Radiation effect integration method, system and device of SPICE model and medium
By extracting radiation-sensitive parameters from the SPICE model and establishing a mapping relationship using a neural network, the SPICE model is updated to include radiation effects. This solves the problem of insufficient simulation accuracy of existing models under radiation environments, and achieves reliability and cost-effectiveness in circuit design.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HARBIN INST OF TECH
- Filing Date
- 2025-12-31
- Publication Date
- 2026-04-10
AI Technical Summary
Existing SPICE models fail to effectively consider radiation effects, resulting in insufficient reliability of circuit designs in space environments and an inability to meet simulation accuracy requirements under radiation conditions.
By extracting radiation-sensitive parameters from the original SPICE model of the target device, electrical characteristic data under different radiation conditions are obtained. A mapping relationship between radiation conditions and radiation-sensitive parameters is established using a neural network, and the SPICE model is updated to include radiation effects.
It improves the accuracy of circuit simulation, enabling accurate simulation of device performance changes under radiation conditions, reducing circuit design and testing costs, and improving circuit reliability.
Smart Images
Figure CN121835561A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of circuit simulation, in particular to a radiation effect integration method, system, device and medium for a SPICE model. BACKGROUND
[0002] With the rapid development of the aerospace field, especially the rise of commercial aerospace, the reliability of integrated circuits in space environment has become a very important research content. High-energy particles in the space radiation environment greatly affect the reliability of electronic components. For example, total dose effect causes oxide layer trap charge accumulation, which causes MOSFET threshold voltage drift and increases leakage current; single event effect may trigger a transient current spike, causing circuit function failure or even permanent damage. Therefore, accurately characterizing the radiation effect of devices during the circuit design phase is a core prerequisite for ensuring the reliability of integrated circuits in special environments.
[0003] In related technologies, as a core part of PDK (Process Design Kit), the physical effects and their accuracy that the SPICE model can contain directly determine the accuracy of circuit design. The existing SPICE model in the PDK is mainly for the conventional environment, only considering the non-ideal effects brought by temperature, voltage, and size effects, without considering radiation effects. The existing PDK technology cannot meet the circuit design in the radiation environment, but rebuilding a SPICE model based on the existing model framework that includes radiation effects will face huge mathematical and physical mechanism challenges, and if a data-driven approach is used, the existing PDK technology cannot be applied. SUMMARY
[0004] The problem solved by the present application is how to improve the circuit simulation accuracy and reliability of the SPICE model in the radiation environment.
[0005] To solve the above problems, the present application provides a radiation effect integration method, system, device and medium for a SPICE model.
[0006] In a first aspect, the present application provides a radiation effect integration method for a SPICE model, comprising: extracting model parameters affected by radiation from an original SPICE model corresponding to a target device as radiation-sensitive parameters; obtaining electrical characteristic data of the target device under different radiation conditions, and extracting the numerical value of the radiation-sensitive parameters under each radiation condition based on the electrical characteristic data to obtain correlation data between different radiation conditions and the numerical value of the radiation-sensitive parameters; based on the correlation data, using a neural network model to establish a mapping relationship between the radiation conditions and the numerical value of the radiation-sensitive parameters; substituting the mapping relationship into the original SPICE model to update the values of the radiation-sensitive parameters, to obtain an enhanced SPICE model containing radiation effects.
[0007] Optionally, the extracting of the model parameters affected by radiation as the radiation-sensitive parameters comprises: obtaining a macroscopic electrical characteristic degradation phenomenon of the target device under a radiation environment; determining a model expression describing the phenomenon in the original SPICE model based on the macroscopic electrical characteristic degradation phenomenon; identifying a model parameter having a physical explanation mechanism from the model expression as the radiation-sensitive parameter according to a radiation effect degradation mechanism of the target device.
[0008] Optionally, the extracting of the model parameters affected by radiation as the radiation-sensitive parameters comprises: obtaining a specific physical process of electrical characteristic degradation of the target device according to a radiation degradation physical mechanism of the target device under the radiation environment; selecting a model parameter used to represent the specific physical process from the original SPICE model as the radiation-sensitive parameter.
[0009] Optionally, the obtaining of the electrical characteristic data of the target device under different radiation conditions and the extracting of the values of the radiation-sensitive parameters under each radiation condition based on the electrical characteristic data to obtain correlation data between different radiation conditions and the values of the radiation-sensitive parameters comprises: obtaining first electrical characteristic data of the target device under a non-radiation condition, and extracting reference values of all model parameters in the original SPICE model based on the first electrical characteristic data; irradiating the target device under each radiation condition and obtaining second electrical characteristic data under each radiation condition respectively; for each radiation condition, performing a parameter extraction operation: setting the values of the model parameters other than the radiation-sensitive parameters in the original SPICE model to the reference values, and running the original SPICE model based on a set of values of the radiation-sensitive parameters under the current radiation condition to obtain a simulation result; fitting the simulation result to the second electrical characteristic data by optimizing the values of the radiation-sensitive parameters, and determining the values of the radiation-sensitive parameters after fitting as the values of the radiation-sensitive parameters under the current radiation condition; The correlation data is formed by associating each radiation condition with the value of the corresponding radiation-sensitive parameter.
[0010] Optionally, establishing the mapping relationship between the radiation conditions and the values of the radiation-sensitive parameters using a neural network model based on the associated data includes: The radiation conditions are used as input features, and the values of the radiation sensitivity parameters corresponding to the radiation conditions are used as output targets to construct a training dataset. The training dataset is trained using an initial neural network, and a quantitative functional relationship between the input features and the output target is obtained from the trained initial neural network, thus forming the mapping relationship.
[0011] Optionally, substituting the mapping relationship into the original SPICE model to update the value of the radiation sensitivity parameter to obtain an enhanced SPICE model that includes radiation effects includes: Define the quantitative functional relationship as a new parameter source for the radiation-sensitive parameter; When simulating the behavior of the target device under the radiation conditions, the radiation conditions are input into the quantitative function relationship to obtain the corresponding values of the radiation-sensitive parameters; The enhanced SPICE model is formed by replacing the original defined values of the corresponding parameters in the original SPICE model with the values of the radiation-sensitive parameters.
[0012] Optionally, the radiation effect integration method of the SPICE model further includes: selecting appropriate radiation conditions according to actual modeling requirements and / or actual radiation resistance circuit index requirements, and using the radiation conditions as input features. The input features include: the cumulative fluence of each radiation condition, the total ionizing dose of each radiation condition, the cumulative fluence of the radiation condition, and the average dose rate or dose rate distribution statistics characterizing the change of radiation intensity over time.
[0013] Secondly, the integrated system for the radiation effect of the SPICE model of the present invention includes: The extraction unit is used to extract the radiation-affected model parameters as radiation-sensitive parameters from the original SPICE model corresponding to the target device. The data processing unit is used to acquire electrical characteristic data of the target device under different radiation conditions, and extract the value of the radiation sensitive parameter under each radiation condition based on the electrical characteristic data, so as to obtain the correlation data between different radiation conditions and the value of the radiation sensitive parameter. The mapping relationship establishment unit is used to establish a mapping relationship between the radiation conditions and the values of the radiation-sensitive parameters based on the associated data and using a neural network model; an updating unit configured to substitute the mapping relationship into the original SPICE model to update the values of the radiation-sensitive parameters, and obtain an enhanced SPICE model containing radiation effects.
[0014] In a third aspect, a computer device of the present application comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the radiation effect integration method of the SPICE model when executing the computer program.
[0015] In a fourth aspect, a computer readable storage medium of the present application stores a computer program, and the computer program is executable on a processor to implement the radiation effect integration method of the SPICE model.
[0016] The radiation effect integration method, system, device and medium of the SPICE model of the present application can extract radiation-sensitive parameters from an original SPICE model, and update these parameters based on electrical characteristic data under different radiation conditions, so that the enhanced SPICE model can accurately reflect the influence of radiation on device performance. For example, the variation rule of the threshold voltage of a MOSFET under different radiation doses obtained from experiments can be integrated into the SPICE model, so that the drift of the threshold voltage of the MOSFET under a specific radiation environment can be accurately simulated during simulation, and the accuracy of circuit simulation is greatly improved. By obtaining electrical characteristic data of the target device under a plurality of different radiation conditions, a wide mapping relationship between the radiation conditions and the values of the radiation-sensitive parameters can be established, that is, the enhanced SPICE model can flexibly adapt to various radiation environments. For example, for a satellite application of an integrated circuit at different orbital altitudes, that is, corresponding to different radiation intensities, the SPICE model obtained by the present application can accurately simulate the performance variation of the devices in the circuit at each orbital altitude, providing a reliable basis for circuit design, and avoiding the problem of inaccurate simulation caused by changes in radiation conditions. Moreover, when obtaining electrical characteristic data, the target device under different batches or different manufacturing process conditions can be tested, and in this way, more comprehensive correlation data of radiation-sensitive parameters and radiation conditions can be obtained, and then the mapping relationship is established by using a neural network model, and these individual difference factors are considered comprehensively. For example, for MOSFET devices of the same batch but different positions, their electrical characteristics may be different after radiation. The enhanced SPICE model of the present application can better reflect the performance of such individual differences in the radiation environment, so as to more accurately simulate the behavior of each device in the real circuit in circuit simulation.
[0017] In the circuit design stage, simulation is performed by using the enhanced SPICE model containing radiation effects, so that potential radiation failure risks, such as transient current spikes caused by single event effects and their propagation paths and influence range, can be found in advance. In view of the functional failure caused by the current spikes that may occur at key circuit nodes, the designer can take measures in advance, including increasing redundancy design, optimizing circuit layout or selecting more radiation-resistant devices, so as to maximize the reduction of failure probability caused by radiation before the circuit is put into space application, thereby improving the reliability of the circuit. At the same time, the enhanced SPICE model provides strong support for the optimization of circuit design parameters. The designer can reasonably select device size, working voltage and other parameters according to the radiation effect information provided by the model. In addition, the enhanced SPICE model can effectively reduce the cost and time of circuit testing. Traditional circuit reliability testing needs a large number of experiments in the actual radiation environment, which is high in cost and time-consuming. The enhanced SPICE model can evaluate the circuit at the simulation stage, screen out unreliable or substandard circuit design schemes, and reduce the number and range of subsequent actual radiation tests. BRIEF DESCRIPTION OF DRAWINGS
[0018] Figure 1 A flowchart of the radiation effect integration method of the SPICE model of the embodiment of the present application is shown in the figure. Figure 2 A comparison diagram of simulation results and test results of the model electrical characteristics of the embodiment of the present application is shown in the figure. Figure 3 A structure diagram of the radiation effect integration system of the SPICE model of the embodiment of the present application is shown in the figure. Figure 4 A structure diagram of the electronic device of the embodiment of the present application is shown in the figure. DETAILED DESCRIPTION
[0019] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings. Although some embodiments of the present application are shown in the drawings, it should be understood that the present application can be implemented in various forms, and should not be interpreted as being limited to the embodiments described herein, on the contrary, these embodiments are provided to make the present application more thorough and complete. It should be understood that the drawings and embodiments of the present application are only for illustrative purposes, and are not intended to limit the scope of protection of the present application.
[0020] It should be understood that each step described in the method embodiments of the present application can be executed in different order and / or in parallel. In addition, the method embodiments can include additional and / or omitted execution steps. The scope of the present application is not limited in this respect.
[0021] The term "include," and variations thereof, as used in this document, is open-ended and means "comprise" or "comprises" but not limited to; the term "based on" means "based, at least in part, on"; the term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments"; the term "optional" means "optional in at least one embodiment". Related terms shall be construed accordingly. It is to be noted the use of "first", "second", "third", etc. in the present document does not connote any order, quantity, or importance, but rather are used to distinguish one set of elements from another, and are used arbitrarily and merely for the sake of convenience.
[0022] It should be noted that the terms "one", "multiple", "a number of", etc. mentioned in the present application are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise explicitly stated in the context, it should be understood as "one or more".
[0023] In combination Figure 1 As shown, the embodiment of the present application provides a radiation effect integrated method of SPICE model, comprising: From the original SPICE model corresponding to the target device, the model parameters affected by radiation are extracted as radiation sensitive parameters.
[0024] Specifically, by studying the physical degradation mechanism of the target device (such as AlGaN / GaN HEMT) in the radiation environment, it is found that the reverse gate current is significantly degraded under proton radiation, while the forward current is basically unchanged. Then, a suitable SPICE model (such as ASM-HEMT model) is selected, and the model expression is carefully analyzed to determine the parameters closely related to the radiation effect, such as RIGDDIO, EBREAKD and RNJGD, which directly affect the degradation behavior of the gate current. Then, by performing parameter extraction under unirradiated and multiple irradiation conditions, the values of the radiation sensitive parameters are obtained by using curve fitting method, and the accuracy is verified by comparing with the experimental data. Finally, the sensitivity analysis is performed to evaluate the contribution degree of these parameters to the change of the electrical characteristics of the device, and it is ensured that the selected parameters can accurately reflect the change of the characteristics of the device in the radiation environment, which provides a key basis for subsequent construction of enhanced SPICE model containing radiation effect. Taking AlGaN / GaN HEMT device as an example, according to the gate current degradation phenomenon of the device under proton radiation, the radiation sensitive parameters RIGDDIO, EBREAKD and RNJGD are determined, which are closely related to the change of the electrical characteristics of the device in the radiation environment and are the key factors affecting the performance of the device. By analyzing the physical mechanism and radiation degradation mechanism of the device, and combining with the change of the electrical characteristics of the device under different radiation conditions, these radiation sensitive parameters are accurately identified, which lays a foundation for subsequent improvement of the model.
[0025] Obtaining electrical characteristic data of the target device under different radiation conditions, and extracting a value of the radiation-sensitive parameter under each of the radiation conditions based on the electrical characteristic data, to obtain correlation data between the different radiation conditions and the value of the radiation-sensitive parameter.
[0026] Specifically, a series of experimental tests under different radiation conditions are performed on a target device, such as an AlGaN / GaN HEMT, including electrical characteristic measurements under non-irradiation conditions and different proton fluences, such as , to obtain the variation law of key parameters such as gate current and threshold voltage. Through experimental observation, it is found that as the proton fluence increases, the reverse gate current of the device significantly degrades, while the forward current remains basically unchanged, providing a basis for determining the radiation-sensitive parameter. Based on the experimental data, using parameter extraction methods such as curve fitting, the specific values of radiation-sensitive parameters closely related to the electrical characteristic changes, such as RIGDDIO, EBREAKD and RNJGD, are extracted for each radiation condition, and a correlation database between different radiation conditions and these parameter values is constructed. These correlation data not only record the electrical characteristic changes of the device in the radiation environment, but also provide key inputs for subsequent use of neural networks to establish a mapping relationship, ensuring that the enhanced SPICE model can accurately reflect the real behavior of the device under radiation conditions, thereby providing reliable support for circuit design in radiation environments.
[0027] Based on the correlation data, a neural network model is used to establish a mapping relationship between the radiation conditions and the values of the radiation-sensitive parameters.
[0028] Specifically, based on the previously constructed data table containing the correlation between different radiation conditions and the values of the radiation-sensitive parameters, an artificial neural network (ANN) technique is used to establish a mapping relationship between the two. Taking an AlGaN / GaN HEMT device as an example, different proton fluences such as are taken as inputs, and the specific values of corresponding radiation-sensitive parameters such as RIGDDIO, EBREAKD and RNJGD are taken as outputs, and the neural network is trained. The neural network learns the complex nonlinear relationship in these data samples, thereby obtaining a series of mapping functions.
[0029] In a preferred embodiment of the present application, the ANN technique is used to construct the relationship between the radiation-sensitive parameters and the radiation conditions, and the explicit expression between the radiation-sensitive parameters and the radiation bias conditions is constructed by extracting the weights and biases, as shown in expression (1): ; (1) These mapping functions can map any given radiation condition, such as proton fluence, to the corresponding radiation-sensitive parameter values, thereby providing the ability to dynamically update the parameters of the subsequent enhanced SPICE model. In addition, in the embodiments of the present application, in order to improve the generalization ability and prediction accuracy of the model, a double data augmentation strategy is also adopted, including: introducing controllable perturbations by Gaussian noise injection to simulate data uncertainty, as shown in expression (2): ; (2) wherein and represent the enhanced input and output data, respectively, and are Gaussian noise, denotes the variance of the Gaussian noise. Intermediate samples are generated in adjacent radiation conditions through linear interpolation techniques, as shown in expression (3): ; (3) wherein and represent the interpolated data, is the interpolation coefficient, and represent the two adjacent original input data points, and correspond to the two original output data points, thereby effectively increasing the diversity and quantity of the data, so that the neural network model can more accurately predict the values of radiation-sensitive parameters when facing new, unseen radiation conditions.
[0030] Substitute the mapping relationship into the original SPICE model to update the values of the radiation-sensitive parameters, and obtain an enhanced SPICE model containing radiation effects.
[0031] Specifically, the mapping relationship obtained by the neural network, such as the mapping function in expression (1), is substituted into the original SPICE model of the target device, thereby updating the values of the radiation-sensitive parameters.
[0032] Taking the ASM-HEMT model of the AlGaN / GaN HEMT device as an example, the expression describing the gate current in the original model includes: ; (4) ; (5) ; (6) wherein denotes the Poole-Frenkel component of the gate current, L represents the length of the device, NF represents a factor related to the device structure, represents one of the radiation sensitive parameters related to the Poole-Frenkel component of the gate current, represents the gate-drain voltage, represents one of the radiation sensitive parameters related to the Poole-Frenkel component of the gate current, represents the Boltzmann constant, represents the operating temperature of the device, represents the electric field strength, represents the gate bar width, represents the total gate current, RIGDDIO represents one of the radiation sensitive parameters related to the degradation of the gate current, EBREAKD represents one of the radiation sensitive parameters related to the degradation of the gate current, RNJGD represents one of the radiation sensitive parameters related to the Poole-Frenkel component of the gate current, as involved in the expressions (4), (5), (6) for the radiation sensitive parameters, such as RIGDDIO, EBREAKD, RNJGD. Replace these parameters with the corresponding expressions output by the neural network model, that is, replace with , with , with , wherein ϕ represents the proton fluence, , , represents the mapping function output by the neural network model, corresponding to the radiation sensitive parameters RIGDDIO, EBREAKD, RNJGD respectively. In this way, the enhanced SPICE model can dynamically update the values of the radiation sensitive parameters according to the input radiation conditions, such as the proton fluence, so as to accurately reflect the changes in the electrical characteristics of the device under different radiation conditions.
[0033] As shown in Figure 2 , the embodiment of the present application is verified by simulation, and the fitting error between the simulation results of the enhanced SPICE model under different proton fluences and the experimental data is within 3%, which proves the high precision and good applicability of the enhanced SPICE model.
[0034] The radiation effect integrated method of the SPICE model of the embodiment can accurately reflect the influence of radiation on the performance of a device by extracting radiation-sensitive parameters from an original SPICE model and updating the parameters based on electrical characteristic data under different radiation conditions. For example, the variation of the threshold voltage of a MOSFET under different radiation doses can be obtained through experiments, and the variation can be integrated into the SPICE model, so that the drift of the threshold voltage of the MOSFET under a specific radiation environment can be accurately simulated during simulation, and the accuracy of circuit simulation is greatly improved. By obtaining electrical characteristic data of a target device under a plurality of different radiation conditions, a wide mapping relationship between radiation conditions and the values of radiation-sensitive parameters can be established, that is, the enhanced SPICE model can be flexibly adapted to various radiation environments. For example, for a satellite application of an integrated circuit at different orbital altitudes, that is, corresponding to different radiation intensities, the SPICE model obtained by the embodiment can accurately simulate the performance variation of the devices in the circuit at each orbital altitude, providing a reliable basis for circuit design and avoiding inaccurate simulation caused by changes in radiation conditions. Moreover, when obtaining the electrical characteristic data, the target devices under different batches or different manufacturing process conditions can be tested. In this way, more comprehensive correlation data of radiation-sensitive parameters and radiation conditions can be obtained, and then the mapping relationship can be established by using a neural network model, considering these individual difference factors. For example, for MOSFET devices of the same batch but different positions, their electrical characteristics may be different after radiation. The enhanced SPICE model of the application can better reflect the performance of the individual differences in the radiation environment, so as to more accurately simulate the behavior of each device in the real circuit in circuit simulation.
[0035] During the circuit design stage, the enhanced SPICE model containing radiation effects is used for simulation, which can discover potential radiation failure risks in advance, such as transient current spikes caused by single event effects and their propagation paths and influence range. In view of the functional failure caused by the current spikes that may occur at key circuit nodes, the designer can take measures in advance, including increasing redundancy design, optimizing circuit layout, or selecting more radiation-resistant devices, so as to maximize the reduction of failure probability caused by radiation before the circuit is put into space application, and improve the reliability of the circuit. At the same time, the enhanced SPICE model provides strong support for circuit design parameter optimization. The designer can reasonably select device size, working voltage and other parameters according to the radiation effect information provided by the model. In addition, the enhanced SPICE model can effectively reduce the cost and time of circuit testing. Traditional circuit reliability testing needs a large number of experiments in actual radiation environment, which is costly and time-consuming. The enhanced SPICE model can evaluate the circuit at the simulation stage, screen out unreliable or substandard circuit design schemes, and reduce the number and range of subsequent actual radiation tests.
[0036] Optionally, the model parameters affected by radiation are extracted from the original SPICE model of the target device as radiation-sensitive parameters, including: Obtaining the macroscopic electrical characteristic degradation phenomenon of the target device in a radiation environment; Based on the macroscopic electrical characteristic degradation phenomenon, determining the model expression in the original SPICE model that describes the phenomenon; According to the radiation effect degradation mechanism of the target device, identifying the model parameters with physical interpretation mechanism from the model expression as the radiation-sensitive parameters.
[0037] Specifically, first, the macroscopic electrical characteristic degradation phenomenon is obtained by implementing experiments in a radiation environment, such as exposing the device to a specific type of radiation source, such as protons, electrons, X-rays, etc., measuring the electrical characteristics of the device before and after radiation, including but not limited to threshold voltage, leakage current, gate current, etc. Key parameters, record and analyze data, identify the changes in electrical characteristics caused by radiation, such as current degradation, voltage drift, etc.
[0038] According to the observed electrical characteristic degradation phenomenon in the experiment, and the physical model and SPICE model library of the device, find the model expression that describes these phenomena, for example, if the observed is the degradation of the gate current, that is, the model expression related to the current flow is selected, such as the Poole-Frankel (PF) emission model. According to the determined related model expression, analyze the physical meaning of each parameter in these expressions and the influence on the electrical characteristics, according to the radiation effect degradation mechanism of the target device, identify and select the parameters directly related to the radiation effect as the radiation-sensitive parameters, wherein the radiation effect degradation mechanism of the target device refers to the specific process that after high-energy particles such as protons or heavy ions enter the semiconductor device, the internal microstructure is physically damaged, and eventually leads to irreversible degradation of the macroscopic electrical performance of the device, such as current, voltage, switching speed. At the same time, in this embodiment, these parameters have clear physical basis, such as parameters related to radiation-induced trap charge or carrier recombination. For example, in an AlGaN / GaN HEMT device, the parameters related to the PF emission model are selected because they directly affect the calculation of the gate current and have a direct relationship with the degradation phenomenon caused by radiation.
[0039] In this optional embodiment, the degradation phenomena of the target device's macroscopic electrical characteristics in the radiation environment are obtained, and based on these phenomena, the corresponding model expressions in the original SPICE model are determined, and then the model parameters with clear physical mechanism interpretation are selected from these expressions as the radiation sensitive parameters, thereby realizing the accurate radiation effect modeling of the SPICE model. Not only the prediction accuracy of the model for the behavior of the device in the radiation environment is improved, but also by selecting the parameters closely related to the physical phenomena, the physical credibility and generalization ability of the model are enhanced. In addition, this parameter selection method based on physical mechanism of the embodiment helps to simplify the complexity of the model while maintaining the sensitivity of the model to key radiation effects, thereby improving the practicability of the model and the reliability of circuit design without sacrificing accuracy.
[0040] Optionally, the model parameters affected by radiation are extracted as radiation sensitive parameters, comprising: According to the radiation degradation physical mechanism of the target device in the radiation environment, the specific physical process of the electrical characteristic degradation of the target device is obtained; From the original SPICE model, the model parameters used to characterize the specific physical process are selected as the radiation sensitive parameters.
[0041] Specifically, by placing the target device in a controlled radiation environment, such as using radiation generated by a particle accelerator to simulate the space radiation environment, the change of the electrical characteristics of the device with the increase of the radiation dose is observed, and the specific mode of device performance degradation, such as threshold voltage drift or leakage current increase, is revealed through experimental data. Then, the physical mechanism is analyzed in combination with the physical structure of the device and the theory of radiation effects. For example, if the leakage current is observed to increase, it may be related to the accumulation of radiation-induced oxide trap charges. In this way, the specific physical process that causes the degradation of electrical characteristics can be determined, providing a basis for subsequent model parameter selection.
[0042] After the specific physical process affecting the degradation of the electrical characteristics of the device is determined, the parameters in the original SPICE model that can characterize these physical processes are identified and selected, and based on the physical meaning of each parameter in the SPICE model, the correlation between them and the determined physical process is evaluated. For example, if the degradation process is related to trap charge accumulation, the parameter in the model that describes the effect of trap charge may be a radiation sensitive parameter. Through parameter sensitivity analysis, the influence degree of these parameters on the simulation results can be further verified to ensure that the selected parameters can effectively reflect the radiation effects. Finally, these parameters are used to update the SPICE model, so that it can more accurately simulate the behavior of the device in the radiation environment.
[0043] In the optional embodiment, by associating the electrical characteristic degradation of the target device in the radiation environment with specific physical processes, and selecting key parameters capable of representing these physical processes from the original SPICE model as radiation-sensitive parameters, important enhancement of the SPICE model is realized, which not only improves the prediction accuracy of the model for the performance degradation of the device in the radiation environment, but also enhances the physical credibility and generalization ability of the model due to the parameter selection based on physical mechanisms. In addition, this method reduces the dependence on a large amount of experimental data, reduces the research and development cost and time, and ensures the effectiveness of the selected parameters through parameter sensitivity analysis, thereby improving the practicability of the model and the reliability of circuit design.
[0044] Optionally, the electrical characteristic data of the target device under different radiation conditions is obtained, and the value of the radiation-sensitive parameter under each radiation condition is extracted based on the electrical characteristic data, to obtain association data between different radiation conditions and the value of the radiation-sensitive parameter, including: The first electrical characteristic data of the target device under a non-radiation condition is obtained, and based on the first electrical characteristic data, the reference value of all model parameters in the original SPICE model is extracted; Under each radiation condition, the target device is irradiated, and the second electrical characteristic data under each radiation condition is obtained respectively; For each radiation condition, a parameter extraction operation is performed: the values of the other model parameters in the original SPICE model except the radiation-sensitive parameters are set to the reference values, and based on the value of a set of radiation-sensitive parameters under the current radiation condition, the original SPICE model is run to obtain a simulation result; By optimizing the value of the radiation-sensitive parameter, the simulation result is fitted with the second electrical characteristic data, and the fitted value of the radiation-sensitive parameter is determined as the value of the radiation-sensitive parameter under the current radiation condition; Each radiation condition is associated with the corresponding value of the radiation-sensitive parameter to form the association data.
[0045] Specifically, first, multiple target device samples are prepared to ensure the statistical significance of the data, and these devices are subjected to detailed electrical property tests in a radiation-shielded environment, recording key parameters such as threshold voltage, leakage current, transconductance, etc. These tests are performed before the devices are affected by any radiation to ensure that the electrical property data obtained reflects the initial state of the devices. Based on these first electrical property data, the reference values of all model parameters are extracted from the original SPICE model using parameter extraction tools or methods, such as through curve fitting, optimization algorithms, or other numerical analysis techniques, to ensure that the extracted parameter values can accurately reproduce the behavior of the devices when not affected by radiation. The appropriate radiation source and irradiation conditions, such as specific particle types, energies, and dose rates, are selected to simulate the radiation environment that the devices may encounter in actual applications. The device samples are exposed to these radiation conditions, and electrical property tests are performed immediately after irradiation to capture the immediate effects of radiation on device performance. These tests should be performed under similar or identical conditions to the irradiation environment to reduce the impact of environmental changes on test results, and the second electrical property data obtained will reflect the performance changes of the devices under different radiation conditions, providing important information for subsequent parameter extraction and model updating.
[0046] For each radiation condition, use the previously extracted baseline parameter values as a starting point, while leaving the values of the radiation-sensitive parameters to be optimized. Run the original SPICE model, input the electrical characteristic test data under the current radiation condition, while setting the values of other model parameters except the radiation-sensitive parameters to the baseline values. Adjust the values of the radiation-sensitive parameters to make the model output match the experimentally observed second electrical characteristic data as much as possible. Define a target function, such as mean square error (MSE), to quantify the difference between the simulation results and the experimental data, and use optimization algorithms to adjust the values of the radiation-sensitive parameters to minimize the target function. For example, automatically adjust algorithm parameters such as learning rate, iteration times, etc. to improve the efficiency and accuracy of the optimization process. In the optimization process, run the SPICE model multiple times to evaluate the simulation results under different parameter combinations, and select the parameter values that minimize the target function as the optimal solution. When the optimal radiation-sensitive parameter values for each radiation condition are found, these values are associated with the corresponding radiation conditions to form a correlation data set. These data sets can be used to train machine learning models such as neural networks to establish the mapping relationship between radiation conditions and radiation-sensitive parameters. In addition, the correlation data can also be used to verify and improve the accuracy of the SPICE model, by comparing the simulation results under different radiation conditions with the experimental data, evaluating the performance of the model, and adjusting and optimizing it as needed. In this way, the correlation data not only provides important training information for the model, but also provides a basis for the verification and improvement of the model. In the embodiment of the present application, the first electrical characteristic data is the baseline electrical characteristic data obtained when the target device is not affected by radiation, including the basic working parameters of the device, such as threshold voltage, drain current, transconductance, saturation current, etc. These data reflect the performance of the device under standard or ideal conditions, providing a reference baseline for subsequent radiation effect analysis. For example, for a MOSFET device, the first electrical characteristic data may include a threshold voltage of 2V and a drain current of 10nA measured in a radiation-free environment. The second electrical characteristic data is the electrical characteristic data obtained after the target device is treated under a specific radiation condition, reflecting the performance changes of the device after being affected by radiation, such as threshold voltage drift, drain current increase, etc., which are used to evaluate the specific effects of radiation on device performance. For example, after a certain dose of proton irradiation, the threshold voltage of the same MOSFET device may drift to 3V and the drain current may increase to 100nA.
[0047] In this optional embodiment, the accurate extraction and correlation of radiation-sensitive parameters are achieved by obtaining the first electrical characteristic data of the target device under non-irradiated conditions and extracting all reference parameter values of the original SPICE model, and then irradiating the device under different radiation conditions to collect the second electrical characteristic data. Using these data, the simulation results are fitted with the experimental data through parameter extraction operations and optimization processes to determine the specific values of the radiation-sensitive parameters of the device under each radiation condition. This not only improves the understanding of the performance changes of the device in the radiation environment, but also provides key information for establishing an enhanced SPICE model that can accurately reflect the behavior of the device under different radiation environments by correlating the radiation conditions with the values of the radiation-sensitive parameters. This model can provide circuit designers with a more reliable prediction tool to evaluate and optimize the performance and reliability of circuits in a radiation environment, thereby significantly improving the accuracy and efficiency of circuit design and reducing potential risks and costs caused by the radiation environment.
[0048] Optionally, the mapping relationship between the radiation conditions and the values of the radiation-sensitive parameters is established based on the correlation data using a neural network model, including: The radiation conditions are taken as input features, and the values of the radiation-sensitive parameters corresponding to the radiation conditions are taken as output targets to construct a training data set; The training data set is trained through an initial neural network, and a quantitative functional relationship between the input features and the output targets is obtained from the trained initial neural network to form the mapping relationship.
[0049] Specifically, first, the values of radiation-sensitive parameters under different radiation conditions are obtained. These values, obtained through the aforementioned parameter extraction operation, represent the performance changes of the device under specific radiation conditions. Next, the characteristics of the radiation conditions are defined, including radiation type (e.g., electrons, protons), radiation dose (e.g., Gray or Rad), and possible energy levels. These radiation conditions are used as input features, and the corresponding radiation-sensitive parameter values are used as output targets, organized into a structured data format, such as a CSV file or database table. Each row of data represents a specific radiation condition and its corresponding radiation-sensitive parameter value, thus forming the training dataset for training the neural network model. A suitable neural network architecture is selected, such as initializing network parameters (e.g., weights and biases) using a multilayer perceptron (MLP). These parameters will be optimized during training. The constructed training dataset is divided into a training set and a validation set for model training and performance evaluation. The neural network is trained using the training set data. Predictions are calculated through forward propagation, and then network parameters are adjusted through backpropagation to minimize the loss function (e.g., mean squared error or cross-entropy). At the end of each training epoch, the model performance is evaluated using a validation set to prevent overfitting, and model parameters or structure are adjusted accordingly. During training, hyperparameters such as learning rate and batch size are adjusted to improve training efficiency and model performance. Once the neural network is trained and its performance is verified to meet requirements, the trained model can be used to establish a mapping relationship between radiation conditions and radiation-sensitive parameter values. Specifically, the features of the radiation conditions are input into the trained neural network, and the network outputs predicted values of radiation-sensitive parameters. The quantitative functional relationship between these outputs and the actual values of radiation-sensitive parameters constitutes the mapping relationship between radiation conditions and radiation-sensitive parameters. This mapping relationship can be directly used to enhance the SPICE model, dynamically predicting the values of radiation-sensitive parameters based on the input radiation conditions, thereby improving the model's prediction accuracy and reliability under radiation environments. Furthermore, this mapping relationship can be used for further analysis and optimization, such as evaluating the impact of different radiation conditions on device performance through sensitivity analysis, or improving prediction accuracy through model optimization.
[0050] In a preferred embodiment of the present application, a machine learning model using an artificial neural network (ANN) is used to establish the mapping relationship between the radiation conditions and the radiation sensitive parameters. Specifically, a multi-layer perceptron (MLP) network can be used, which is a type of feedforward neural network that contains an input layer, one or more hidden layers, and an output layer. The number of nodes in the input layer is equal to the number of features of the radiation conditions, such as the type of radiation (proton, electron), the radiation dose (Gy or Rad), the radiation energy (MeV), etc. The hidden layer or layers each contain a number of neurons that are used to learn the non-linear relationship between the input features and the output. The number of nodes in the output layer is equal to the number of radiation sensitive parameters, such as RIGDDIO, EBREAKD, RNJGD, etc. In this embodiment, second electrical characteristic data under different radiation conditions and corresponding radiation sensitive parameter values are collected, and these data are used to train the MLP network by adjusting the network weights and biases to minimize the error between the predicted and actual values of the radiation sensitive parameters. A portion of the data not involved in the training is used to verify the predictive ability of the model to ensure that the model has good generalization performance. The trained MLP network is integrated into the SPICE model to dynamically predict the values of the radiation sensitive parameters according to the input radiation conditions, thereby achieving accurate simulation of the performance of the device in a radiation environment. In this way, the MLP network can learn the complex relationship between the radiation conditions and the radiation sensitive parameters and provide fast and accurate parameter prediction in actual applications, thereby improving the reliability and efficiency of circuit design.
[0051] Exemplarily, in combination with Figure 2As shown, a artificial neural network (ANN) model, particularly a multi-layer perceptron (MLP) network, is used to establish the mapping relationship between the radiation conditions and the radiation-sensitive parameters. First, a SPICE model suitable for the target device is selected, and the parameters affected by radiation in the model are determined. Then, the original electrical characteristic data of the device is obtained before radiation, and the model parameters are extracted. The device is irradiated under different radiation conditions, and the electrical characteristic data after irradiation is obtained, thereby extracting the values of the radiation-sensitive parameters. These data are used to construct a training data set, in which the radiation conditions are used as input features and the values of the radiation-sensitive parameters are used as output targets. Then, a suitable neural network architecture, such as a multi-layer perceptron (MLP), is selected, and the network parameters are initialized. The neural network is trained using the training set data, and the network parameters are adjusted through forward propagation and back propagation to minimize the loss function. After each training cycle, the model performance is evaluated using the validation set to prevent overfitting, and the model parameters or structure are adjusted accordingly. During the training process, the learning rate, batch size, and other hyperparameters may need to be adjusted to improve the training efficiency and model performance. After training, the trained MLP network is integrated into the SPICE model, and the values of the radiation-sensitive parameters are dynamically predicted according to the input radiation conditions, thereby achieving accurate simulation of the performance of the device in a radiation environment. In this way, the MLP network can learn the complex relationship between the radiation conditions and the radiation-sensitive parameters, and provide fast and accurate parameter prediction in practical applications, thereby improving the reliability and efficiency of circuit design. Not only does this improve the understanding of the performance changes of the device in a radiation environment, but the correlation data formed by associating the radiation conditions with the values of the radiation-sensitive parameters provides key information for establishing an enhanced SPICE model that can accurately reflect the behavior of the device in different radiation environments.
[0052] In this optional embodiment, a training data set containing radiation conditions as input features and corresponding radiation-sensitive parameter values as output targets is constructed, and then a neural network model is trained using this data set to ultimately form a quantitative functional relationship between the input features and the output targets, thereby constructing a mapping model that can accurately reflect the performance changes of the device under different radiation conditions, and further providing dynamic updated radiation-sensitive parameters for the enhanced SPICE model, so that circuit designers can more accurately predict and evaluate the performance and reliability of the circuit in the actual radiation environment.
[0053] Optionally, the mapping relationship is substituted into the original SPICE model to update the values of the radiation-sensitive parameters, and an enhanced SPICE model containing radiation effects is obtained, comprising: defining the quantitative functional relationship as a new parameter source of the radiation-sensitive parameters; When simulating the behavior of the target device under the radiation condition, the radiation condition is input into the quantitative function relationship to obtain the numerical value of the corresponding radiation-sensitive parameter; The numerical value of the radiation-sensitive parameter is substituted for the original defined value of the corresponding parameter in the original SPICE model to form the enhanced SPICE model.
[0054] Specifically, after the training of the neural network is completed and the mapping relationship between the radiation condition and the radiation-sensitive parameter is established, the mapping relationship is implemented in the form of a function, which is used to output the numerical value of the corresponding radiation-sensitive parameter according to the input radiation condition. In the embodiment of the present application, the function can be the neural network model itself or a mathematical expression extracted therefrom. In the SPICE model, the function is defined as a new parameter source of the radiation-sensitive parameter, that is, when the numerical value of the parameter is needed, the function is called to obtain the numerical value instead of directly using the original defined value in the model. When performing circuit simulation, the radiation condition required for simulation, such as a specific radiation type and dose, is first determined. Then, the radiation condition is input as input, and the numerical value of the radiation-sensitive parameter of the device under the radiation condition is calculated through the previously defined quantitative function relationship, that is, the mapping model. The numerical value reflects the performance change of the device under the radiation environment and is the key to accurate simulation.
[0055] After obtaining the numerical value of the radiation-sensitive parameter under the radiation condition, the numerical value is used to update the corresponding parameter in the original SPICE model, that is, in the parameter definition part of the model, the original static parameter value is replaced with the dynamically obtained parameter value calculated according to the current radiation condition through the mapping model. In this way, the original SPICE model is enhanced to reflect the performance change of the device under different radiation conditions, thereby forming an enhanced SPICE model containing radiation effects, which provides more accurate simulation results for circuit designers to evaluate and optimize the performance and reliability of the circuit in the radiation environment.
[0056] In the optional embodiment, the mapping relationship between the radiation condition and the radiation-sensitive parameter established by the neural network model is defined as a new parameter source, and the numerical value of the radiation-sensitive parameter is dynamically obtained when simulating the behavior of the target device under a specific radiation condition, and then the original defined value of the corresponding parameter in the original SPICE model is replaced, thereby realizing significant enhancement of the SPICE model. The enhanced SPICE model can dynamically adjust the radiation-sensitive parameter according to the input radiation condition, thereby more accurately simulating and predicting the electrical behavior and performance change of the device in the actual radiation environment. Not only does it improve the evaluation accuracy of radiation effects in the circuit design stage, but also reduces the dependence on a large amount of experimental data, reduces the research and development cost and time, and improves the generalization ability and applicability of the model.
[0057] Optionally, the radiation effect integrated method of the SPICE model further comprises: selecting a corresponding radiation condition according to actual modeling requirements and / or actual anti-radiation circuit index requirements, and taking the radiation condition as an input feature, the input feature comprising: a cumulative fluence of each of the radiation conditions, a total ionizing dose of each of the radiation conditions, a cumulative fluence of the radiation conditions, an average dose rate or a dose rate distribution statistical value representing a change of radiation intensity over time.
[0058] Specifically, the radiation condition is not limited to the current radiation environment, but also includes selecting one or more related radiation conditions as model inputs according to actual modeling requirements, i.e., the target or scene of simulation analysis, and / or actual anti-radiation circuit index requirements, i.e., hard standards or specifications that the circuit design must meet, according to the actual radiation scene to be simulated, such as low earth orbit, deep space exploration, nuclear reactor periphery, etc., and the circuit design index, such as anti-total dose level, single particle flip threshold, etc. The selected radiation conditions constitute an input feature set, which specifically includes at least one type of quantitative feature. Specifically, the cumulative fluence refers to the total fluence of a single or multiple radiation conditions that the device is subjected to in a radiation environment, i.e., the product of the number of particles and the area of the device. The cumulative fluence is an important index for evaluating radiation effects, and the cumulative fluence is directly related to the total amount of radiation energy absorbed by the device. The total ionizing dose is a measure of the energy deposited by radiation energy in a unit mass of material, usually measured in Gray (Gy), and the total ionizing dose reflects the total energy deposition of the device in a radiation environment, which is a key parameter for evaluating the degree of radiation damage. The dose rate is the ionizing radiation energy deposited per unit time, and the average dose rate represents the average value of the change of radiation intensity over time. This parameter is used to explain the dynamic process of radiation effects and reflects the speed of the impact of radiation on device performance. The dose rate distribution statistical value includes the distribution of the dose rate, such as the maximum value, the minimum value, the median, etc. These statistical values help to better understand the complexity of the radiation environment and its possible various impacts on device performance.
[0059] In this optional embodiment, by taking these radiation conditions as input features and as inputs of the neural network model, a more accurate mapping relationship can be established to dynamically update the radiation-sensitive parameters in the enhanced SPICE model. This embodiment not only reflects the current radiation condition, but also considers the cumulative effect of the device in the radiation environment, thereby more accurately predicting the performance changes of the device in actual application.
[0060] In combination with Figure 3 As shown in the figure, the radiation effect integrated system of the SPICE model of the present application comprises: An extraction unit is configured to extract model parameters affected by radiation from an original SPICE model corresponding to a target device as radiation-sensitive parameters. a data processing unit configured to acquire electrical characteristic data of the target device under different radiation conditions, and extract a value of the radiation-sensitive parameter under each of the radiation conditions based on the electrical characteristic data, to obtain correlation data between the different radiation conditions and the value of the radiation-sensitive parameter; a mapping relationship establishing unit configured to establish a mapping relationship between the radiation conditions and the value of the radiation-sensitive parameter based on the correlation data by using a neural network model; an updating unit configured to substitute the mapping relationship into the original SPICE model to update the value of the radiation-sensitive parameter, to obtain an enhanced SPICE model containing radiation effects.
[0061] The radiation effect integrated system of the SPICE model has the same advantages as the radiation effect integrated method of the SPICE model compared with the prior art, which will not be repeated here.
[0062] In combination with Figure 4 The computer device of the present application comprises a memory, a processor and a computer program stored on the memory and executable on the processor, and the processor implements the above-mentioned radiation effect integrated method of the SPICE model when executing the computer program.
[0063] The computer device of the present application has the same advantages as the radiation effect integrated method of the SPICE model compared with the prior art, which will not be repeated here.
[0064] The present application also provides a computer readable storage medium having a computer program stored thereon, and the computer program is executable on a processor to implement the above-mentioned radiation effect integrated method of the SPICE model.
[0065] The computer readable storage medium of the present application has the same advantages as the radiation effect integrated method of the SPICE model compared with the prior art, which will not be repeated here.
[0066] Although the present application is disclosed as above, the protection scope of the present application is not limited to this. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the present application, and these changes and modifications will fall within the protection scope of the present application.
Claims
1. A method for integrating the radiation effects of a SPICE model, characterized in that, include: From the original SPICE model corresponding to the target device, the model parameters affected by radiation are extracted as radiation-sensitive parameters; The electrical characteristics data of the target device under different radiation conditions are obtained, and the values of the radiation-sensitive parameters under each radiation condition are extracted based on the electrical characteristics data to obtain the correlation data between different radiation conditions and the values of the radiation-sensitive parameters. Based on the associated data, a neural network model is used to establish a mapping relationship between the radiation conditions and the values of the radiation-sensitive parameters; Substituting the mapping relationship into the original SPICE model to update the values of the radiation sensitivity parameters, an enhanced SPICE model incorporating radiation effects is obtained.
2. The method for integrating the radiation effects of the SPICE model according to claim 1, characterized in that, The step of extracting radiation-affected model parameters as radiation-sensitive parameters from the original SPICE model corresponding to the target device includes: The macroscopic electrical property degradation phenomenon of the target device under radiation environment is obtained; Based on the aforementioned macroscopic electrical property degradation phenomenon, the model expression describing this phenomenon in the original SPICE model is determined; Based on the radiation effect degradation mechanism of the target device, model parameters with physical interpretation mechanisms are identified from the model expression as radiation-sensitive parameters.
3. The method for integrating the radiation effects of the SPICE model according to claim 2, characterized in that, The extraction of radiation-affected model parameters as radiation-sensitive parameters includes: Based on the physical mechanism of radiation degradation of the target device under the radiation environment, the specific physical process of electrical characteristic degradation of the target device is obtained; From the original SPICE model, model parameters used to characterize the specific physical process are selected as the radiation-sensitive parameters.
4. The method for integrating the radiation effects of the SPICE model according to claim 3, characterized in that, The process of acquiring electrical characteristic data of the target device under different radiation conditions, and extracting the value of the radiation-sensitive parameter under each radiation condition based on the electrical characteristic data to obtain correlation data between different radiation conditions and the value of the radiation-sensitive parameter, includes: Obtain the first electrical characteristic data of the target device under non-radiation conditions, and based on the first electrical characteristic data, extract the baseline values of all model parameters in the original SPICE model; Under each of the radiation conditions, the target device is irradiated, and second electrical characteristic data are acquired under each of the radiation conditions. For each radiation condition, a parameter extraction operation is performed: the values of other model parameters in the original SPICE model, except for the radiation-sensitive parameters, are set as the baseline values, and the original SPICE model is run based on a set of values of the radiation-sensitive parameters under the current radiation conditions to obtain simulation results. By optimizing the values of the radiation-sensitive parameters, the simulation results are fitted with the second electrical characteristic data, and the fitted values of the radiation-sensitive parameters are determined as the values of the radiation-sensitive parameters under the current radiation conditions. The correlation data is formed by associating each radiation condition with the value of the corresponding radiation-sensitive parameter.
5. The method for integrating the radiation effects of the SPICE model according to claim 4, characterized in that, The step of establishing a mapping relationship between the radiation conditions and the values of the radiation-sensitive parameters using a neural network model based on the associated data includes: The radiation conditions are used as input features, and the values of the radiation sensitivity parameters corresponding to the radiation conditions are used as output targets to construct a training dataset. The training dataset is trained using an initial neural network, and a quantitative functional relationship between the input features and the output target is obtained from the trained initial neural network, thus forming the mapping relationship.
6. The method for integrating the radiation effects of the SPICE model according to claim 5, characterized in that, The step of substituting the mapping relationship into the original SPICE model to update the value of the radiation sensitivity parameter, thereby obtaining an enhanced SPICE model that includes radiation effects, includes: Define the quantitative functional relationship as a new parameter source for the radiation-sensitive parameter; When simulating the behavior of the target device under the radiation conditions, the radiation conditions are input into the quantitative function relationship to obtain the corresponding values of the radiation-sensitive parameters; The enhanced SPICE model is formed by replacing the original defined values of the corresponding parameters in the original SPICE model with the values of the radiation-sensitive parameters.
7. The method for integrating the radiation effects of the SPICE model according to any one of claims 1 to 6, characterized in that, Also includes: Select appropriate radiation conditions based on actual modeling requirements and / or actual radiation resistance circuit performance requirements, and use the radiation conditions as input features; The input features include: the cumulative fluence of each radiation condition, the total ionizing dose of each radiation condition, the cumulative fluence of each radiation condition, and the average dose rate or dose rate distribution statistics characterizing the change of radiation intensity over time.
8. An integrated system for the radiation effects of a SPICE model, characterized in that, include: The extraction unit is used to extract the radiation-affected model parameters as radiation-sensitive parameters from the original SPICE model corresponding to the target device. The data processing unit is used to acquire electrical characteristic data of the target device under different radiation conditions, and extract the value of the radiation sensitive parameter under each radiation condition based on the electrical characteristic data, so as to obtain the correlation data between different radiation conditions and the value of the radiation sensitive parameter. The mapping relationship establishment unit is used to establish a mapping relationship between the radiation conditions and the values of the radiation-sensitive parameters based on the associated data and using a neural network model; The update unit is used to substitute the mapping relationship into the original SPICE model to update the value of the radiation sensitivity parameter and obtain an enhanced SPICE model that includes the radiation effect.
9. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the radiation effect integration method of the SPICE model according to any one of claims 1 to 7.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the radiation effect integration method of the SPICE model according to any one of claims 1 to 7.