Machine State Prediction From Event Clustering for Industrial Maintenance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Industrial machines face challenges in predicting failures due to constraints related to data quality and quantity, real-time processing needs, accuracy of data processing, availability for maintenance, and the difficulty for human operators to decode event codes quickly, leading to inefficiencies in supervising and maintaining these machines.
Innovation Solution
A computer-implemented method generates a prediction model by clustering event features from industrial machine data into vector clusters, assigning them to machine states, and providing transition probabilities, which can be used to predict the behavior of machines under supervision, optimizing processing times and improving efficiency and effectiveness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If human operators manually decode event codes to supervise machine operation, then data quality and quantity can be processed, but recognition time and reaction time are insufficient to prevent collisions or failures
Solution Approach 1:
The patent replaces the mechanical human operator decoding process with an automated computer system that processes event codes. The computer system uses algorithms to automatically interpret event codes, identify machine states, and predict failures, eliminating the time delay associated with manual decoding while maintaining or improving accuracy.
Solution Approach 2:
The patent introduces an intermediate computer system that acts as a mediator between the machine's event code generation and the operator's decision-making. This intermediary automatically decodes event codes, determines machine states, and provides predictions to operators, bridging the time gap between event occurrence and operator response.
2Loss of time
If computers process event data in real-time with minimized delays, then reaction time is improved, but processing complexity and computational requirements increase
Solution Approach 1:
The patent segments the complex processing task into distinct modules: event code decoding, machine state determination, and failure prediction. Each module handles a specific aspect of the processing, allowing for optimized computation at each stage and reducing overall complexity while maintaining real-time performance.
Solution Approach 2:
The patent changes the processing parameters by using predefined thresholds and state transition rules that simplify real-time computation. Instead of complex continuous analysis, the system uses discrete parameter comparisons and state machine approaches that can be executed rapidly with minimal computational resources.
3Ease of repair
If the machine is stopped for maintenance at any time, then maintenance can be performed, but technical resources are wasted and productivity is reduced
Solution Approach 1:
The patent performs preliminary failure prediction by continuously analyzing event codes and determining machine states before actual failures occur. This allows maintenance to be scheduled in advance during planned downtime periods rather than requiring immediate machine stops, enabling proactive maintenance that preserves productivity while ensuring maintenance needs are met.
Solution Approach 2:
The patent implements a feedback loop where the computer system continuously monitors machine operation, predicts potential failures, and provides early warnings to operators. This feedback enables maintenance to be planned and scheduled at optimal times rather than reacting to actual failures, balancing maintenance needs with productivity requirements.
4Reliability
If the computer stops the machine on incorrect data processing (false alarms), then safety is improved, but productivity is reduced due to unnecessary interruptions
Solution Approach 1:
The patent applies different levels of confidence thresholds to different machine states and failure predictions. Not all predictions trigger immediate machine stops; instead, the system uses localized quality control where only high-confidence predictions with significant risk levels trigger protective actions, while lower-confidence predictions are monitored without causing interruptions.
Solution Approach 2:
The patent uses a graduated response system where not all predicted failures result in immediate machine stops. Instead, the system applies partial actions such as warnings, monitoring intensification, or scheduled maintenance for lower-confidence predictions, reserving full machine stops for high-confidence, high-risk predictions, thereby reducing false alarm interruptions while maintaining safety.
Data Source
AI summary
Disclosed is a computer-implemented method for generating a prediction model. The model can be for use in processing machine event data to predict behavior of a plurality of industrial machines under supervision. The prediction model can be configured to determine current and future states of the industrial machines. The method can include: extracting event features from event codes and structuring the event features into feature vectors; and generating the prediction model by clustering the feature vectors into a plurality of vector clusters, the vector clusters being assigned to respective machine states. The prediction model can be constructed based on event data from a first industrial machine and be applied to control an operating state of a second industrial machine.