Choose your reading route
eduKateSG · Why Mathematics? · Observability and Identifiability for Sensor-Fault Isolation: Rank Tests, Fault Signatures, Redundancy and Sensor Placement
Start with the question, audit the worked example, or jump to the assumptions and practice. Every chapter stays open for uninterrupted reading.
Understand the question
Define the quantity and follow the mechanism.Route 2 · Chapters 6–10
Reproduce the example
Audit the numbers one step at a time.Route 3 · Chapters 11–15
Check assumptions
See where design and calibration matter.Route 4 · Chapters 16–20
Explore applications
Connect the method to a concrete investigation.Route 5 · Chapters 21–24
Study and verify
Practise, inspect failure cases and read further.
Full chapter index · Practice and checking · Mathematics Learning Hub
Why mathematics is important becomes unavoidable when a diagnostic system must decide whether its sensors contain enough independent information to reveal a hidden state or separate one fault from another. Observability asks whether internal state can be reconstructed from outputs over time. Fault identifiability asks whether distinct fault patterns produce distinguishable evidence.
For a two-state teaching system with A=[[1,1],[0,1]] and C=[1,0], the two-step observability matrix is [C;CA]=[[1,0],[1,1]]. Its determinant is 1 and rank is 2, so the state is observable in this ideal linear example. For fault isolation, signatures [1,0] and [0,1] are linearly independent and distinguishable in the chosen residual space; signatures [1,0] and [1,0] are identical and cannot be isolated from those outputs, no matter how confident an algorithm sounds.
Did You Know? More measurements do not guarantee more information. Two sensors that provide the same effective equation may add redundancy against random noise but no new direction for identification.
This article answers one intent: The mechanism uses matrix rank, fault signatures and redundancy structure to test whether states and faults are distinguishable before designing a detector.
Observability and identifiability test whether the available measurements can support the claim a detector will make. The small matrix example makes rank and signature independence visible. Later chapters add conditioning, redundancy, sensor placement, augmentation, common causes and unknown inputs. Students transfer matrix and vector reasoning; practitioners transfer the habit of testing information structure before tuning thresholds.
Full chapter index
Chapters 1–5 · Understand the question
Chapters 6–10 · Reproduce the example
Chapters 11–15 · Check assumptions
Chapters 16–20 · Explore applications
Chapters 21–24 · Study and verify
Begin with the object being measured, because a detector name is less useful than the statistic it changes. No detector can isolate causes that create identical observations. The importance of mathematics here is that the statement becomes a quantity or logical test that another reader can reproduce.
Two faults share one residual signature. Follow the order of operations, keep the time index and units, and check whether the direction makes sense before trusting the number.
Algorithm complexity cannot manufacture missing information. Test distinguishability early. Transfer the idea by changing one input at a time and explaining the direction before computing.
A useful audit begins by writing the claim in plain language, then identifying the observation, reference model and comparison rule. In this chapter, the central claim is: No detector can isolate causes that create identical observations. Test it first on a short sequence where every intermediate value can be inspected. Then change one condition and explain why the conclusion does or does not survive.
The mathematical value appears when a pattern is connected to one reproducible calculation. Represent dynamics x(k+1)=Ax(k)+Bu(k) and measurements y(k)=Cx(k). The importance of mathematics here is that the statement becomes a quantity or logical test that another reader can reproduce.
A describes state evolution and C maps states to sensors. Follow the order of operations, keep the time index and units, and check whether the direction makes sense before trusting the number.
Wrong structure invalidates rank conclusions. Document the model. Keep raw evidence beside the transformed statistic so diagnosis remains possible.
For transfer, sketch a before-and-after case and label what is measured, what is predicted and what is inferred. The concrete case here is: A describes state evolution and C maps states to sensors. That example is illustrative rather than evidence about a particular school, learner or machine. A sound explanation keeps calculation, interpretation and action in separate sentences.
Separate the mechanism from the interpretation so a number does not become unsupported certainty. Stack C,CA,…,CA^(n−1) for an n-state linear time-invariant system. The importance of mathematics here is that the statement becomes a quantity or logical test that another reader can reproduce.
The teaching matrix is [[1,0],[1,1]]. Follow the order of operations, keep the time index and units, and check whether the direction makes sense before trusting the number.
Numerical rank depends on tolerance and scaling. Inspect singular values. Compare with a simple baseline and an alternative cause before deciding.
Reliability work needs a counterexample as well as a successful example. Here the important boundary is: Numerical rank depends on tolerance and scaling. Ask what other mechanism could create the same observed pattern, what additional measurement would separate the explanations, and how uncertainty should be recorded. This turns a formula into a defensible decision process.
Keep timing, units and assumptions visible before applying a threshold. Full column rank n gives classical state observability. The importance of mathematics here is that the statement becomes a quantity or logical test that another reader can reproduce.
Rank 2 for two states passes the ideal test. Follow the order of operations, keep the time index and units, and check whether the direction makes sense before trusting the number.
The test is structural and model-specific. Validate operating points. Explain the key equation aloud and point to the term carrying each meaning.
A student can practise this idea without specialised equipment. Start with a tiny table, reproduce the stated operation, and predict how the output changes when one entry is doubled, delayed or removed. The practical next step is: Validate operating points. Explain the result aloud, including the assumption that would most change the answer.
Make the example auditable by predicting direction, calculating and checking a nearby case. For a square two-by-two observability matrix, determinant 1 confirms nonsingularity. The importance of mathematics here is that the statement becomes a quantity or logical test that another reader can reproduce.
1·1−0·1=1. Follow the order of operations, keep the time index and units, and check whether the direction makes sense before trusting the number.
Determinants are unstable diagnostics in larger systems. Use rank-revealing factorizations. Report observation, assumption, calculation and authorised action as separate layers.
A useful audit begins by writing the claim in plain language, then identifying the observation, reference model and comparison rule. In this chapter, the central claim is: For a square two-by-two observability matrix, determinant 1 confirms nonsingularity. Test it first on a short sequence where every intermediate value can be inspected. Then change one condition and explain why the conclusion does or does not survive.
Begin with the object being measured, because a detector name is less useful than the statistic it changes. Outputs over a time window contain enough information to distinguish initial states. The importance of mathematics here is that the statement becomes a quantity or logical test that another reader can reproduce.
Position measurements across time reveal velocity. Follow the order of operations, keep the time index and units, and check whether the direction makes sense before trusting the number.
Observability does not guarantee accurate estimation under noise. Assess conditioning. Transfer the idea by changing one input at a time and explaining the direction before computing.
For transfer, sketch a before-and-after case and label what is measured, what is predicted and what is inferred. The concrete case here is: Position measurements across time reveal velocity. That example is illustrative rather than evidence about a particular school, learner or machine. A sound explanation keeps calculation, interpretation and action in separate sentences.
| Object | Matrix or vectors | Result |
|---|---|---|
| Observability matrix | [[1,0],[1,1]] | determinant 1; rank 2 |
| Fault signatures | [1,0] and [0,1] | independent; isolable in this example |
| Coincident signatures | [1,0] and [1,0] | not distinguishable from these outputs |
The mathematical value appears when a pattern is connected to one reproducible calculation. The Gramian measures accumulated output sensitivity over a horizon. The importance of mathematics here is that the statement becomes a quantity or logical test that another reader can reproduce.
Small eigenvalues indicate weakly observed directions. Follow the order of operations, keep the time index and units, and check whether the direction makes sense before trusting the number.
Full rank can still be practically ill-conditioned. Report eigenvalues or condition numbers. Keep raw evidence beside the transformed statistic so diagnosis remains possible.
Reliability work needs a counterexample as well as a successful example. Here the important boundary is: Full rank can still be practically ill-conditioned. Ask what other mechanism could create the same observed pattern, what additional measurement would separate the explanations, and how uncertainty should be recorded. This turns a formula into a defensible decision process.
Separate the mechanism from the interpretation so a number does not become unsupported certainty. Choose measurements that add independent and well-conditioned information. The importance of mathematics here is that the statement becomes a quantity or logical test that another reader can reproduce.
A second sensor measures the hidden state directly. Follow the order of operations, keep the time index and units, and check whether the direction makes sense before trusting the number.
Cost, failure modes and environment constrain placement. Optimise with engineering limits. Compare with a simple baseline and an alternative cause before deciding.
A student can practise this idea without specialised equipment. Start with a tiny table, reproduce the stated operation, and predict how the output changes when one entry is doubled, delayed or removed. The practical next step is: Optimise with engineering limits. Explain the result aloud, including the assumption that would most change the answer.
Keep timing, units and assumptions visible before applying a threshold. Use relationships among sensors and model predictions to create residual checks. The importance of mathematics here is that the statement becomes a quantity or logical test that another reader can reproduce.
Three related measurements allow one inconsistent channel to be challenged. Follow the order of operations, keep the time index and units, and check whether the direction makes sense before trusting the number.
Redundancy is conditional on independent failure modes. Map common causes. Explain the key equation aloud and point to the term carrying each meaning.
A useful audit begins by writing the claim in plain language, then identifying the observation, reference model and comparison rule. In this chapter, the central claim is: Use relationships among sensors and model predictions to create residual checks. Test it first on a short sequence where every intermediate value can be inspected. Then change one condition and explain why the conclusion does or does not survive.
Make the example auditable by predicting direction, calculating and checking a nearby case. Describe how a specific fault affects residuals across channels and time. The importance of mathematics here is that the statement becomes a quantity or logical test that another reader can reproduce.
A bias in sensor one creates one signature vector. Follow the order of operations, keep the time index and units, and check whether the direction makes sense before trusting the number.
Magnitude and dynamics can change the pattern. Validate signatures experimentally. Report observation, assumption, calculation and authorised action as separate layers.
For transfer, sketch a before-and-after case and label what is measured, what is predicted and what is inferred. The concrete case here is: A bias in sensor one creates one signature vector. That example is illustrative rather than evidence about a particular school, learner or machine. A sound explanation keeps calculation, interpretation and action in separate sentences.
Begin with the object being measured, because a detector name is less useful than the statistic it changes. Distinct directions support isolation; identical or proportional signatures may not. The importance of mathematics here is that the statement becomes a quantity or logical test that another reader can reproduce.
[1,0] and [0,1] are independent. Follow the order of operations, keep the time index and units, and check whether the direction makes sense before trusting the number.
Near-collinear signatures are fragile under noise. Measure angular separation. Transfer the idea by changing one input at a time and explaining the direction before computing.
Reliability work needs a counterexample as well as a successful example. Here the important boundary is: Near-collinear signatures are fragile under noise. Ask what other mechanism could create the same observed pattern, what additional measurement would separate the explanations, and how uncertainty should be recorded. This turns a formula into a defensible decision process.
The mathematical value appears when a pattern is connected to one reproducible calculation. A fault must influence at least one monitored residual. The importance of mathematics here is that the statement becomes a quantity or logical test that another reader can reproduce.
A nonzero signature makes detection possible in the ideal model. Follow the order of operations, keep the time index and units, and check whether the direction makes sense before trusting the number.
Small effects can be buried in noise. Study signal-to-noise ratio. Keep raw evidence beside the transformed statistic so diagnosis remains possible.
A student can practise this idea without specialised equipment. Start with a tiny table, reproduce the stated operation, and predict how the output changes when one entry is doubled, delayed or removed. The practical next step is: Study signal-to-noise ratio. Explain the result aloud, including the assumption that would most change the answer.
Separate the mechanism from the interpretation so a number does not become unsupported certainty. Each target fault must be distinguishable from other relevant faults. The importance of mathematics here is that the statement becomes a quantity or logical test that another reader can reproduce.
Independent signatures support a two-fault teaching case. Follow the order of operations, keep the time index and units, and check whether the direction makes sense before trusting the number.
Detectable does not imply isolable. Use a confusion matrix. Compare with a simple baseline and an alternative cause before deciding.
A useful audit begins by writing the claim in plain language, then identifying the observation, reference model and comparison rule. In this chapter, the central claim is: Each target fault must be distinguishable from other relevant faults. Test it first on a short sequence where every intermediate value can be inspected. Then change one condition and explain why the conclusion does or does not survive.
Keep timing, units and assumptions visible before applying a threshold. Model bias faults as extra states and test augmented observability or identifiability. The importance of mathematics here is that the statement becomes a quantity or logical test that another reader can reproduce.
A constant sensor bias gains its own state. Follow the order of operations, keep the time index and units, and check whether the direction makes sense before trusting the number.
Augmentation can introduce unobservable modes. Run the rank test again. Explain the key equation aloud and point to the term carrying each meaning.
For transfer, sketch a before-and-after case and label what is measured, what is predicted and what is inferred. The concrete case here is: A constant sensor bias gains its own state. That example is illustrative rather than evidence about a particular school, learner or machine. A sound explanation keeps calculation, interpretation and action in separate sentences.
Make the example auditable by predicting direction, calculating and checking a nearby case. Multiple active faults create combined signatures. The importance of mathematics here is that the statement becomes a quantity or logical test that another reader can reproduce.
Two biases sum their effects. Follow the order of operations, keep the time index and units, and check whether the direction makes sense before trusting the number.
Combinations can mimic a third fault or state change. Define the fault catalogue. Report observation, assumption, calculation and authorised action as separate layers.
Reliability work needs a counterexample as well as a successful example. Here the important boundary is: Combinations can mimic a third fault or state change. Ask what other mechanism could create the same observed pattern, what additional measurement would separate the explanations, and how uncertainty should be recorded. This turns a formula into a defensible decision process.
Begin with the object being measured, because a detector name is less useful than the statistic it changes. Shared power, reference or environment can move several sensors together. The importance of mathematics here is that the statement becomes a quantity or logical test that another reader can reproduce.
Two sensors drift in the same direction. Follow the order of operations, keep the time index and units, and check whether the direction makes sense before trusting the number.
Assuming independence gives false reassurance. Include shared-cause models. Transfer the idea by changing one input at a time and explaining the direction before computing.
A student can practise this idea without specialised equipment. Start with a tiny table, reproduce the stated operation, and predict how the output changes when one entry is doubled, delayed or removed. The practical next step is: Include shared-cause models. Explain the result aloud, including the assumption that would most change the answer.
The mathematical value appears when a pattern is connected to one reproducible calculation. Loads and disturbances can occupy the same residual directions as faults. The importance of mathematics here is that the statement becomes a quantity or logical test that another reader can reproduce.
A force disturbance resembles actuator loss. Follow the order of operations, keep the time index and units, and check whether the direction makes sense before trusting the number.
Decoupling may reduce sensitivity to real faults too. Check unknown-input observability. Keep raw evidence beside the transformed statistic so diagnosis remains possible.
A useful audit begins by writing the claim in plain language, then identifying the observation, reference model and comparison rule. In this chapter, the central claim is: Loads and disturbances can occupy the same residual directions as faults. Test it first on a short sequence where every intermediate value can be inspected. Then change one condition and explain why the conclusion does or does not survive.
Separate the mechanism from the interpretation so a number does not become unsupported certainty. Use local linearisation, differential geometry or trajectory-based sensitivity for more complex models. The importance of mathematics here is that the statement becomes a quantity or logical test that another reader can reproduce.
Sensor informativeness changes with operating point. Follow the order of operations, keep the time index and units, and check whether the direction makes sense before trusting the number.
One rank result may not hold globally. Test across regimes. Compare with a simple baseline and an alternative cause before deciding.
For transfer, sketch a before-and-after case and label what is measured, what is predicted and what is inferred. The concrete case here is: Sensor informativeness changes with operating point. That example is illustrative rather than evidence about a particular school, learner or machine. A sound explanation keeps calculation, interpretation and action in separate sentences.
Keep timing, units and assumptions visible before applying a threshold. Identifiability analysis can prevent unjustified automatic isolation. The importance of mathematics here is that the statement becomes a quantity or logical test that another reader can reproduce.
An ambiguous signature triggers broader inspection. Follow the order of operations, keep the time index and units, and check whether the direction makes sense before trusting the number.
It does not choose the operational action alone. Combine evidence, consequence and authority. Explain the key equation aloud and point to the term carrying each meaning.
Reliability work needs a counterexample as well as a successful example. Here the important boundary is: It does not choose the operational action alone. Ask what other mechanism could create the same observed pattern, what additional measurement would separate the explanations, and how uncertainty should be recorded. This turns a formula into a defensible decision process.
Make the example auditable by predicting direction, calculating and checking a nearby case. The topic connects matrices, determinants, vectors, geometry and logical limits. The importance of mathematics here is that the statement becomes a quantity or logical test that another reader can reproduce.
A learner can compute the two-by-two rank and compare signatures. Follow the order of operations, keep the time index and units, and check whether the direction makes sense before trusting the number.
Full rank is not perfect certainty. Explain noise and conditioning. Report observation, assumption, calculation and authorised action as separate layers.
A student can practise this idea without specialised equipment. Start with a tiny table, reproduce the stated operation, and predict how the output changes when one entry is doubled, delayed or removed. The practical next step is: Explain noise and conditioning. Explain the result aloud, including the assumption that would most change the answer.
Begin with the object being measured, because a detector name is less useful than the statistic it changes. Specify A,C,fault signatures,horizon, rank tolerance, conditioning, common causes and validation data. The importance of mathematics here is that the statement becomes a quantity or logical test that another reader can reproduce.
Tests include each fault, combined faults and unknown disturbances. Follow the order of operations, keep the time index and units, and check whether the direction makes sense before trusting the number.
A rank statement without units and scaling is weak. Archive assumptions. Transfer the idea by changing one input at a time and explaining the direction before computing.
A useful audit begins by writing the claim in plain language, then identifying the observation, reference model and comparison rule. In this chapter, the central claim is: Specify A,C,fault signatures,horizon, rank tolerance, conditioning, common causes and validation data. Test it first on a short sequence where every intermediate value can be inspected. Then change one condition and explain why the conclusion does or does not survive.
The mathematical value appears when a pattern is connected to one reproducible calculation. Observability concerns state information; identifiability concerns distinguishing fault explanations; detection statistics decide from noisy evidence. The importance of mathematics here is that the statement becomes a quantity or logical test that another reader can reproduce.
The layers depend on one another. Follow the order of operations, keep the time index and units, and check whether the direction makes sense before trusting the number.
A good detector cannot repair non-identifiability. Order the design correctly. Keep raw evidence beside the transformed statistic so diagnosis remains possible.
For transfer, sketch a before-and-after case and label what is measured, what is predicted and what is inferred. The concrete case here is: The layers depend on one another. That example is illustrative rather than evidence about a particular school, learner or machine. A sound explanation keeps calculation, interpretation and action in separate sentences.
Separate the mechanism from the interpretation so a number does not become unsupported certainty. Adding a duplicate sensor can improve precision and availability yet leave rank unchanged. The importance of mathematics here is that the statement becomes a quantity or logical test that another reader can reproduce.
Duplicate the first row of C and recompute. Follow the order of operations, keep the time index and units, and check whether the direction makes sense before trusting the number.
More sensors is not the same as more independent equations. Ask what new direction appears. Compare with a simple baseline and an alternative cause before deciding.
Reliability work needs a counterexample as well as a successful example. Here the important boundary is: More sensors is not the same as more independent equations. Ask what other mechanism could create the same observed pattern, what additional measurement would separate the explanations, and how uncertainty should be recorded. This turns a formula into a defensible decision process.
Keep timing, units and assumptions visible before applying a threshold. NASA and aerospace research characterise sensor and actuator fault identifiability; structured residual research uses extended observability matrices for isolation. The importance of mathematics here is that the statement becomes a quantity or logical test that another reader can reproduce.
Recent eduKate articles cover MMAE and distributed filtering. Follow the order of operations, keep the time index and units, and check whether the direction makes sense before trusting the number.
Ideal rank tests require practical conditioning checks. Validate with representative faults. Explain the key equation aloud and point to the term carrying each meaning.
A student can practise this idea without specialised equipment. Start with a tiny table, reproduce the stated operation, and predict how the output changes when one entry is doubled, delayed or removed. The practical next step is: Validate with representative faults. Explain the result aloud, including the assumption that would most change the answer.
The nearest published companions are innovation whiteness tests, CUSUM monitoring, generalized likelihood-ratio fault detection and multiple-model adaptive estimation. The eduKateSG Mathematics Learning Hub, eduKateSingapore Mathematics World, eduKate Sengkang Mathematics Hub, Bukit Timah Tutor Mathematics Hub and eduKate Punggol Mathematics Pathway provide wider learning routes. Primary sources: https://ntrs.nasa.gov/citations/20140006008, https://doi.org/10.2514/1.62523, https://doi.org/10.1016/S0959-1524(01)00046-4.
Frequently asked questions
What is observability?
It is the ability, under a model, to reconstruct internal state from outputs over time.
What is fault identifiability?
It is the ability to distinguish one fault pattern from other relevant explanations.
Does full rank guarantee accurate estimation?
No. Poor conditioning and noise can make an observable system practically difficult to estimate.
Can identical fault signatures be isolated?
No. If they create the same monitored evidence, additional independent information is required.
Does adding a sensor always increase rank?
No. A duplicate measurement may add precision or redundancy without adding a new independent direction.
Extended practice laboratory
Practice case 1: Build observability matrix
Setup: Use A=[[1,1],[0,1]], C=[1,0].
Checking target: Get [[1,0],[1,1]].
Practice case 2: Compute determinant
Setup: Use the two-by-two matrix.
Checking target: Confirm determinant 1.
Practice case 3: Change sensor
Setup: Set C=[0,1].
Checking target: Check rank.
Practice case 4: Duplicate sensor
Setup: Add another [1,0] row.
Checking target: Observe limited new direction.
Practice case 5: Compare signatures
Setup: Use [1,0] and [0,1].
Checking target: Confirm independence.
Practice case 6: Coincident signatures
Setup: Use [1,0] twice.
Checking target: Explain non-isolability.
Practice case 7: Near collinearity
Setup: Use [1,0] and [1,.01].
Checking target: Inspect conditioning.
Practice case 8: Augment bias
Setup: Add a constant bias state.
Checking target: Rebuild the matrix.
Practice case 9: Unknown input
Setup: Add a disturbance signature.
Checking target: Check confusion.
Practice case 10: Decision memo
Setup: Report observable state but ambiguous faults.
Checking target: Recommend sensor redesign.