Search

CN-117744188-B - Automatic model parameter selection method and system based on hidden Markov

CN117744188BCN 117744188 BCN117744188 BCN 117744188BCN-117744188-B

Abstract

The application discloses a hidden Markov-based model parameter automatic selection method and a hidden Markov-based model parameter automatic selection system, wherein the method comprises the following steps of responding to a selected instruction input by a user in a visual operation interface to determine a selected area in a model and constructing a selected area set; the method comprises the steps of constructing an implicit state set, wherein the implicit state set comprises adjustable parameters of a model, acquiring a state transition probability matrix and an observation probability matrix, and calculating an implicit state sequence by using a Viterbi algorithm based on the selected area and the acquired state transition probability matrix and observation probability matrix to obtain a predicted parameter combination. According to the technical scheme, expert experience is solidified in the state transition matrix and the observation probability matrix based on the hidden Markov algorithm, and when a user selects certain areas, parameter combination is automatically predicted and prompted to guide the user to use, so that experience thresholds used by the user are reduced, and efficiency of parameter selection of technicians is improved.

Inventors

  • LI YONGSHENG
  • SHI KAI
  • GU XIAO

Assignees

  • 上海概伦电子股份有限公司

Dates

Publication Date
20260508
Application Date
20231201

Claims (10)

  1. 1. The automatic model parameter selection method based on the hidden Markov is characterized by comprising the following steps of: Responding to a selected instruction input by a user in a visual operation interface to determine a selected area in the semiconductor device model, and constructing a selected area set; Constructing an implicit state set, wherein the implicit state set comprises adjustable parameters of the semiconductor device model; Acquiring a state transition probability matrix and an observation probability matrix solidified by expert experience; and calculating an implicit state sequence by using a Viterbi algorithm based on the selected area and the acquired state transition probability matrix and observation probability matrix to obtain a predicted parameter combination.
  2. 2. The method of automatic selection of hidden markov based model parameters of claim 1, wherein constructing the set of selected regions in response to a selected instruction entered by a user in the visual operation interface to determine the selected regions in the model further comprises: Acquiring attribute information of a characterization graph based on the characterization graph selected by a cursor in a current visual operation interface; Acquiring parameter information of all points in a marked state in the characterization graph; Determining a corresponding selected area in the model based on attribute information of the characterization graph and parameter information of all points in a marked state in the characterization graph; a set of selected regions v= { V 1 ,v 2 ,…,v m }, m representing the number of possible observations is constructed.
  3. 3. The automatic selection method of model parameters based on hidden markov according to claim 1, wherein constructing a set of hidden states, wherein the set of hidden states includes adjustable parameters of the model, and further includes: Constructing an implicit state set Q= { Q 1 ,q 2 ,…,q n }, wherein n represents a possible state number; The model employs one of BSIM3, BSIM4, BSIM6, BSIM-CMG, BSIM-IMG, BSIMSOI, UTSOI, hiSIM2, hiSIM _ HV, PSP, GP-BJT, RPITFT, each state q n in the implicit state set corresponding to an adjustable parameter in the model.
  4. 4. A method for automatically selecting parameters of a hidden markov based model according to claim 1 or 3, wherein constructing a set of hidden states, wherein the set of hidden states includes adjustable parameters of the model further includes: Each state q n in the implicit state set corresponds to one adjustable parameter or a combination of a plurality of adjustable parameters, and the adjustable parameters are adjustable parameters corresponding to the model.
  5. 5. The method of claim 1, further comprising, after computing the implicit state sequences using a viterbi algorithm, performing a de-duplication process on the implicit state sequences to obtain the predicted parameter combinations.
  6. 6. The method for automatically selecting model parameters based on hidden Markov as set forth in claim 1, wherein the state transition probability matrix A is A= [ a ij ] N×N , Where a ij ＝P(I t+1 ＝q j |I t ＝q i ),i＝1,2,…,N;j＝1,2,…,N;a ij represents the probability that t is in the q i state and t+1 transitions to the q j state.
  7. 7. The method for automatically selecting model parameters based on hidden Markov as set forth in claim 6, wherein the observation probability matrix B is B= [ B j (k)] N×M , Where b j (k)＝P(o t ＝v k |I t ＝q j ),k＝1,2…,M;j＝1,2,…,N;b j (k) represents the probability that the state of q j at time t generates the observation v k , also called the generation probability and the emission probability.
  8. 8. The method of claim 7, wherein computing an implicit state sequence using a viterbi algorithm based on the selected region and the obtained state transition probability matrix and observation probability matrix to obtain a predicted parameter combination further comprises: determining an observation sequence o= (O 1 ,o 2 ,…o T ) based on the selected regions, wherein T is the number of selected regions; And calculating r hidden state sequences possibly generating an observation event sequence by adopting a Viterbi algorithm in combination with the state transition probability matrix A, the observation probability matrix B and the number T of the selected areas so as to obtain a predicted r group parameter combination.
  9. 9. A hidden markov based model parameter automatic selection system comprising: the selected area acquisition module is used for responding to a selected instruction input by a user in the visual operation interface to determine a selected area in the semiconductor device model and constructing a selected area set; The parameter combination construction module is used for constructing an implicit state set, wherein the implicit state set comprises adjustable parameters of the semiconductor device model; The matrix acquisition module is used for acquiring a state transition probability matrix and an observation probability matrix solidified by expert experience; and the prediction module is used for calculating an implicit state sequence by using a Viterbi algorithm based on the selected area, the acquired state transition probability matrix and the acquired observation probability matrix to obtain a predicted parameter combination.
  10. 10. An electronic device, comprising: a memory for storing a processing program; a processor, which when executing the processing program, implements the automatic selection method of model parameters based on hidden markov according to any one of claims 1 to 8.

Description

Automatic model parameter selection method and system based on hidden Markov Technical Field The invention belongs to the technical field of integrated circuit computer aided design, and particularly relates to a method and a system for automatically selecting model parameters based on hidden Markov. Background The continued development of semiconductor and integrated circuit technology has made the importance of integrated circuit Computer Aided Design (CAD) or Electronic Design Automation (EDA) platforms increasingly prominent. One fundamental function of the EDA platform is the extraction of parameters of a device model, i.e., the extraction of model parameters of a semiconductor device manufactured in a particular integrated circuit process based on some standard device model. After the model parameters are extracted, various operating characteristics of the semiconductor device can be mathematically depicted in conjunction with the corresponding standard device model for device simulation in subsequent circuit design. The BSIM model is a metal oxide field effect transistor (MOSFET) model developed by berkeley division, university of california, usa, which is suitable for digital and analog circuit design and simulation. In an actual parameter extraction operation, the model parameters of the MOSFET device can be extracted by selecting various BSIM models (e.g., BSIM4, BSIM-Bulk, BSIM-CMG, etc.) corresponding to the actual MOSFET device to process the test data (e.g., I-V curves C-V curves of MOSFETs of different sizes, etc.) of the MOSFET device. In the prior art, various device model parameters are extracted by firstly selecting a selected area to be fitted, then manually inputting parameters according to experience, and then modifying the parameters through an optimization algorithm so as to fit the selected area. However, this method of selecting parameters is too empirical to be suitable for inexperienced persons. Disclosure of Invention In order to solve the problems, the invention aims to provide a hidden Markov-based automatic model parameter selection method and a hidden Markov-based automatic model parameter selection system, which solidify expert experience on a state transition matrix and an observation probability matrix based on a hidden Markov algorithm, and automatically predict and prompt parameter combinations after a user selects certain areas to guide the user to use, reduce experience thresholds used by the user and improve the parameter selection efficiency of technicians. The first technical scheme provided by the application is that the automatic model parameter selection method based on the hidden Markov comprises the following steps of responding to a selected instruction input by a user in a visual operation interface to determine a selected area in a model, constructing a selected area set, constructing an implicit state set, wherein the implicit state set comprises adjustable parameters of the model, acquiring a state transition probability matrix and an observation probability matrix, and calculating an implicit state sequence by adopting a Viterbi algorithm based on the selected area and the acquired state transition probability matrix and the observation probability matrix to obtain a predicted parameter combination. Preferably, the selected area set is constructed by responding to a selected instruction input by a user in a visual operation interface to determine a selected area in the model, wherein the selected area set further comprises the steps of acquiring attribute information of a characteristic graph selected by a cursor in the visual operation interface, acquiring parameter information of all points in a marked state in the characteristic graph, determining a corresponding selected area in the model based on the attribute information of the characteristic graph and the parameter information of all points in the marked state in the characteristic graph, and constructing the selected area set V= { V 1,v2,…,vm }, wherein m represents a possible observation number. Preferably, constructing a set of implicit states, wherein the set of implicit states includes adjustable parameters of the model, further includes constructing a set of implicit states, Q= { Q 1,q2,…,qn }, n representing a number of possible states, the model including, but not limited to, BSIM3, BSIM4, BSIM6, BSIM-CMG, BSIM-IMG, BSIMSOI, UTSOI, hiSIM, hiSIM _ HV, PSP, GP-BJT, RPITFT, each state Q n in the set of implicit states corresponding to an adjustable parameter in the model. Preferably, constructing an implicit state set, wherein the implicit state set contains adjustable parameters of the model, and further comprises obtaining each state q n in the implicit state set to correspond to one adjustable parameter or a combination of a plurality of adjustable parameters, and the adjustable parameters are adjustable parameters corresponding to the model. Preferably, the method further comprises, after