Fault models are used to evaluate the effect of hardware errors on the entire electrical/mechanical/structural system. A good fault model is expected to reflect the conduct of a physical defect correctly and should be computationally effective in fault simulation and generation of test patterns. According to Memon, when the fault model engenders false-positive results, the best approach is to employ different statistical quality control measures to scrutinize the results. The measures include;
- Check Sheet. I will use it to identify defects in the system. For instance, if there are 30 faulty model results, there are 30 defects, out of which there are major and minor defects that may be repeated if not rectified.
- Stratification. It involves converting the data assembled into a meaningful form for analysis. For example, if 40 faulty model results are generated in a day, stratification can break down the data into hourly defects. This will help indicate when most defects were produced that would be the best to remedy the system.
- Histogram. A histogram is a chart that can graphically represent the rate of occurrence of major and minor defects. It will be easy to spot the faults with the highest frequency.
- Pareto Chart. It is another visual illustration of the fault model errors. It helps to indicate the reason behind a small percentage of a major problem, say 30 percent, which could result in a higher rate of defects being produced by the fault model, say 85 percent.
- Cause and Effect diagram. It is also a visual illustration of the root cause of a problem and its effect on the system in producing defects.
- Scatter Diagram. The graphic illustration is drawn to show the relationship between the fault model results and the defects on an X and Y plot, where X is the independent variable and Y is the dependent variable.
- Control Chart. The visual depiction helps to show the organization can control the defects and work toward meeting the product or consumer expectations. It also indicates the upper tolerance and lower tolerance limits.