to see if there is a prior chip of the same type within the similarity threshold. If there is, the target chip is directly controlled in the next moment using the control command of the prior chip of the same type. If there is no chip, the deviation between the second state data at the current moment and the average value of the first state data of the prior chip of the same type at the same moment is calculated. The deviation value is then used as the control guide for the state data at the next moment. The large model generates corresponding control commands to control the relevant equipment at the next moment. The method proposed in this invention can cope with complex scenarios of multi-parameter coupling and nonlinear disturbances, and can ensure the stable manufacturing of the target chip even under extreme conditions.

[0019] In a preferred embodiment of the present invention, specifically, acquiring a digital twin model of multiple identical chips pre-established for the target chip includes, The first state data of the same type of chip during the manufacturing process is obtained by collecting sensor data from pre-deployed related sensors. Based on this, a digital twin model is generated using a digital twin algorithm. A database is then built based on this data. By reading the data in the database, a digital twin model of any chip can be obtained.

[0020] In a preferred embodiment of the present invention, specifically, calculating the similarity between the second state data and the first state data at the same time includes, The corresponding weights of all data types in the first state data and the second state data are predetermined. After normalizing all data types, the similarity between the second state data and the first state data at the same time is calculated by weighted summation.

[0021] In this preferred embodiment, by calculating the similarity using the weighted Euclidean distance in the above manner, it is possible to find the prior chip with similar state data, and then use the relevant control methods of the prior chip to guide the control of the next moment.

[0022] In a preferred embodiment of the present invention, the corresponding weights of all data types in the first state data and the second state data are predetermined using the entropy weight method. The specific calculation process is as follows. Construct the original data matrix. ; in, This represents the value of the j-th type of state data under the i-th sample, where m is the total number of samples and n is the total number of state data types. The original data matrix is ​​standardized to obtain a standardized matrix. ; in The data is dimensionless and standardized. For the j-th type of state data, calculate the proportion of the value of the i-th sample in the total sample sum of the index. , , And on Superimposed Minimum Then re-normalize to ensure computational stability; Then the information entropy of the j-th type of state data The calculation formula is as follows: , in, The entropy normalization coefficient is used to ensure... ∈[0,1]; Calculate information utility value ; The information utility values ​​are normalized to obtain the final weights of the j-th class of state data. , , And it satisfies the following constraints: , .

[0023] In this preferred embodiment, the corresponding weights of all data types in the first state data and the second state data are determined by the entropy weight method, which can greatly eliminate the influence of subjective factors and obtain accurate weight coefficients.

[0024] In a preferred embodiment of the present invention, specifically, calculating the deviation between the second state data at the current moment and the first state data at the same moment includes, ={ }; Where D is the deviation value. This represents the normalized value of the j-th type of state data in the second state data at the current time t. Let be the normalized average value of the j-th class of state data in the first state data at the current time t, where j∈[1, n].

[0025] In this preferred embodiment, by combining the difference between each state data and the average value under the historical curve in its database to form an orange deviation value D, the control commands generated by the large model are coupled with multiple parameters when input into the subsequent large model.

[0026] In a preferred embodiment of the present invention, specifically, the control instruction based on the control guidance, which calls a large model to calculate the state data for the next time step, includes: A structured Prompt is constructed based on the aforementioned control guidance, the second state data at the current moment, device constraints, and output format requirements; The control instructions based on the structured Prompt call to the large model output a standardized JSON structure. The standardized JSON structure includes fields such as device identifier, control parameter name, target adjustment value, adjustment step size, and expected control effect, which can be directly used for subsequent device control.

[0027] In this preferred embodiment, the control command is generated using a direct Prompt call method, as detailed below: The control guidance, current status data, equipment constraints, and output format requirements are encapsulated into structured prompt words, which can then directly call the text generation / completion interface of the large model.

[0028] Input: The Prompt is constructed in modules, and it is fixed to include the control target and guidance rules, the full state data at the current time, the boundary constraints of the adjustable parameters of the equipment, and the output format requirements (mandatory JSON structure). It can be supplemented with 1 to 3 sets of historical control examples for small sample guidance.

[0029] Call format: Directly call the large model dialogue completion API, and complete the generation in a single round of requests.

[0030] Output constraints: The Prompt function forces the large model to output a standardized JSON structure, which includes fields such as device identifier, control parameter name, target control value, control step size, and expected control effect, and can be directly used for subsequent device control.

[0031] In a preferred embodiment of the present invention, the method further includes, after the manufacturing of the target chip is completed, generating a digital twin model of the target chip using a digital twin algorithm and storing it in the database to update the database.

[0032] In this preferred embodiment, after each target chip is manufactured, its digital twin model is built to update the database, thereby providing more accurate guidance for subsequent chip manufacturing processes.

[0033] Example 2: This invention also proposes a real-time chip data modeling system for digital twins, comprising the following: The first data acquisition module is used to acquire digital twin models of multiple chips of the same type that have been pre-established for the target chip, and to read the first state data of any chip of the same type at any time during the manufacturing process through the digital twin model. The second data acquisition module is used to acquire real-time status data of the target chip during the manufacturing process as second status data; The similarity calculation module is used to calculate the similarity between the second state data and the first state data at the same time, and to determine whether there is a first state data with a similarity greater than a first threshold. The judgment module is used to make the following judgments: if they exist, the digital twin model corresponding to the first state data with the highest similarity is used to guide the adjustment of the state data at the next moment; if they do not exist, the deviation value between the second state data at the current moment and the first state data at the same moment is calculated, and the deviation value is used as the adjustment guide for the state data at the next moment. The equipment control module is used to call the control instructions based on the control guidance to calculate the state data of the next moment from the large model, and to control the operation of the relevant equipment based on the control instructions.

[0034] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0035] If the integrated module is implemented as a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or system capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0036] Although the description of the invention has been quite detailed and particularly of several described embodiments, it is not intended to limit it to any of these details or embodiments or any particular embodiment, but should be considered as providing a broad possible interpretation of the claims by referring to the appended claims and taking into account the prior art, thereby effectively covering the intended scope of the invention. Furthermore, the invention has been described above with respect to embodiments foreseeable by the inventors in order to provide a useful description, and non-substantial modifications to the invention that have not yet been foreseen may still represent equivalent modifications.

[0037] The above description is merely a preferred embodiment of the present invention. The present invention is not limited to the above-described embodiments. Any embodiment that achieves the technical effects of the present invention using the same means should fall within the protection scope of the present invention. Within the protection scope of the present invention, various modifications and variations can be made to the technical solutions and / or implementation methods.

Claims

1. A chip real-time data modeling method for digital twins, characterized in that, Including the following: Step 110: Obtain digital twin models of multiple chips of the same type that were previously established for the target chip, and read the first state data of any chip of the same type at any moment in the manufacturing process through the digital twin models; Step 120: Obtain the real-time status data of the target chip during the manufacturing process as the second status data; Step 130: Calculate the similarity between the second state data and the first state data at the same time, and determine whether there is any first state data with a similarity greater than the first threshold; Step 140: If it exists, use the digital twin model corresponding to the first state data with the highest similarity to guide the adjustment of the state data at the next moment; if it does not exist, calculate the deviation value between the second state data at the current moment and the first state data at the same moment, and use the deviation value as the guidance for the adjustment of the state data at the next moment. Step 150: Based on the control guidance, call the large model to calculate the control instructions for the state data at the next moment, and control the corresponding operation of the relevant equipment based on the control instructions.

2. The chip real-time data modeling method for digital twins according to claim 1, characterized in that, Specifically, acquiring digital twin models of multiple similar chips pre-established for the target chip, including, The first state data of the same type of chip during the manufacturing process is obtained by collecting sensor data from pre-deployed related sensors. Based on this, a digital twin model is generated using a digital twin algorithm. A database is then built based on this data. By reading the data in the database, a digital twin model of any chip can be obtained.

3. The chip real-time data modeling method for digital twins according to claim 1, characterized in that, Specifically, calculating the similarity between the second state data and the first state data at the same time includes, The corresponding weights of all data types in the first state data and the second state data are predetermined. After normalizing all data types, the similarity between the second state data and the first state data at the same time is calculated by weighted summation.

4. The chip real-time data modeling method for digital twins according to claim 3, characterized in that, Specifically, the corresponding weights of all data types in the first-state data and the second-state data are pre-determined using the entropy weight method. The specific calculation process is as follows: Construct the original data matrix. ； in, This represents the value of the j-th type of state data under the i-th sample, where m is the total number of samples and n is the total number of state data types. The original data matrix is ​​standardized to obtain a standardized matrix. ； in The data is dimensionless and standardized. For the j-th type of state data, calculate the proportion of the value of the i-th sample in the total sample sum of the index. , ， And on Superimposed Minimum Then re-normalize to ensure computational stability; Then the information entropy of the j-th type of state data The calculation formula is as follows: ， in, The entropy normalization coefficient is used to ensure... ∈[0,1]; Calculate information utility value ; The information utility values ​​are normalized to obtain the final weights of the j-th class of state data. , ， And it satisfies the following constraints: , .

5. The chip real-time data modeling method for digital twins according to claim 4, characterized in that, Specifically, the deviation between the second state data at the current moment and the first state data at the same moment is calculated, including: ={ }； Where D is the deviation value. This represents the normalized value of the j-th type of state data in the second state data at the current time t. Let be the normalized average value of the j-th class of state data in the first state data at the current time t, where j∈[1, n].

6. The chip real-time data modeling method for digital twins according to claim 1, characterized in that, Specifically, the control instructions, based on the aforementioned control guidance, call the large model to calculate the state data for the next time step, include: A structured Prompt is constructed based on the aforementioned control guidance, the second state data at the current moment, device constraints, and output format requirements; The control instructions based on the structured Prompt call to the large model output a standardized JSON structure. The standardized JSON structure includes fields such as device identifier, control parameter name, target adjustment value, adjustment step size, and expected control effect, which can be directly used for subsequent device control.

7. The chip real-time data modeling method for digital twins according to claim 2, characterized in that, The method further includes, after the manufacturing of the target chip is completed, generating a digital twin model of the target chip using a digital twin algorithm and storing it in the database to update the database.

8. A chip real-time data modeling system for digital twins, characterized in that, Including the following: The first data acquisition module is used to acquire digital twin models of multiple chips of the same type that have been pre-established for the target chip, and to read the first state data of any chip of the same type at any time during the manufacturing process through the digital twin model. The second data acquisition module is used to acquire real-time status data of the target chip during the manufacturing process as second status data; The similarity calculation module is used to calculate the similarity between the second state data and the first state data at the same time, and to determine whether there is a first state data with a similarity greater than a first threshold. The judgment module is used to make the following judgments: if they exist, the digital twin model corresponding to the first state data with the highest similarity is used to guide the adjustment of the state data at the next moment; if they do not exist, the deviation value between the second state data at the current moment and the first state data at the same moment is calculated, and the deviation value is used as the adjustment guide for the state data at the next moment. The equipment control module is used to call the control instructions based on the control guidance to calculate the state data of the next moment from the large model, and to control the operation of the relevant equipment based on the control instructions.

Images