Heterogeneous GAN Training for Monitoring-Station Error Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing air quality and weather prediction models suffer from measurement errors introduced by monitoring stations, leading to inaccurate predictions due to limited station coverage and sensor inconsistencies, which are magnified during data learning.
Innovation Solution
A training method for a heterogeneous generative adversarial network model that includes a generator and discriminator, utilizing measurement data from multiple types of stations to perform joint prediction, with a total objective function that includes a first objective function for the generator and second objective functions for the discriminator to enhance accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If measurement data from monitoring stations is used for prediction, then prediction capability is provided, but measurement errors are introduced that reduce prediction accuracy
Solution Approach 1:
The patent introduces a Generative Adversarial Network (GAN) as an intermediary system between the raw measurement data and the prediction output. The GAN includes a generator that creates corrected measurement data and a discriminator that validates the corrections, effectively mediating the transformation of error-prone original data into accurate prediction inputs without directly using the flawed measurement data
2Area of stationary object
If more monitoring stations are deployed to improve coverage, then spatial coverage is increased, but system complexity and cost increase
Solution Approach 1:
The patent creates virtual copies of monitoring station data through the GAN-generated synthetic measurements. Instead of deploying physical monitoring stations to every location, the system generates virtual measurement data that replicates what additional stations would measure, providing extended spatial coverage without the complexity and cost of physical deployment
3Productivity
If traditional deep learning models are used for prediction, then prediction function is achieved, but measurement errors are magnified during learning process
Solution Approach 1:
The patent implements a feedback mechanism through the adversarial training process where the discriminator provides continuous feedback to the generator about the quality of generated measurements. This feedback loop allows the system to iteratively improve the generated data quality and prevent error magnification by constantly validating and correcting the learning process
Data Source
AI summary
Provided are a training method and device for a heterogeneous generative adversarial network model, an equipment, a program and a storage medium. In the training method, measurement data of a heterogeneous station is acquired, the measurement data of the heterogeneous station is set as a training sample, and joint training is performed on the heterogeneous generative adversarial network model according to a total objective function. A generator is configured to predict environment data at a future occasion according to environment data of the heterogeneous station at a historical occasion so as to output predicted data. A discriminator is configured to be input the predicted data output by the generator and corresponding measurement data, and discriminate a similarity between the measurement data and the predicted data; a total objective function includes a first objective function of the generator and a second objective function of the discriminator.