Autoencoder Validation for Time-Compressed Printing System Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing production printers face challenges in predicting performance due to varying and asynchronous data formats, making it difficult to maintain them effectively without unnecessary downtime and labor costs.
Innovation Solution
A system using an autoencoder to time-compress data from production printers, validated through partial F tests to ensure predictive power is preserved, allowing for efficient training of machine learning models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If preventive maintenance is performed regularly, then reliability is improved, but productivity deteriorates due to increased downtime
Solution Approach 1:
The system performs preliminary analysis of printer data using autoencoders and machine learning models to predict potential failures before they occur. This allows maintenance to be scheduled based on actual predicted needs rather than following a fixed preventive maintenance schedule, thereby maintaining reliability while minimizing unnecessary downtime and preserving productivity.
2Measurement precision
If data is collected at high frequency, then measurement precision is improved, but device complexity increases due to data management requirements
Solution Approach 1:
The autoencoder extracts only the most relevant features from the high-frequency printer data, separating the essential performance indicators from the vast amount of raw data. This extraction process maintains measurement precision by focusing on critical features while significantly reducing data processing complexity through dimensionality reduction and feature selection.
Solution Approach 2:
The system collects comprehensive high-frequency data but processes only a partial subset of this data through the autoencoder's feature extraction. By applying partial action (processing only relevant features rather than all collected data), the system maintains high measurement precision for critical parameters while avoiding the full computational burden of processing every data point, thus reducing device complexity.
3Loss of substance
If data is compressed to reduce storage, then loss of information may occur, but device complexity increases due to compression validation requirements
Solution Approach 1:
The system implements feedback mechanisms through validation tests that compare original and compressed data representations. The autoencoder's reconstruction process provides feedback on compression quality, and validation tests provide additional feedback to ensure predictive power is maintained. This feedback loop allows the system to achieve efficient data compression while managing complexity through automated validation rather than manual processes.
Solution Approach 2:
The autoencoder creates compressed representations (copies) of the original printer data that retain the essential predictive information. Rather than storing all original high-resolution data, the system uses these compressed copies for analysis and training, significantly reducing storage requirements while maintaining the ability to accurately predict printer performance and failures.
4Productivity
If machine learning models are trained with compressed data, then productivity is improved, but measurement precision may deteriorate
Solution Approach 1:
The system applies partial action by training machine learning models on compressed data representations that contain only the most relevant features extracted by the autoencoder. This partial representation is sufficient for maintaining prediction accuracy while dramatically improving training productivity. The validation tests ensure that this partial data representation does not compromise measurement precision beyond acceptable thresholds.
Data Source
AI summary
Systems and methods are provided for validating an autoencoder for a printing system. In one embodiment, a system stores values associated with individual features at a printing system and further stores times indicating when the values were determined at the printing system. The system also operates an autoencoder to perform time-compression upon the values and the times. Validating the autoencoder includes: determining predictive powers of the individual features prior to time-compression via partial F tests, determining predictive powers of the individual features after time-compression via partial F tests, and determining differences between the predictive powers of the individual features before time-compression and the predictive powers of the individual features after time-compression. Depending upon differences in the predictive powers, a report is generated indicating validity or invalidity of the autoencoder.