CN113743006A - Parameter optimization method, device, equipment and storage medium - Google Patents
Parameter optimization method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN113743006A CN113743006A CN202111002258.1A CN202111002258A CN113743006A CN 113743006 A CN113743006 A CN 113743006A CN 202111002258 A CN202111002258 A CN 202111002258A CN 113743006 A CN113743006 A CN 113743006A
- Authority
- CN
- China
- Prior art keywords
- thermal model
- algorithm
- parameters
- parameter
- simulation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/08—Thermal analysis or thermal optimisation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application discloses a method, a device, equipment and a storage medium for optimizing parameters, wherein the method comprises the following steps: generating candidate parameters of a thermal model by a particle swarm algorithm, wherein the thermal model is a simulation model established aiming at the thermal characteristics of chip packaging; running a thermal model according to the candidate parameters to obtain a simulated value; evaluating the thermal model according to the error between the simulated value and the reference value to obtain an evaluation result of the thermal model running under the candidate parameters; repeating the steps until the algorithm is converged; and determining candidate parameters with optimal evaluation results as the parameters of the thermal model. According to the method and the device, candidate parameters of the thermal model are generated through a particle swarm algorithm, new candidate parameters are generated through repeated iteration and the thermal model is adjusted until the algorithm converges, so that the optimal parameters are output, the parameter generation of the chip packaging thermal model is realized, and the problems of poor accuracy and poor accuracy caused by manual mode calibration of the thermal model parameters are solved.
Description
Technical Field
The present application relates to the field of integrated circuit design technologies, and in particular, to a method, an apparatus, a device, and a storage medium for parameter optimization.
Background
Thermal analysis of chip packages is an important part in integrated circuit design, and simulation analysis can be performed on different areas by establishing a thermal system simulation model (referred to as a thermal model in the present application) of the chip packages to obtain the temperature characteristics of key nodes, so that whether the chip packages meet the thermal characteristic requirements or not is verified at the integrated circuit design stage.
In order to accurately analyze the thermal performance of the chip package, digital modeling can be performed on a simulation platform based on physical equations. For example, a standard model library may be constructed for devices on the PCBA (printed circuit board assembly), and based on the model library, a thermal model of the PCBA may be quickly built for simulation analysis. Although the thermal model can be quickly built by parametric configuration by means of a standard model library, the thermal model still needs to be calibrated under specific conditions to ensure the accuracy and universality of the thermal model. Therefore, parameter identification of calibration parameters in a thermal model is an important part of the final application state of the model.
In view of this, in the related art, it is necessary to manually perform parameter calibration on the thermal model of the chip. However, calibrating a thermal model usually requires simultaneous calibration of a dynamic condition and a steady-state condition, and a plurality of Key Performance Indicators (KPIs) (e.g., temperatures of a plurality of attention areas) need to be considered at the same time, and after calibration is required, a plurality of KPIs need to achieve higher precision at the same time, so that difficulty in parameter identification is higher, and precision and accuracy of calibration by a manual method are poor.
Disclosure of Invention
The application provides a parameter optimization method, a parameter optimization device and a parameter optimization storage medium, and can solve the problems of poor parameter calibration accuracy and poor accuracy of a thermal model of chip packaging in a manual mode in the related art.
In one aspect, an embodiment of the present application provides a method for optimizing a parameter, including:
generating candidate parameters of a thermal model by a particle swarm algorithm, wherein the thermal model is a simulation model established aiming at the thermal characteristics of the chip package;
running the thermal model according to the candidate parameters to obtain a simulated value;
evaluating the thermal model according to the error between the simulated value and a reference value to obtain an evaluation result of the thermal model running under the candidate parameter;
repeating the steps until the algorithm is converged;
and determining the candidate parameter with the optimal evaluation result as the parameter of the thermal model.
Optionally, the method further includes:
determining that the algorithm converges when the error no longer decreases or the oscillation converges.
Optionally, the method further includes:
determining that the algorithm converges when the number of times the candidate parameter is generated exceeds a quantity threshold.
Optionally, the running the thermal model according to the candidate parameter to obtain a simulated value includes:
assigning the thermal model according to the candidate parameters to obtain an assigned model;
and operating the assigned model to obtain the simulation value.
Optionally, the thermal model includes at least two key contribution indicators, and the key contribution indicators correspond to different positions of the chip package;
the simulation value comprises a plurality of simulation curves, and each simulation curve corresponds to the key contribution index;
the reference value includes a plurality of reference curves, each of the reference curves corresponding to the key contribution indicator.
Optionally, the evaluating the candidate parameter according to the error between the simulated value and the reference value to obtain an evaluation result includes:
sampling each simulation curve according to a preset condition to obtain simulation sampling data of each key contribution index;
sampling each reference curve according to the preset condition to obtain reference sampling data of each key contribution index;
calculating the mean square error of each simulation sampling data and the corresponding reference sampling data;
and calculating to obtain the evaluation result according to each mean square error.
Optionally, the sampling each simulation curve according to a preset condition includes:
performing discrete sampling on each simulation curve according to a preset frequency;
the sampling each reference curve according to the preset condition includes:
and performing the discrete sampling on each reference curve according to the preset frequency.
Optionally, the calculating the evaluation result according to each mean square error includes:
calculating the average value of the mean square deviations according to each mean square deviation;
and taking the opposite number of the mean value of the mean square error as the evaluation result.
Optionally, the generating candidate parameters of the thermal model by the particle swarm algorithm includes:
updating an individual optimal solution and a global optimal solution of the algorithm according to the evaluation result, wherein the individual optimal solution is a solution which enables the evaluation result of each population in the algorithm to be optimal, and the global optimal solution is a solution which enables the evaluation results of all the populations in the algorithm to be comprehensively optimal;
outputting the global optimal solution as the candidate parameter;
updating the position and the speed of each population in the algorithm according to the evaluation result to obtain the updated position and speed;
updating the population of the algorithm according to the updated position and speed;
and repeating the steps.
Optionally, the candidate parameter with the optimal evaluation result is a final global optimal solution of the algorithm.
On the other hand, an embodiment of the present application provides an apparatus for optimizing a parameter, including:
the parameter generation module is used for generating candidate parameters of a thermal model through a particle swarm algorithm, wherein the thermal model is a simulation model established aiming at the thermal characteristics of the chip package; running the thermal model according to the candidate parameters to obtain a simulated value; evaluating the thermal model according to the error between the simulated value and a reference value to obtain an evaluation result of the thermal model running under the candidate parameter; repeating the steps until the algorithm is converged;
and the parameter output module is used for determining the candidate parameter with the optimal evaluation result as the parameter of the thermal model.
In another aspect, an embodiment of the present application provides a computer device, including a processor and a memory, where the memory stores at least one instruction or program, and the instruction or program is loaded and executed by the processor to implement the parameter optimization method as described in any one of the above.
In another aspect, an embodiment of the present application provides a computer-readable storage medium, where at least one instruction is stored in the storage medium, and the instruction is loaded and executed by a processor to implement the parameter optimization method as described in any one of the above.
The technical scheme at least comprises the following advantages:
candidate parameters of the thermal model are generated through a particle swarm algorithm, new candidate parameters are generated through multiple iterations to adjust the thermal model until the algorithm converges, and therefore the optimal parameters are output, the parameter generation of the chip packaging thermal model is achieved, the problem that the accuracy and the precision are poor due to manual calibration of the thermal model parameters is solved, and the accuracy and the precision of the thermal model are improved.
Drawings
In order to more clearly illustrate the detailed description of the present application or the technical solutions in the prior art, the drawings needed to be used in the detailed description of the present application or the prior art description will be briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative efforts.
FIG. 1 is a block diagram of a computer device provided by an exemplary embodiment of the present application;
FIG. 2 is a flow chart of a method for optimizing parameters provided by an exemplary embodiment of the present application;
FIG. 3 is a flow chart of a parameter evaluation method provided by an exemplary embodiment of the present application;
FIG. 4 is a flow chart of a method for generating parameters provided by an exemplary embodiment of the present application;
FIG. 5 is a flow chart of a method for optimizing parameters provided by an exemplary embodiment of the present application;
fig. 6 is a block diagram of an apparatus for optimizing parameters according to an exemplary embodiment of the present application.
Detailed Description
The technical solutions in the present application will be described clearly and completely with reference to the accompanying drawings, and it is obvious that the described embodiments are some, but not all embodiments of the present application. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
In the description of the present application, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings, and are only for convenience of description and simplicity of description, and do not indicate or imply that the device or element being referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus, should not be construed as limiting the present application. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.
In the description of the present application, it is to be noted that, unless otherwise explicitly specified or limited, the terms "mounted," "connected," and "connected" are to be construed broadly, e.g., as meaning either a fixed connection, a removable connection, or an integral connection; the connection can be mechanical connection or electrical connection; the two elements may be directly connected or indirectly connected through an intermediate medium, or may be communicated with each other inside the two elements, or may be wirelessly connected or wired connected. The specific meaning of the above terms in the present application can be understood in a specific case by those of ordinary skill in the art.
In addition, the technical features mentioned in the different embodiments of the present application described below may be combined with each other as long as they do not conflict with each other.
Referring to FIG. 1, a block diagram of a computer device provided by an exemplary embodiment of the present application is shown. The computer device may be a Personal Computer (PC) or a server (or server group). It includes: a processor 110 and a memory 120.
The processor 110 may be a Central Processing Unit (CPU), a Network Processor (NP), or a combination of a CPU and an NP. The processor 110 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a Programmable Logic Device (PLD), or a combination thereof. The PLD may be a Complex Programmable Logic Device (CPLD), a field-programmable gate array (FPGA), a General Array Logic (GAL), or any combination thereof.
The memory 120 is connected to the processor 110 through a bus or other means, and at least one instruction, at least one program, a code set, or a set of instructions is stored in the memory 120, and is loaded and executed by the processor 110 to implement the parameter optimization method provided in any one of the following embodiments. The memory 120 may be a volatile memory (volatile memory), a non-volatile memory (non-volatile memory), or a combination thereof. The volatile memory may be a random-access memory (RAM), such as a Static Random Access Memory (SRAM) or a Dynamic Random Access Memory (DRAM). The nonvolatile memory may be a Read Only Memory (ROM), such as a Programmable Read Only Memory (PROM), an Erasable Programmable Read Only Memory (EPROM), and an electrically erasable programmable read-only memory (EEPROM). The nonvolatile memory may also be a flash memory (flash memory), a magnetic memory such as a magnetic tape (magnetic tape), a floppy disk (floppy disk), and a hard disk. The non-volatile memory may also be an optical disc.
The memory 120 stores therein a thermal model 121, and the thermal model 121 is a simulation model established for the thermal characteristics of the chip package. Illustratively, the thermal model 121 may be a simulation model built on a Dymola system (modeling and simulation software available from Dassault Systems, france): the original thermal model can be constructed in the Dymola system and imported in FMU file format into Simulink (a visual simulation tool in MATLAB) of MATLAB (commercial math software available from MathWorks, inc., usa). The thermal model 121 may be optimized for parameters by any of the following method embodiments.
Referring to fig. 2, a flow chart of a method for optimizing parameters provided by an exemplary embodiment of the present application is shown, the method being executable by a computer device in the embodiment of fig. 1, and the method including:
and step 201, generating candidate parameters of the thermal model through a particle swarm algorithm.
Wherein the thermal model may be any of the thermal models described above. Illustratively, a chip to be packaged is applied to a vehicle, parameters related to thermal characteristics of the chip package include various parameters to be identified under various working conditions, and candidate parameters can be selected from the parameters to be identified through a particle swarm algorithm. In the initial stage of iteration, a group of candidate parameters can be randomly generated from the parameters to be identified, and the algorithm is initialized to serve as the initial candidate parameters.
Optionally, step 202 includes, but is not limited to: the thermal model can be assigned according to the candidate parameters to obtain an assigned model; and operating the assigned model to obtain a simulated value.
For example, in this embodiment of the application, the thermal model includes at least two key contribution indicators, the key contribution indicators correspond to different positions of the chip package, and after the assigned thermal model is run, a simulation curve of each key contribution indicator is obtained as a simulation value, so that the simulation value includes a plurality of simulation curves, and each simulation curve has its corresponding key contribution indicator. For example, the temperature of at least two areas of the junction, pad, lead (both sides), and package surface may be used as key contributors.
And step 203, evaluating the thermal model according to the error between the simulated value and the reference value to obtain an evaluation result of the thermal model running under the candidate parameters.
Where the error between the simulated value and the reference value is calculated, it can be described in various ways, for example, it can be the difference (or the average of the difference) between the simulated value and the reference value, the mean square error (or the average of the mean square error) between the simulated value and the reference value, and so on.
Illustratively, when the thermal model includes at least two key contribution indicators, the simulation value output is a simulation curve of each key contribution indicator, for example, the thermal model includes the above five key contribution indicators (junction, pad, lead (both sides), and package surface temperature), the simulation value is a simulation curve (thermal profile, time-temperature curve) of the five key contribution indicators, correspondingly, the reference value includes a plurality of reference curves, each reference curve has its corresponding key contribution indicator, and an error of the thermal model operating under the candidate parameter can be calculated according to the simulation curve and the reference curve corresponding to each key contribution indicator.
After the error is calculated, the thermal model may be evaluated according to the error, generally speaking, the smaller the error is, the better the evaluation result is, and the evaluation mode may be set according to the requirement, which is not described herein. If the model comprises at least two key contribution indexes, the error between the simulation curve and the reference curve corresponding to each key contribution index needs to be calculated, and the evaluation result of the iteration is obtained according to the error calculation corresponding to different key contribution indexes.
When the algorithm is determined to be converged, entering step 205; when it is determined that the algorithm does not converge, step 201 is entered.
And repeating the steps to generate new candidate parameters, operating the thermal model according to the new candidate parameters so as to evaluate to obtain an evaluation result, updating the parameters in the algorithm according to the evaluation result, and further generating the new candidate parameters until the particle swarm optimization converges. The basis for determining the convergence of the algorithm may be any of the following: (1) the error is not reduced any more; (2) converging error oscillation; (3) the number of times the candidate parameter is generated (number of iterations) exceeds a quantity threshold.
And step 205, determining candidate parameters with the optimal evaluation result as the parameters of the thermal model.
And when the particle swarm optimization is determined to be converged, determining the candidate parameters with the optimal evaluation result as the parameters of the thermal model, and constructing the thermal model according to the parameters.
In summary, in the embodiment of the present application, the candidate parameters of the thermal model are generated through the particle swarm algorithm, and new candidate parameters are generated through multiple iterations to adjust the thermal model until the algorithm converges, so as to output the optimal parameters, thereby realizing the parameter generation of the chip packaging thermal model, solving the problem of poor accuracy and precision caused by manually calibrating the parameters of the thermal model, and improving the accuracy and precision of the thermal model.
Referring to fig. 3, which shows a flowchart of a parameter evaluation method provided in an exemplary embodiment of the present application, the method may be executed by a computer device in the embodiment of fig. 1, and the method may be an optional implementation of step 202 in the embodiment of fig. 2, and the method includes:
Wherein the preset condition comprises a preset frequency. Illustratively, each simulation curve may be subjected to discrete sampling according to a preset frequency to obtain simulation sampling data of each key contribution index.
For example, each reference curve may be discretely sampled according to the same preset frequency to obtain reference sample data of each key contribution index.
And for any key contribution index, calculating corresponding simulation sampling data and reference sampling data to obtain the mean square error of each key contribution index.
And step 304, calculating according to each mean square error to obtain an evaluation result.
Illustratively, the mean of the mean square deviations can be calculated according to the mean square deviation of each key contribution index; the evaluation result was the inverse of the mean square error. Thus, the larger the value of the evaluation result, the smaller the average error of the thermal model over the plurality of key contribution indicators, and the better the model performance.
Referring to fig. 4, which shows a flowchart of a method for generating parameters according to an exemplary embodiment of the present application, where the method may be executed by a computer device in the embodiment of fig. 1, and the method may be an optional implementation of step 201 in the embodiment of fig. 2, and the method includes:
and step 401, updating the individual optimal solution and the global optimal solution of the particle swarm algorithm according to the evaluation result.
The individual optimal solution is a solution which enables the evaluation result of each population in the algorithm to be optimal, and the global optimal solution is a solution which enables the evaluation results of all the populations in the algorithm to be comprehensively optimal.
When the particle swarm algorithm is operated for the first time, initialization is needed, and the initialization step comprises the following steps: determining algorithm parameters, wherein the algorithm parameters comprise at least one of iteration times of the algorithm, the size of a population, learning factors of particles in the algorithm and inertia factors of the particles; and randomly generating a group of initial feasible solutions as initial candidate parameters in a reasonable value range according to the characteristics of the thermal model.
And step 402, outputting the global optimal solution as a candidate parameter.
And outputting the global optimal solution of the iteration as a candidate parameter by the particle swarm algorithm, namely evaluating the candidate parameter to obtain a corresponding evaluation result.
And step 403, updating the position and the speed of each population in the algorithm according to the evaluation result to obtain the updated position and speed.
And step 404, updating the population of the algorithm according to the updated position and speed.
For example, the position and the speed of each population can be updated according to the evaluation result based on the algorithm parameters, so as to obtain the updated position and speed of the population; and updating the population of the algorithm according to the updated position and the updated speed of the population.
When the algorithm is not converged, the method enters step 401; when the algorithm converges, the iteration ends.
Referring to fig. 5, which shows a flowchart of a method for optimizing parameters provided in an exemplary embodiment of the present application, as shown in fig. 5, the method includes:
And step 503, constructing a fitness function.
The fitness function is a function used for evaluating the performance of the thermal model under different candidate parameters in the method. In the following, a thermal model simulation of a certain chip package is taken as an example to explain:
the model has a plurality of (more than 30) working conditions and a plurality of (more than 50) parameters to be identified. The model needs to identify a plurality of parameters under the various working conditions, so that the final model can achieve higher simulation precision when the temperature of the junction, the bonding pad, the pin (two sides) and the packaging surface is used as a key contribution index.
In view of this, each time a new set of candidate parameters is generated by the particle swarm algorithm, the new set of candidate parameters is assigned to the corresponding Simulink model for evaluation. The model obtains simulation curves about the five key contribution indexes, and the five simulation curves are subjected to discrete sampling according to a certain sampling frequency.
Because the given sampling time in the reference value comprises a plurality of granularities, the five simulation curves can be sampled according to the sampling time with the coarsest granularity, the reference value is also subjected to down-sampling according to the coarsest time granularity, sampling points with simulation values of fixed sampling time granularity and the reference value are obtained, the sampling points are respectively corresponding on a time axis to obtain sampling data (comprising simulation sampling data and reference sampling data), errors of the simulation sampling data and the reference sampling data in the sampling data of five key contribution indexes of the model are respectively calculated through a mean square error function (mean square error function), and the number of the opposite of the error mean is taken as a fitness function value (cost function) of the model under set parameters.
From the above, the larger the fitness value is, the smaller the average error of the model on five key contribution indexes is, and the better the model performance is, so that the fitness function can be used for evaluating the model performance.
Taking the above model as an example, the model has a plurality of parameters to be identified in common, and the particle swarm algorithm cannot directly process the parameters, so the parameters need to be encoded into a form that can be processed and calculated by the algorithm. In the thermal model, since all parameters exist in the form of independent numerical parameters, all parameters are spliced together in a certain order in the spatial decoding stage and reconstructed into a vector form. That is, if there are 52 parameters to be identified, the solution generated by the algorithm will be a 52-dimensional vector each time.
And 505, performing particle swarm optimization.
After the fitness function design and the spatial coding are completed, reference function (benchmark) data are read through a particle swarm algorithm to generate and iteratively update a population of particles consisting of a cluster of feasible solutions until the algorithm converges or the iteration times reach a time threshold, so that a historical optimal solution of the population is output and is used as a final identification result and a parameter of a thermal model. The calculation of the algorithm can be referred to above, and is not described herein.
And step 506, outputting parameters with the optimal evaluation results.
The output parameters with the optimal evaluation results can be used as the parameters of the thermal model to construct the thermal model.
Referring to fig. 6, a block diagram of a parameter optimization apparatus provided in an exemplary embodiment of the present application is shown, and the apparatus may be implemented as a computer device in any of the above embodiments through software, hardware or a combination of both. The device includes: a parameter generation module 610 and a parameter output module 620, wherein:
a parameter generating module 610, configured to generate candidate parameters of a thermal model by using a particle swarm algorithm, where the thermal model is a simulation model established for thermal characteristics of a chip package; running a thermal model according to the candidate parameters to obtain a simulated value; evaluating the thermal model according to the error between the simulated value and the reference value to obtain an evaluation result of the thermal model running under the candidate parameters; and repeating the steps until the algorithm converges.
And a parameter output module 620, configured to determine a candidate parameter with an optimal evaluation result as a parameter of the thermal model.
Optionally, the parameter generating module 610 is further configured to determine that the algorithm converges when the error is no longer reduced or the oscillation converges.
Optionally, the parameter generating module 610 is further configured to determine that the algorithm converges when the number of times of generating the candidate parameter exceeds the number threshold.
Optionally, the parameter generating module 610 is further configured to assign a value to the thermal model according to the candidate parameter to obtain an assigned model; and operating the assigned model to obtain a simulated value.
Optionally, the thermal model includes at least two key contribution indexes, and the key contribution indexes correspond to different positions of the chip package; the simulation value comprises a plurality of simulation curves, and each simulation curve corresponds to a key contribution index; the reference value includes a plurality of reference curves, each reference curve corresponding to a key contribution indicator.
Optionally, the parameter generating module 610 is further configured to sample each simulation curve according to a preset condition to obtain simulation sampling data of each key contribution index; sampling each reference curve according to a preset condition to obtain reference sampling data of each key contribution index; calculating the mean square error of each simulation sampling data and the corresponding reference sampling data; and calculating according to each mean square error to obtain an evaluation result.
Optionally, the parameter generating module 610 is further configured to perform discrete sampling on each simulation curve according to a preset frequency; and performing discrete sampling on each reference curve according to a preset frequency.
Optionally, the parameter generating module 610 is further configured to calculate an average value of the mean square deviations according to each mean square deviation; the evaluation result was the inverse of the mean square error.
Optionally, the parameter generating module 610 is further configured to update an individual optimal solution and a global optimal solution of the algorithm according to the evaluation result, where the individual optimal solution is a solution that optimizes the evaluation result of each population in the algorithm, and the global optimal solution is a solution that optimizes the evaluation results of the populations in the algorithm; outputting the global optimal solution as a candidate parameter; updating the position and the speed of each population in the algorithm according to the evaluation result to obtain the updated position and speed; updating the population of the algorithm according to the updated position and speed; and repeating the steps.
Optionally, the candidate parameter with the optimal evaluation result is a final global optimal solution of the algorithm.
The present application further provides a computer-readable storage medium having stored therein at least one instruction, at least one program, a set of codes, or a set of instructions, which is loaded and executed by the processor to implement the method for optimizing parameters according to any of the above embodiments.
The present application also provides a computer program product, which when run on a computer, causes the computer to execute the method for optimizing parameters provided by the above method embodiments.
It should be understood that the above examples are only for clarity of illustration and are not intended to limit the embodiments. Other variations and modifications will be apparent to persons skilled in the art in light of the above description. And are neither required nor exhaustive of all embodiments. And obvious variations or modifications of this invention are intended to be covered by the scope of the invention as expressed herein.
Claims (13)
1. A method for optimizing a parameter, comprising:
generating candidate parameters of a thermal model by a particle swarm algorithm, wherein the thermal model is a simulation model established aiming at the thermal characteristics of the chip package;
running the thermal model according to the candidate parameters to obtain a simulated value;
evaluating the thermal model according to the error between the simulated value and a reference value to obtain an evaluation result of the thermal model running under the candidate parameter;
repeating the steps until the algorithm is converged;
and determining the candidate parameter with the optimal evaluation result as the parameter of the thermal model.
2. The method of claim 1, further comprising:
determining that the algorithm converges when the error no longer decreases or the oscillation converges.
3. The method of claim 1, further comprising:
determining that the algorithm converges when the number of times the candidate parameter is generated exceeds a quantity threshold.
4. A method according to any one of claims 1 to 3, wherein said running said thermal model according to said candidate parameters, resulting in simulated values, comprises:
assigning the thermal model according to the candidate parameters to obtain an assigned model;
and operating the assigned model to obtain the simulation value.
5. The method of claim 4, wherein the thermal model includes at least two key contribution indicators corresponding to different locations of the chip package;
the simulation value comprises a plurality of simulation curves, and each simulation curve corresponds to the key contribution index;
the reference value includes a plurality of reference curves, each of the reference curves corresponding to the key contribution indicator.
6. The method of claim 5, wherein the evaluating the candidate parameter according to the error between the simulated value and the reference value to obtain an evaluation result comprises:
sampling each simulation curve according to a preset condition to obtain simulation sampling data of each key contribution index;
sampling each reference curve according to the preset condition to obtain reference sampling data of each key contribution index;
calculating the mean square error of each simulation sampling data and the corresponding reference sampling data;
and calculating to obtain the evaluation result according to each mean square error.
7. The method of claim 6, wherein the sampling each simulation curve according to a preset condition comprises:
performing discrete sampling on each simulation curve according to a preset frequency;
the sampling each reference curve according to the preset condition includes:
and performing the discrete sampling on each reference curve according to the preset frequency.
8. The method of claim 7, wherein said calculating said evaluation result from each of said mean square deviations comprises:
calculating the average value of the mean square deviations according to each mean square deviation;
and taking the opposite number of the mean value of the mean square error as the evaluation result.
9. The method of claim 8, wherein generating candidate parameters for the thermal model by a particle swarm algorithm comprises:
updating an individual optimal solution and a global optimal solution of the algorithm according to the evaluation result, wherein the individual optimal solution is a solution which enables the evaluation result of each population in the algorithm to be optimal, and the global optimal solution is a solution which enables the evaluation results of all the populations in the algorithm to be comprehensively optimal;
outputting the global optimal solution as the candidate parameter;
updating the position and the speed of each population in the algorithm according to the evaluation result to obtain the updated position and speed;
updating the population of the algorithm according to the updated position and speed;
and repeating the steps.
10. The method of claim 9, wherein the candidate parameter with the best evaluation result is a final global optimal solution of the algorithm.
11. An apparatus for optimizing a parameter, comprising:
the parameter generation module is used for generating candidate parameters of a thermal model through a particle swarm algorithm, wherein the thermal model is a simulation model established aiming at the thermal characteristics of the chip package; running the thermal model according to the candidate parameters to obtain a simulated value; evaluating the thermal model according to the error between the simulated value and a reference value to obtain an evaluation result of the thermal model running under the candidate parameter; repeating the steps until the algorithm is converged;
and the parameter output module is used for determining the candidate parameter with the optimal evaluation result as the parameter of the thermal model.
12. A computer device comprising a processor and a memory, said memory having stored therein at least one instruction or program that is loaded and executed by said processor to implement a method of optimizing parameters according to any one of claims 1 to 10.
13. A computer-readable storage medium having stored thereon at least one instruction which is loaded and executed by a processor to implement a method of optimizing a parameter according to any one of claims 1 to 10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111002258.1A CN113743006A (en) | 2021-08-30 | 2021-08-30 | Parameter optimization method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111002258.1A CN113743006A (en) | 2021-08-30 | 2021-08-30 | Parameter optimization method, device, equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113743006A true CN113743006A (en) | 2021-12-03 |
Family
ID=78733705
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111002258.1A Pending CN113743006A (en) | 2021-08-30 | 2021-08-30 | Parameter optimization method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113743006A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115495854A (en) * | 2022-11-21 | 2022-12-20 | 苏州数算软云科技有限公司 | Parameter calibration method, device, equipment and medium of computer aided engineering model |
| CN116307131A (en) * | 2023-02-14 | 2023-06-23 | 北京市燃气集团有限责任公司 | System parameter determining method and device for natural gas recovery system |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101017510A (en) * | 2006-03-13 | 2007-08-15 | ***电子第五研究所 | Heat resistance network model and method for calculating junction temperature of microelectronics using same |
| CN105159085A (en) * | 2015-09-09 | 2015-12-16 | 东南大学 | real time parameter identification method of air-conditioning belonged building second order equivalent thermal parameter model |
-
2021
- 2021-08-30 CN CN202111002258.1A patent/CN113743006A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101017510A (en) * | 2006-03-13 | 2007-08-15 | ***电子第五研究所 | Heat resistance network model and method for calculating junction temperature of microelectronics using same |
| CN105159085A (en) * | 2015-09-09 | 2015-12-16 | 东南大学 | real time parameter identification method of air-conditioning belonged building second order equivalent thermal parameter model |
Non-Patent Citations (3)
| Title |
|---|
| 万萌等: "功率器件集总参数热路模型及其参数提取研究", 电工技术学报, vol. 30, no. 21, pages 1 - 4 * |
| 师蔚等: "基于粒子群优化算法的永磁电机热网络参数识别", 电机与控制应用, vol. 2019, no. 3, pages 1 - 2 * |
| 徐铭伟;周雒维;杜雄;沈刚;杨旭;: "NPT型IGBT电热仿真模型参数提取方法综述", 电力自动化设备, no. 01 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115495854A (en) * | 2022-11-21 | 2022-12-20 | 苏州数算软云科技有限公司 | Parameter calibration method, device, equipment and medium of computer aided engineering model |
| CN115495854B (en) * | 2022-11-21 | 2023-04-07 | 苏州数算软云科技有限公司 | Parameter calibration method, device, equipment and medium of computer aided engineering model |
| CN116307131A (en) * | 2023-02-14 | 2023-06-23 | 北京市燃气集团有限责任公司 | System parameter determining method and device for natural gas recovery system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113743006A (en) | Parameter optimization method, device, equipment and storage medium | |
| CN112699504B (en) | Assembly physical digital twin modeling method and device, electronic equipment and medium | |
| US9465773B2 (en) | Data-driven distributionally robust optimization | |
| CN112070216B (en) | Method and system for training graph neural network model based on graph computing system | |
| US8665731B1 (en) | Reliability estimation methods for large networked systems | |
| US11409591B2 (en) | Anomaly determination apparatus, anomaly determination method, and non-transitory computer readable medium storing program | |
| CN113011055B (en) | Motor high-temperature fault simulation method and system based on finite element technology | |
| US8078446B2 (en) | Linear time-invariant system modeling apparatus and method of generating a passive model | |
| KR20220082003A (en) | Die-level product modeling without die-level input data | |
| Stratigopoulos et al. | Multidimensional analog test metrics estimation using extreme value theory and statistical blockade | |
| JP7440984B2 (en) | Highly efficient circuit simulation methods, devices, equipment and storage media | |
| Cevallos et al. | The extended Kalman filter in the dynamic state estimation of electrical power systems | |
| US20210232941A1 (en) | Method, system, and computer program product for determining causality | |
| CN114663074B (en) | Product data analysis method and system based on digital twins | |
| CN112297435A (en) | Spatial difference measurement | |
| CN111812445B (en) | Intelligent capacitor error correction method and device, computer equipment and storage medium | |
| US11199833B2 (en) | Quality determination method, quality determination device, quality determination system and computer-readable non-transitory medium | |
| Campomanes-Álvarez et al. | Mesh simplification for 3D modeling using evolutionary multi-objective optimization | |
| US20060158287A1 (en) | Parameter adjuster | |
| EP4006767A1 (en) | Sensor model generation method and system, and sensor measurement method and system | |
| US20090276171A1 (en) | Data processing apparatus and method | |
| Wu et al. | A correlation-based approach to trustworthy sensing for cyber-physical systems | |
| CN116933490A (en) | Pedal testing method and device, electronic equipment and storage medium | |
| Kofanov et al. | Investigation Methods of Parameters Random Variations in Control Systems | |
| Schär et al. | A data-driven surrogate model for uncertainty quantification of dynamical systems |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |