Q1. How is prediction accuracy verified?  

Ans. After the prediction by, it is verified by a capacity testing.  This is a very time-consuming process, but the IDs of the batteries (DUTs) are linked and matched.

Q2. In which situations should be used?    

Ans.  Many are used for pre-screening.

• •  Relatively easy to implement in the field, it can be employed in a large number of dispersed locations.
    

  •  Capacity testing involves load testing for quality assurance and is often suitable for centralised operations.
    

  • As a total business design, depending on the prediction accuracy of the pre-screening, the idea of supplementing it with a free replacement within the warranty period makes economic sense. This means that the debate is whether the prediction accuracy should be 95.5% (2σ) or 99.7% (3σ).　

Q3.  Is the prediction accuracy the correlation coefficient (R²)?    

Ans.  The correlation coefficient is a straightforward way of assessing the goodness of fit of a prediction model to the sample as a whole.  In real business, however, profitability is only related to the accuracy of the predictions at the thresholds that determine the pass/fail criteria in pre-screening.

• • For this reason, we present not only the correlation coefficient, but also the prediction accuracy for each threshold value, applying the idea of Bayesian probability.
    

  • Several threshold values can be set.
    

Q4.  How should threshold values be set?   

Ans. It is up to your marketing strategy.

• • Performance trends can be monitored for each threshold value. Information is at your hand on how much the threshold value should be to maximise revenues.
    

  • Simulations can be performed on how to divide the grades of remanufactured battery packs according to market demand.
    

Q5, What is on-board diagnostics?

• • Diagnosis of the degree of deterioration of the drive battery while the vehicle is still on board.

  • This allows batteries to be evaluated in car rental and used car assessments.