Radar Fill-Level Sensing With ML for Multiple Reflection Signals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing radar-based fill-level measurement systems struggle to accurately detect the fill level and additional layers due to interference from impurities, additives, and internal container fittings, leading to uncertain signal maximum identification.
Innovation Solution
Implementing a radar-based fill-level measuring system with a machine learning algorithm to interpret received signals, particularly using artificial neural networks for deep learning, to enhance the certainty of fill-level detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If radar-based measurement principles (pulse transit time, FMCW, TDR) are used to measure fill level continuously and without contact, then measurement reliability and robustness are improved, but measurement precision deteriorates when impurities, additives, or foreign substances form additional product layers that create multiple reflections and interfere with signal maximum identification
Solution Approach 1:
The patent segments the received signal into multiple components by identifying individual signal maxima and their corresponding distance values. The evaluation unit processes each signal maximum separately, assigning distance values to multiple layers (product layer, additive layer, foam layer) rather than treating the signal as a single entity. This segmentation allows precise identification of each layer's fill level independently.
Solution Approach 2:
The patent performs preliminary signal processing by storing the received signal and conducting offline evaluation using machine learning algorithms. The signal is pre-processed to identify signal maxima, calculate distance values, and determine fill levels before final interpretation. This preliminary action prepares the data structure to handle multiple reflections and interfering echoes systematically.
2Measurement precision
If machine learning algorithms are implemented to interpret complex signals and identify signal maxima with certainty, then measurement precision is improved, but device complexity increases due to the additional computational requirements and processing overhead
Solution Approach 1:
The patent replaces traditional mechanical signal processing methods with machine learning algorithms. Instead of using conventional threshold-based or amplitude-based detection methods to identify signal maxima, the system employs artificial neural networks and deep learning models that automatically learn to distinguish valid signal maxima from interfering echoes caused by impurities, additives, and container fittings.
Solution Approach 2:
The patent introduces an intermediary evaluation unit that acts as a bridge between the raw received signal and the final fill level measurement. This evaluation unit contains the machine learning algorithm that processes the complex signal patterns, identifies signal maxima, and determines distance values. The intermediary layer handles the computational complexity internally while presenting simplified, accurate results to the user.
3Speed
If the fill level is determined by identifying the signal maximum and its corresponding distance value, then measurement speed is improved, but measurement precision deteriorates when additional filling material layers or interfering objects generate multiple reflections that make signal maximum identification uncertain
Solution Approach 1:
The patent performs preliminary signal storage and offline evaluation, preparing the signal data in advance for systematic analysis. By storing the received signal and conducting thorough evaluation using machine learning algorithms before final interpretation, the system ensures that all signal maxima are identified and processed with high precision, even in complex scenarios with multiple reflections.
Solution Approach 2:
The patent segments the received signal to identify and process each signal maximum individually. The evaluation unit calculates distance values for each identified signal maximum and assigns them to different layers (product layer, additive layer, foam layer). This segmentation approach maintains measurement speed by processing signals systematically while achieving high precision through careful analysis of each signal component.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The system reliably determines the fill level and additional layers by accurately interpreting complex signals, even under challenging conditions, thereby increasing measurement certainty and reliability.
Implementation Method 1
radar-based measuring methods have become established, since they are robust and low-maintenance
Implementation Method 2
the high-frequency signal in the measuring sensor is reflected at the fill level of the filling material surface and guided back along the measuring sensor towards the fill-level measuring device
Implementation Method 3
This variant of radar-based, fill-level measurement is also known by the term, 'TDR' (time-domain reflectometry)
Data Source
AI summary
A measuring system and a corresponding measuring method for reliably determining a fill level of a filling material in a container comprises a radar-based measuring device having a transmission unit via which high-frequency signals can be transmitted towards the filling material and received as received signals after reflection on the filling material surface; furthermore having a signal generation unit which generates the high-frequency signal to be transmitted; and having a receiving unit which records the received signal. In an evaluation unit of the measuring system, a machine learning algorithm is designed such that the fill level can be detected on the basis of the received signal. By the machine learning algorithm, received signals which are recorded under complex measuring conditions such as interference and multiple reflections can be interpreted much more reliably.