Multihead Weigher Signal Estimation for Faster Stable Weighing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing signal processing methods for multihead weighers are slow to react to changes in weight signals due to interference variables, leading to prolonged measurement times and reduced throughput, especially in combination scales where accurate weight determination is critical for quick product handling.
Innovation Solution
A signal processing method involving repeated sampling of weight signals, preprocessing to extract key parameters, and using artificial neural networks to estimate the weight signal, which allows for faster and more accurate determination of the weight signal, even in the presence of interference, by providing an early estimate of the weight signal stability and confidence in accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional filtering methods are used to eliminate disturbances in weight signals, then measurement accuracy is improved, but measurement time increases significantly
Solution Approach 1:
The system performs preliminary classification of signal fluctuations into disturbance categories (periodic, aperiodic, high-frequency noise) and applies appropriate filtering strategies in advance. By pre-identifying the nature of disturbances and preparing corresponding filter configurations, the system can quickly switch to the optimal filtering approach without extensive processing, thus reducing measurement time while maintaining accuracy
Solution Approach 2:
The filtering system is made dynamic and adaptive rather than static. The system continuously monitors signal characteristics and automatically adjusts filter parameters based on the detected disturbance type. This dynamic adaptation allows the system to use minimal filtering (and thus minimal time) for clean signals while applying stronger filtering only when disturbances are detected, resolving the contradiction between accuracy and speed
2Reliability
If strong filtering is applied to suppress disturbances, then measurement reliability is improved, but response speed to weight changes decreases
Solution Approach 1:
Different filtering strengths are applied to different frequency components and disturbance types locally. Instead of using a uniform strong filter across all signal characteristics, the system applies targeted filtering: strong suppression for identified periodic disturbances, moderate filtering for aperiodic disturbances, and minimal filtering for high-frequency noise. This localized approach maintains reliability for specific disturbance types while preserving response speed for legitimate weight changes
Solution Approach 2:
The system dynamically changes filter parameters (cutoff frequencies, filter orders, attenuation levels) based on the detected disturbance characteristics. When periodic disturbances are detected, parameters are adjusted to target those specific frequencies; when aperiodic disturbances occur, different parameters are applied. This parameter adaptation allows the system to maintain high reliability only when and where disturbances are present, rather than constantly using strong filtering that would reduce response speed
3Measurement precision
If measurement time is extended to achieve accurate weight determination, then throughput is reduced, but measurement accuracy is improved
Solution Approach 1:
The system performs preliminary analysis of the weight signal to determine the nature and magnitude of disturbances before committing to extended measurement times. By pre-assessing signal quality and disturbance levels, the system can make an early decision about whether extended measurement is necessary, thus avoiding unnecessary delays for clean signals while applying extended measurement only when accuracy is compromised by disturbances
Solution Approach 2:
The measurement process is made dynamic with adaptive time extension. Rather than using a fixed extended measurement period for all cases, the system continuously evaluates signal stability and disturbance levels, extending measurement time only as long as necessary to achieve required accuracy. This dynamic time adaptation maximizes throughput by minimizing measurement time for stable signals while ensuring accuracy is achieved for disturbed signals
Data Source
Figure 1
Figure 2a~2b
Figure 2c
AI summary
The present invention relates to a signal processing method for weight signals (W) from scales, in particular combination scales (K). The signal processing is carried out using pre-processed discrete values (W(i)) of the weight signal (W), which are supplied to at least one artificial neural network. Using this at least one artificial neural network, an estimated value (SW) for the actual weight is determined, for example, in a weighing unit of a combination scale. This is done faster than if one were to wait for the actual weight signal. The estimated values (SW) are passed on to the combination scale (KW), which uses them for combination calculations.