Obsolescence Forecasting Using Bayesian Neural Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current asset management systems are inadequate in forecasting asset obsolescence due to the large volume and variety of rapidly changing asset-related information, inability to correlate asset features with relevant data, and the complex nature of assets and supply chains, leading to inaccurate predictions and unsatisfactory replacement plans.
Innovation Solution
A product obsolescence forecast system that uses machine instructions to access and analyze data from various sources, including structured, semi-structured, and unstructured data, applies probabilistic models like Bayesian neural networks to forecast obsolescence dates, and determines impacts on systems, including identifying replacement products and scheduling availability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional asset management systems are used to track asset information, then basic asset tracking is possible, but the systems cannot handle large volume and variety of rapidly changing asset-related information, leading to inadequate forecasting capability
Solution Approach 1:
The patent segments asset management into multiple specialized modules: data collection module for gathering information from diverse sources, data processing module for cleaning and standardizing data, probabilistic modeling module for forecasting obsolescence dates, and impact analysis module for assessing system effects. This segmentation allows each module to handle specific aspects of the complex task, improving forecasting reliability without overwhelming a single system
Solution Approach 2:
The patent introduces probabilistic models as intermediaries between raw asset data and forecasting outcomes. These models process uncertain and varied asset information through Bayesian neural networks that incorporate multiple data sources and uncertainties, transforming complex, unstructured data into reliable obsolescence date predictions while managing system complexity
2Measurement precision
If comprehensive data from multiple sources is collected to improve forecast accuracy, then prediction reliability increases, but the volume and variety of data to be processed increases significantly
Solution Approach 1:
The patent extracts only the most relevant features and data elements from comprehensive asset information using the probabilistic model. The Bayesian neural network identifies and extracts key predictive features from large volumes of diverse data, separating signal from noise and focusing computational resources on the most informative aspects of asset data for accurate obsolescence forecasting
Solution Approach 2:
The patent transforms raw asset data into standardized parameters and features suitable for probabilistic modeling. The system converts diverse data formats and structures into uniform parameters that the Bayesian neural network can process efficiently, changing the representation of data to optimize both accuracy and processing efficiency
3Reliability
If probabilistic models are applied to forecast obsolescence dates, then accurate predictions can be made, but the complexity of analyzing and acting on the results increases
Solution Approach 1:
The patent implements feedback loops where probabilistic forecast results automatically trigger impact analysis and replacement planning processes. The system feeds obsolescence date predictions back into the asset management workflow, where they automatically initiate assessments of system impacts and generation of replacement recommendations, making the complex probabilistic outcomes actionable through automated feedback mechanisms
Solution Approach 2:
The patent performs preliminary impact analysis and replacement planning based on probabilistic forecasts before actual obsolescence occurs. By anticipating obsolescence dates and proactively analyzing system impacts and preparing replacement plans in advance, the system reduces the complexity of implementing replacements by having preliminary assessments ready before critical decisions are needed
4Adaptability or versatility
If the system determines impacts on systems and supply chains, then comprehensive asset management is achieved, but the time and resources required for analysis increase
Solution Approach 1:
The patent performs preliminary impact analysis by evaluating system dependencies and supply chain relationships in advance of actual obsolescence events. The system pre-assesses which systems and supply chain components will be affected by potential obsolescence, preparing impact assessments beforehand so that when obsolescence occurs, comprehensive analysis is already available or can be quickly completed
Data Source
AI summary
A product obsolescence forecast system includes machine instructions stored in a non-transitory computer readable storage medium, the machine instructions. A processor executes the instructions to receive an identity of a first product and identities of one or more second products similar to the first product, each of the second products having gone obsolete; receive a determinant of obsolescence of one or more of the obsolete second products; generate one or more observations related to the first product by inputting each received determinant to a trained network; and generate a statistical analysis of the one or more observations to provide an expected value of an actual obsolescence date for the first product.