Filter Characteristic Inversion for Raw Measurement Noise Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technical systems, such as radar and satellite navigation, often transmit only reduced information due to limited data rates, losing important details about measurement conditions and noise, which are crucial for further processing and assessment.
Innovation Solution
A method to determine statistical properties of raw measured values, including measurement noise and average errors, using output values from a filter process without precise knowledge of the filter, by learning filter characteristics during stable conditions and reconstructing raw values from useful signals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If only reduced information is transmitted due to limited data rate, then data transmission efficiency is improved, but measurement quality information is lost
Solution Approach 1:
The patent extracts only the essential reduced information (filtered output values) for transmission while retaining the full raw measured values at the receiving end. The filter characteristics are learned and transmitted, enabling reconstruction of statistical properties without transmitting the complete raw data set, thus achieving efficient transmission while preserving measurement quality information.
Solution Approach 2:
The patent performs preliminary filtering and learning of filter characteristics at the transmitting end before transmission. The filter characteristics are learned during stable measurement conditions and stored for later use in reconstructing statistical properties at the receiving end, enabling efficient transmission of only essential information.
2Measurement precision
If filter characteristics are learned during stable conditions, then accuracy of statistical property determination is improved, but additional processing time is required
Solution Approach 1:
The filter characteristics are learned in advance during stable measurement conditions when the system is not under stress. This preliminary learning phase allows the system to store accurate filter characteristics that can be quickly retrieved and used for reconstructing statistical properties during operation, avoiding time-consuming calculations when rapid response is needed.
Solution Approach 2:
The system uses feedback from stable measurement conditions to continuously refine and update filter characteristics. During stable periods, the system learns from the measurements and adjusts the filter characteristics, creating a self-improving system that maintains high accuracy without requiring continuous processing time.
Data Source
Figure 1
AI summary
In order to determine, at least approximately, statistical properties of raw measured values, particularly measurement noise and/or mean errors, without detailed knowledge of a filter procedure or filter for the raw measured values and with only the aid of output values of a wanted signal obtained from said raw measured values by means of the filter procedure, i.e. in order to make statements regarding the measurement conditions, assuming some common boundary conditions but without knowledge of the exact filter, a method is suggested for determining statistical properties of raw measured values (102) from a wanted signal (104) obtained from a filter procedure in a filter (103) and formed from a temporal sequence of output values from said filter. In a first method step, a filter characteristic of the filter is determined from a temporal sequence of output values obtained in stable measurement conditions, and in a second method step, the filter characteristic that has been determined is inverted, raw measured values being reconstructed from the inverse of the filter characteristic and the wanted signal and, from said reconstructed raw measured values, the statistical properties thereof being determined and/or these statistical properties being determined directly from the inverse of the filter characteristic and said wanted signal.