Circuit Simulation Parameter Identification via Statistical Dispersion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current simulation techniques for circuit design, such as those using SPICE, face challenges in accurately setting input parameter values and evaluating errors between calculated and actual measurement values, often relying on subjective weight adjustments and lacking objective, quantitative methods.
Innovation Solution
An input parameter value set identifying method that calculates an indicator value for dispersion, such as standard deviation, to evaluate errors and automatically adjust parameter values, minimizing overall error through iterative simulations and optimization algorithms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional error function with weight adjustment is used to adjust input parameter values, then fitting can be carried out according to user's sense, but it is difficult to appropriately set the weight and repeated adjustments are required
Solution Approach 1:
The system automatically determines weights based on actual measurement data without requiring manual user input. The weight for each measurement point is calculated autonomously using the standard deviation of actual values at that point, eliminating the need for repeated manual weight adjustments while improving error evaluation accuracy
Solution Approach 2:
The system uses the calculated standard deviation of actual measurement values as feedback to automatically adjust weights. This feedback mechanism allows the system to learn from the data distribution and appropriately weight different measurement points, resolving the contradiction between accurate error evaluation and time-consuming manual adjustment
2Ease of operation
If conventional techniques are used for setting input parameters, then parameter setting can be performed, but the weight is not considered and objective basis for error evaluation is lacking
Solution Approach 1:
The system automatically calculates weights based on the statistical properties (standard deviation) of the actual measurement data itself. This self-service approach eliminates the need for manual weight setting while providing an objective, data-driven basis for error evaluation, thereby improving both ease of operation and measurement precision
3Adaptability or versatility
If user manually adjusts weight for fitting, then fitting can be performed according to user's sense, but the setting depends on user's skill and objective basis is difficult to indicate
Solution Approach 1:
The system replaces subjective user judgment with objective statistical calculation. By automatically computing weights from the standard deviation of actual measurement values, the system maintains fitting flexibility while eliminating dependence on user skill and providing a clear objective basis for error evaluation
Solution Approach 2:
The system changes the approach from fixed manual weight setting to dynamic weight calculation based on data characteristics. The weight for each measurement point is automatically adjusted according to its standard deviation, allowing the system to adapt to different data conditions while maintaining objectivity
Data Source
AI summary
For each input variable value set, an indicator value associated with dispersion of actually measured output variable values is calculated from data including, for each input variable value set, the actually measured output variable values. Then, a processing to cause a simulator to compute a calculated output variable value for each combination of a candidate input parameter value sets and one input variable value set, and a processing to calculate, for each candidate input parameter value set, an entire error obtained by taking into consideration, with respect to all input variable value sets, partial errors obtained respectively by evaluating, by the indicator value for a corresponding input variable value set, difference between the calculated and actually measured output variable values for the corresponding input variable value set are repeated to identify the candidate input parameter value set making the entire error minimum.