Conveyance Traffic Prediction Using Vehicle Log Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional conveyance systems in semiconductor manufacturing plants cannot predict the degree of future traffic congestion accurately, especially before an event occurs, limiting effective conveyance control and optimization.
Innovation Solution
A prediction device using machine-learning to analyze historical log data on conveyance vehicle assignments and positions, allowing for the prediction of future traffic congestion by training a model to forecast the number of conveyance vehicles in a target area, taking into account various scenarios such as vehicle locations and travel routes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional event-based prediction methods are used, then traffic congestion can be detected after events occur, but prediction cannot be performed before events occur or at any given timing
Solution Approach 1:
The system performs preliminary actions by continuously collecting and storing log information about conveyance vehicle operations, transport commands, and system states before congestion events occur. This historical data is prepared in advance to enable predictions at any future timing, not just after events happen.
Solution Approach 2:
The patent replaces conventional event-based mechanical prediction triggers with a machine-learning-based intelligent system. The ML model analyzes patterns in historical log data to predict congestion probability, replacing the need for specific congestion events to trigger predictions.
2Measurement precision
If machine-learning models are trained on comprehensive log information, then prediction accuracy improves, but system complexity and data processing requirements increase
Solution Approach 1:
The system extracts only the necessary features from comprehensive log information that are relevant for congestion prediction. The ML model is trained to identify and utilize key patterns in transport commands, vehicle positions, and system states, filtering out unnecessary data complexity while maintaining prediction accuracy.
Solution Approach 2:
The prediction system is designed with multi-functionality to handle various prediction scenarios using a single unified ML model. The model can predict congestion for different target areas, time periods, and congestion conditions, reducing the need for multiple specialized systems and thereby reducing overall complexity.
Data Source
AI summary
A prediction device include: a storage unit storing a prediction model trained by machine-learning to receive the input data and output the output data, the input data being based on log information of a plurality of conveyance vehicles during a first period prior to a reference point in time, the output data indicating a prediction result of a degree of increase or decrease in the number of conveyance vehicles in a target area during a second period subsequent to the reference point in time; an acquisition unit acquiring data for prediction, based on the log information during a past period prior to a prediction execution point in time; and a prediction unit acquiring prediction information indicating a prediction result of a degree of increase or decrease in the number of conveyance vehicles in the target area during a future period by entering the data for prediction into the prediction model.