Action State Estimation via Statistical Signal Analysis

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional action state estimation systems require significant processing load to convert measurement signals into frequency components, leading to increased processing burdens while compromising estimation accuracy.

Innovation Solution

An action state estimation system that samples displacement measurement signals directly, calculates statistics, and uses an action state model to estimate loaded muscle states without frequency conversion, reducing processing load while maintaining desired accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If measurement signals are converted into frequency components using conventional methods, then estimation accuracy can be maintained, but processing load increases significantly

Engineering Contradiction:

Improveestimation accuracyVSAvoidprocessing load

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent extracts only the essential statistical features (mean, variance, skewness, kurtosis) from the displacement measurement data, rather than performing complete frequency transformation. This selective extraction maintains the necessary information for muscle loaded state estimation while eliminating unnecessary processing steps, thereby reducing computational complexity while preserving estimation accuracy.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

Instead of converting time-domain signals to frequency-domain components as in conventional methods, the patent inverts the approach by directly analyzing statistical properties of the time-domain displacement data. This inversion eliminates the need for computationally intensive Fourier transforms while still capturing the essential characteristics needed for accurate muscle loaded state estimation.

Inventive Principle:

Principle #13The other way round (Inversion)

2Loss of information

If frequency conversion is performed on measurement signals, then detailed signal analysis is possible, but processing time increases

Engineering Contradiction:

Improvesignal analysis capabilityVSAvoidprocessing time

Core Design Contradiction:

Loss of informationVSLoss of time

Solution Approach 1:

The patent extracts only the four essential statistical parameters (mean, variance, skewness, kurtosis) from displacement measurement data, discarding redundant information. This selective extraction maintains sufficient signal analysis capability for muscle loaded state estimation while dramatically reducing processing time by avoiding complete frequency transformation of the entire signal.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent performs partial frequency analysis by calculating only the necessary statistical moments (up to fourth order) rather than complete spectral decomposition. This partial action provides sufficient information for estimation purposes while avoiding the excessive processing time required for full frequency domain conversion.

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS20230148908A1Action state estimation apparatus, action state estimation method, action state learning apparatus, and action state learning method

Publication Date: 2023.05.18 MURATA MFG CO LTD

AI summary

An action state estimation apparatus is provided that includes a sampling portion, a statistic calculation portion, an action state model storage, and an estimation calculation portion. The sampling portion samples a displacement measurement signal within a predetermined time and generates displacement measurement data. The statistic calculation portion calculates a statistic of the displacement measurement data. The action state model storage stores an action state model modeled by associating the statistic with a loaded state of a muscle of the test subject. The estimation calculation portion estimates the loaded state by setting the statistic as an input vector and using the action state model.