1. Definition of Control State Recognition and Decision

Control State Recognition and Decision is the control domain that uses sensors, actuators, communication devices, and internal control information to recognize the actual condition of the machine and determine whether that condition is normal, whether the next operation may proceed, or whether adjustment, stopping, or recovery is required.

A machine cannot operate reliably by executing commands alone.

It must continuously answer questions such as:

• Did the device actually move as commanded?

• Is the product in the required position?

• Has the target position, speed, or force been reached?

• Is the machine ready to execute the next function?

• Can the current signal be trusted?

• Are multiple signals consistent with one another?

• Has the machine deviated from normal conditions?

• Should operation continue, be restricted, or stop?

• Can the current machine state be regarded as known?

The essence of state recognition and decision control is therefore not simply reading signals.

The key is to convert physical conditions into meaningful machine states and determine the next action based on those states.

2. Position within the Control System

The control flow of general machinery can be expressed as follows:

Function Execution → Physical Change → Signal Detection → State Recognition → State Decision → Next Command or Response

The role of each area is as follows:

A sensor detects a physical quantity, but the sensor signal itself does not necessarily represent the actual machine state.

For example, the fact that a proximity sensor is ON does not always mean that the product is correctly positioned.

• The sensor may be contaminated.

• The product may be jammed in an abnormal orientation.

• The sensor bracket may have shifted.

• Another object may have activated the sensor.

Detection and state recognition must therefore be distinguished.

3. Distinction Between Signal, State, and Decision

3.1 Signal

A signal is raw information provided by a sensor or control device.

Examples:

• Sensor ON/OFF

• Encoder position value

• Pressure: 0.48 MPa

• Temperature: 82°C

• Servo Ready

• Inverter Fault

• Communication data received

• Safety-door switch status

3.2 State

A state is a meaningful machine condition defined by interpreting one or more signals.

Examples:

• Product arrival complete

• Clamp normal

• Axis stopped

• Ready for processing

• Insufficient pressure

• Position uncertain

• Suspected product jam

• Automatic operation permitted

3.3 Decision

A decision determines how the control system should respond to the recognized state.

Examples:

• Permit the next function

• Maintain the current operation

• Reduce speed

• Retry

• Generate a warning

• Perform a normal stop

• Perform a fault stop

• Perform a safety stop

• Request operator confirmation

A signal is the input, a state is the interpreted result, and a decision is the control response to that state.

4. Hierarchy of State Recognition

Machine states are not defined at only one level.

In experienced control design, states are divided into several layers.

4.1 Physical State

The actual physical quantity or condition.

• Position

• Speed

• Force

• Pressure

• Temperature

• Current

• Vibration

• Product presence

4.2 Device State

The current operating condition of an individual device.

• Motor stopped

• Servo Ready

• Cylinder forward complete

• Valve actuated

• Heater operating

• Brake applied

• Sensor fault

4.3 Functional State

The progress state of a specific functional unit.

• Waiting for feeding

• Positioning in progress

• Clamping complete

• Processing in progress

• Inspection complete

• Discharge failed

4.4 Process State

The progress condition of the product or overall process.

• Product not loaded

• Ready for processing

• Process in progress

• Inspection passed

• Rework required

• Process interrupted

4.5 Machine State

The operating condition of the entire machine.

• Initialization

• Ready

• Automatic operation

• Manual operation

• Paused

• Fault stop

• Recovery

• Maintenance

4.6 Safety State

The condition related to hazardous motion and safety functions.

• Safety conditions satisfied

• Hazardous-area access detected

• Safety stop in progress

• Safety output disabled

• Stop confirmation complete

• Restart prohibited

These state layers must be connected.

For example, the ON signal from a cylinder forward sensor represents a device state. If normal pressure and the correct product-position signal are also confirmed, the functional state may be recognized as Clamping Complete.

5. Definition of State Variables

A state variable is a value used to determine the current condition of the machine.

5.1 Discrete State Variables

These have a limited number of possible values.

Examples:

• ON/OFF

• Open/Closed

• Ready/Not Ready

• Present/Absent

• Normal/Fault

• Automatic/Manual/Maintenance

5.2 Continuous State Variables

These are expressed as continuous numerical values.

Examples:

• Position

• Speed

• Temperature

• Pressure

• Torque

• Current

• Vibration value

5.3 Derived State Variables

A derived state variable is generated by combining one or more signals.

Examples:

• If the position is within the target range and the speed is below the stop threshold, recognize Position Reached.

• If the pressure is above the required level and maintained for a defined time, recognize Clamp Stable.

• If motor current increases while there is no position change, recognize Suspected Product Jam.

• If both door sensors are normal and the safety relay is Ready, recognize Safety Door Normal.

Derived states are highly important in actual control decisions.

It is preferable for the controller to operate based on meaningful states rather than individual raw signals.

6. Direct State Recognition and Indirect State Estimation

6.1 Direct State Recognition

The final target condition itself is measured directly.

Examples:

• Detecting table position with a linear scale

• Directly confirming product presence with a photoelectric sensor

• Measuring clamping force with a load cell

• Measuring rotational speed with a speed sensor

Direct detection provides higher state reliability, but may increase sensor cost and installation complexity.

6.2 Indirect State Estimation

The state is estimated using other signals.

Examples:

• Estimating table position from a motor-shaft encoder

• Estimating contact from motor current

• Estimating clamping state from cylinder pressure

• Estimating cylinder position from valve output and elapsed time

Indirect estimation can simplify the structure, but has the following limitations:

• Transmission-mechanism failure may not be detected.

• The result may be affected by load and friction variation.

• The actual final result may differ from the estimate.

• Errors may accumulate.

For critical functions, the final result should be verified directly whenever possible.

7. Types of Signals Required for State Recognition

7.1 Position and Displacement Signals

• Proximity sensors

• Limit switches

• Photoelectric sensors

• Linear scales

• Rotary encoders

• Displacement sensors

7.2 Speed and Motion Signals

• Encoder speed

• Rotation sensors

• Travel time

• Axis following error

• Motion-controller status

7.3 Force and Load Signals

• Load cells

• Torque sensors

• Motor current

• Servo load ratio

• Cylinder pressure

7.4 Thermal and Process Signals

• Temperature

• Pressure

• Flow rate

• Vacuum

• Weight

• Level

• Tension

7.5 Product and Environmental Signals

• Product presence

• Product orientation

• Product size

• Door state

• Material supply condition

• Cooling-water condition

• Lubrication condition

7.6 Internal Control-Device Signals

• Drive Ready

• Servo On

• In Position

• Torque Limit

• Alarm

• Communication Healthy

• Safety Ready

State recognition should use not only external sensors but also internal information from drives and controllers.

8. Signal Validity Verification

A received sensor value should not be accepted immediately as a valid state.

The signal must first be verified.

8.1 Range Check

Verifies whether the measured value is within a physically possible range.

Examples:

• Does the pressure exceed the sensor measurement range?

• Is the temperature abnormally low or high?

• Is the encoder value outside the axis travel limit?

8.2 Rate-of-Change Check

Verifies whether the speed of change is physically possible.

Examples:

• Temperature changes by hundreds of degrees in one second.

• Position changes suddenly while the axis is stopped.

• Pressure changes at an unrealistic rate.

8.3 Temporal Consistency

Verifies whether the signal remains active for the required duration.

Examples:

• Does the door-closed signal remain active for at least the minimum time?

• Is the position-reached signal only momentary?

8.4 Cross Consistency

Verifies whether related signals are logically consistent.

Examples:

• Are both forward and reverse sensors ON simultaneously?

• Is encoder speed detected while the motor-stop signal is active?

• Does the cylinder continue moving while the valve is OFF?

8.5 Command–Response Consistency

Verifies whether the actual response matches the execution command.

Examples:

• After a forward command, does the position change in the forward direction?

• After heater output is applied, does the temperature rise?

• After a motor-stop command, does the speed decrease?

9. Signal Filtering and Stabilization

Physical sensor signals may contain noise, chattering, and momentary fluctuations.

9.1 Debouncing

Removes rapid ON/OFF repetition from mechanical contacts or digital sensors.

9.2 On-Delay and Off-Delay

Recognizes a state change only after the signal remains active or inactive for a defined time.

9.3 Averaging and Filtering

Reduces momentary variations in analog signals.

• Moving average

• Low-pass filter

• Median filter

• Outlier removal

9.4 Hysteresis

Prevents repeated state changes around a threshold.

Example:

• Pressure recognized as normal at 0.50 MPa or higher

• Pressure recognized as insufficient at 0.45 MPa or lower

Filtering should reduce noise without delaying important abnormality detection.

10. State Certainty and Uncertainty

A machine condition cannot always be classified clearly as either Normal or Abnormal.

There are also states in which the actual condition cannot be trusted.

10.1 Confirmed State

The required signals are consistent and the state has been clearly verified.

10.2 Unconfirmed State

The required information is not yet sufficient, or completion has not yet been confirmed.

10.3 Unknown State

The actual state is unknown because of power recovery, communication loss, loss of position information, or a similar event.

10.4 Conflicting State

Related signals contradict one another.

Examples:

• Both forward and reverse completion signals are ON.

• Door Open and Door Closed signals occur simultaneously.

• The axis-position value indicates stop while a speed signal is present.

10.5 Estimated State

The state cannot be verified directly and is estimated from other signals.

Critical operations should not begin automatically in Unknown, Conflicting, or Estimated states.

11. Basic Structure of State Decision

State decision generally follows this sequence:

Signal Acquisition → Signal Validation → State Generation → State Classification → Priority Evaluation → Control Decision

11.1 Signal Acquisition

Collects data from sensors, drives, PLCs, and external equipment.

11.2 Signal Validation

Verifies range, timing, rate of change, and consistency.

11.3 State Generation

Combines valid signals to create meaningful states.

11.4 State Classification

Classifies states as normal, caution, warning, fault, or safety-related.

11.5 Priority Evaluation

Determines which response has priority when multiple decisions occur simultaneously.

11.6 Control Decision

• Continue operation

• Transition to the next step

• Adjust output

• Restrict function

• Retry

• Stop

• Perform a safety action

12. Decision Methods

12.1 Threshold Decision

Determines the state by comparing a measured value with a set range.

Examples:

• Generate a warning at 90°C or higher.

• Prohibit operation below 0.4 MPa.

• Recognize completion when position error is within ±0.2 mm.

12.2 Logical Decision

Determines the state by combining multiple digital conditions.

Example:

Product Present AND Clamp Complete AND Door Closed

12.3 Temporal Decision

Includes signal order and duration in the judgment.

Examples:

• Completion must occur within two seconds after the command.

• Pressure must be maintained for at least one second.

• Two signals must become active within 100 ms of each other.

12.4 State-Transition Decision

Determines the next state based on the current state and input conditions.

Examples:

• Ready + Start Command → Automatic

• Automatic + Timeout → Fault Stop

• Fault Stop + Reset + Condition Verified → Recovery

12.5 Trend-Based Decision

Uses not only the current value but also the direction and pattern of change.

Examples:

• Continuous increase in bearing temperature

• Gradual increase in cycle time

• Long-term increase in motor current

• Repeated increase in position error

13. Operation-Permission Decisions

State-recognition results are used as permission conditions for function execution.

13.1 Start Permission

Determines whether a function may begin.

Examples:

• Device Ready

• Previous function complete

• Product condition normal

• Safety conditions satisfied

13.2 Continuation Permission

Determines whether the current function may continue.

Examples:

• Pressure remains normal

• Product position has not shifted

• No overload condition

• No access to the hazardous area

13.3 Completion Permission

Determines whether the function may be recognized as successfully completed.

Examples:

• Target position reached

• Pressure maintained

• Product result verified

• Completed within the allowable time

13.4 Next-Step Permission

Determines whether the system may transition to the next function.

Examples:

• Current function complete

• Next device Ready

• Product-transfer path normal

• Interlock conditions satisfied

14. Control Decisions According to State

14.1 Continue

The current function continues because normal conditions are maintained.

14.2 Transition

The current function is complete, and the system moves to the next function or state.

14.3 Adjust

Output is adjusted to reduce the difference between the target and actual values.

14.4 Hold

The current operation is suspended until the required condition is restored.

14.5 Restrict

Speed, output, or permitted functions are limited.

14.6 Retry

The same function is executed again under safe and limited conditions.

14.7 Fault Stop

Normal function execution is no longer possible, so operation stops and operator intervention is required.

14.8 Safety Stop

A safety-related stop is performed to limit risk to personnel and equipment.

15. Classification of State Severity

Not all abnormal states should be handled in the same way.

Severity should not be determined by the magnitude of the measured value alone.

The following should also be considered:

• Effect on machine function

• Effect on product quality

• Possibility of equipment damage

• Risk to personnel

• Recoverability

• Reliability of the recognized state

16. State Decision Using Multiple Signals

Critical states should preferably not be determined from only one signal.

16.1 Redundant Detection of the Same State

Two or more sensors confirm the same condition.

Examples:

• Two door sensors

• Two position sensors

• Encoder and limit sensor

16.2 Cross-Verification Using Different Physical Quantities

Different types of signals are combined to determine the state.

Examples:

• Position + pressure to determine clamping complete

• Speed + brake feedback to determine stopped

• Motor current + position change to determine jamming

• Temperature + heater output to determine heating failure

16.3 Comparison of Command, Response, and Result

The following three elements are compared:

• What command was issued?

• How did the device respond?

• Was the final result achieved?

This comparison is effective for early detection of transmission failure and device malfunction.

17. State Decision According to Operating Mode

The same signal may be interpreted differently depending on the operating mode.

17.1 Automatic Operation

• All function-completion conditions applied

• Strict process interlocks applied

• Automatic state transition

• Automatic timeout evaluation

17.2 Manual Operation

• Focus on individual device states

• Automatic state transition restricted

• Essential collision-prevention conditions maintained

• State continuously monitored during operator control

17.3 Setup Operation

• State reevaluated after setting changes

• Limited speed and limited stroke applied

• State verified step by step

17.4 Recovery Mode

• Reliability of existing position information reevaluated

• Incomplete function state checked

• Product and tool positions verified

• Re-entry into automatic operation evaluated

17.5 Maintenance Mode

• Some standard conditions may be bypassed

• Safety and restriction conditions maintained

• Device states displayed directly

• Bypass state recorded

18. Event-Based and Continuous State Decision

18.1 Event-Based Decision

A decision is made when a specific signal change occurs.

Examples:

• Product arrival

• Button input

• Completion signal

• Alarm occurrence

• Door opening

18.2 Continuous State Decision

A state value is continuously monitored.

Examples:

• Position error

• Temperature

• Pressure

• Current

• Speed

• Vibration

Event-based decisions are suitable for sequence transitions, while continuous decisions are suitable for feedback control and abnormality monitoring.

Actual machines use both methods together.

19. Abnormal States and Diagnostics

State recognition should not only detect abnormality but also help distinguish possible causes.

19.1 Signal Faults

• Sensor wire break

• Short circuit

• Stuck signal

• Electrical noise

• Communication error

• Physically unrealistic measured value

19.2 Device Faults

• Motor does not operate

• Cylinder does not complete

• Valve does not respond

• Brake does not engage

• Heater output abnormal

19.3 Mechanism Faults

• Belt breakage

• Coupling disengagement

• Mechanical jam

• Excessive friction

• Increased backlash

19.4 Process Faults

• Product does not arrive

• Duplicate product

• Positioning defect

• Clamping defect

• Inspection failure

19.5 State Inconsistency

• Command and actual operation do not match

• Sensors contradict each other

• Process state and product state do not match

• Internal controller state and actual machine state do not match

20. State Recovery and Revalidation

A cleared fault does not mean that the state should immediately be changed to normal.

The following procedure is required:

1. Remove the cause of the abnormality.

2. Confirm that related signals have returned to normal.

3. Verify the actual position and condition of the device.

4. Perform homing if required.

5. Confirm the product and tool conditions.

6. Resolve any incomplete function.

7. Recheck interlocks and safety conditions.

8. Confirm an intentional operator reset.

9. Transition to the Ready state.

After recovery, the state must be recognized and validated again.

Simply restoring the previous state may be hazardous.

21. Priority of State Decisions

When multiple decisions occur simultaneously, priority must be defined.

A typical priority order is:

1. Safety-Related Decision

2. Equipment Protection Decision

3. Fault Decision

4. Process Quality Decision

5. Functional Transition Decision

6. Performance Optimization Decision

For example, even if the conditions for transitioning to the next function are satisfied, a safety-door opening must cause a safety stop with higher priority.

Without defined priorities, conflicting control commands may occur.

22. Relationship with Control Function Execution

Control Function Execution defines the function and command to be executed.

Control State Recognition and Decision determines:

• Whether the function may begin

• Whether the function is progressing normally

• Whether the function is complete

• Whether the next function may begin

• Whether interruption, retry, or recovery is required

In other words:

Function Execution creates the action, while State Recognition and Decision evaluates the result of that action and determines the next action.

23. Relationship with Feedback Control

State recognition verifies the current value, while feedback control continuously adjusts output to reduce the difference between the target and actual values.

Examples:

• State recognition: Is the temperature within the normal range?

• Feedback control: Adjust heater output to maintain the set temperature.

• State recognition: Has the position reached the target range?

• Feedback control: Adjust motor output to reduce position error.

State recognition provides the judgment, while feedback control performs continuous regulation.

24. Relationship with Interlock Control

The results of state recognition are used as input conditions for interlocks.

Examples:

• If Clamp Complete is not recognized, prohibit machining-axis descent.

• If Product Present is not recognized, prohibit pressing.

• If Axis Stopped is not recognized, prohibit door unlock.

• If Pressure Normal is not recognized, prohibit automatic operation.

An interlock does not recognize the state itself. It uses the recognized state to permit or block operation.

25. Relationship with Safety-Related Control

Safety-related states require higher reliability and independence than general process states.

Examples:

• Safety-door state

• Emergency-stop state

• Safe-speed state

• Stop-confirmed state

• Brake state

• Hazardous-area access state

Safety-state decisions should consider:

• Detection of wire break and contact welding

• Dual channels

• Cross-monitoring

• Safety-controller diagnostics

• Safety-output feedback

• Restart prevention

Safety functions must not be replaced by process-state evaluation in a standard PLC.

26. Interface with Assembly

The reliability of state recognition is strongly affected by the actual installation condition of sensors.

Required Assembly Information

• Sensor mounting reference surface

• Sensing direction

• Sensing distance

• Adjustment range

• Bracket stiffness

• Cable direction

• Protection from contamination and impact

• Relationship between sensor and target

• Reference position for repeated assembly

• Readjustment method after replacement

If assembly error is larger than the state-decision tolerance, signal repeatability cannot be ensured.

Sensor installation tolerance and state-decision tolerance must therefore be reviewed together.

27. Interface with Manufacturing

The same state decision must be reproduced on every manufactured machine.

Items to Manage

• Sensor settings

• Thresholds

• Hysteresis

• Debounce time

• Filter conditions

• Timeout

• Allowable tolerance

• Warning and stop criteria

• Signal logic

• Sensor calibration values

Manufacturing and Commissioning Tests

• Sensor ON/OFF test

• Analog-range test

• Signal wire-break test

• Contradictory-signal test

• Timeout test

• State-transition test

• Mode-specific decision test

• Communication-loss test

• State test after power recovery

• Reset test after sensor replacement

28. Major Design Deliverables

29. Review Questions for State Recognition and Decision

State Definition

• Are the required states clearly defined?

• Are signals and states distinguished?

• Are device states and final functional states distinguished?

• Are Unknown and Conflicting states defined?

Signal Reliability

• Does the sensor signal represent the actual state?

• Are range and rate-of-change checks provided?

• Are noise and chattering handled?

• Can contradictory signals be detected?

• Is the relationship between command and response verified?

Decision Criteria

• Are normal and abnormal ranges defined?

• Are thresholds and hysteresis appropriate?

• Are time conditions included?

• Are multiple signals combined appropriately?

• Is decision priority defined?

Operation Decision

• Are start, continuation, and completion permissions distinguished?

• Are Hold, Restrict, Retry, and Stop conditions clear?

• Are next-state transition conditions defined?

• Is operation blocked when the state is uncertain?

Recovery

• Is there a recovery method when the state cannot be trusted?

• Is the need for homing defined?

• Is the state revalidated after recovery?

• Are conditions for re-entry into automatic operation clear?

Assembly and Manufacturing

• Are sensor-position tolerances defined?

• Can decision parameters be reproduced consistently?

• Can every state be tested?

• Are signal faults and contradictory states tested?

30. Conclusion

Control State Recognition and Decision is the control domain that converts physical machine signals into meaningful states and determines function execution, continuation, completion, interruption, and recovery based on those states.

Its purpose is not to increase the number of sensors.

The important task is to clearly design the following relationships:

• Which states must be known

• Which signals are used to verify those states

• Whether those signals can be trusted

• Whether multiple signals are consistent

• Whether the state should be classified as normal, abnormal, or uncertain

• Which operations should be permitted or blocked in that state

• When the system should transition to the next state

• How the machine should recover and revalidate the state when state information is lost

Good state recognition and decision control does not process only normal signals.

It is designed to prevent incorrect machine decisions even under uncertain conditions such as sensor failure, contradictory signals, loss of position information, communication failure, and power recovery.

This allows Control Function Execution to operate based on the actual machine state and enables Interlock, Fault Recovery, and Safety-Related Control to function reliably using consistent state information.