API Reference: Sensors

API Reference: Sensors

The SensorThreshold library provides data containers with state-dependent threshold rules.

Sensor

Data container with state channels and condition-dependent thresholds.

Constructor

s = Sensor('pressure');
s = Sensor('pressure', 'Name', 'Chamber Pressure', 'ID', 101);
s = Sensor('pressure', 'MatFile', 'data.mat', 'KeyName', 'p_chamber');

Properties

Property	Type	Description
Key	string	Unique identifier
Name	string	Human-readable display name
ID	number	Numeric sensor ID
Source	string	Path to original data file
MatFile	string	Path to .mat file
KeyName	string	Field name in .mat file (defaults to Key)
X	1xN double	Time stamps (datenum)
Y	1xN double	Sensor values
StateChannels	cell array	StateChannel objects
ThresholdRules	cell array	ThresholdRule objects
ResolvedThresholds	struct array	Precomputed threshold lines (after resolve)
ResolvedViolations	struct array	Precomputed violation points (after resolve)
ResolvedStateBands	struct	Precomputed state region bands (after resolve)

Methods

Method	Description
s.load()	Load data from external source
s.addStateChannel(sc)	Attach a StateChannel
s.addThresholdRule(condition, value, ...)	Add a condition-dependent threshold
s.resolve()	Precompute thresholds, violations, and bands
s.getThresholdsAt(t)	Evaluate active rules at a specific time

addThresholdRule Parameters

s.addThresholdRule(condition, value, 'Direction', 'upper', 'Label', 'HH');

• condition — struct with state keys/values (e.g., struct('machine', 1))
• value — threshold value (scalar)
• Direction — 'upper' (y > th) or 'lower' (y < th)
• Label — display label
• Color — RGB triplet override

Complete Example

% Create sensor with data
s = Sensor('pressure', 'Name', 'Chamber Pressure');
s.X = linspace(0, 100, 1e6);
s.Y = randn(1, 1e6) * 10 + 50;

% Attach state channel (machine mode)
sc = StateChannel('machine');
sc.X = [0 30 60 80];
sc.Y = [0 1 2 1];     % idle=0, run=1, boost=2
s.addStateChannel(sc);

% Condition-dependent thresholds
s.addThresholdRule(struct('machine', 1), 70, 'Direction', 'upper', 'Label', 'Run HI');
s.addThresholdRule(struct('machine', 2), 55, 'Direction', 'upper', 'Label', 'Boost HI');
s.addThresholdRule(struct(), 80, 'Direction', 'upper', 'Label', 'Absolute HI');

% Resolve and plot
s.resolve();
fp = FastPlot('Theme', 'dark');
fp.addSensor(s);
fp.render();

---

StateChannel

Time-varying discrete state (e.g., machine mode: idle=0, run=1, boost=2).

Constructor

sc = StateChannel('machine');
sc.X = [0 20 40 60];     % time points
sc.Y = [0 1 2 1];        % state values

Properties

Property	Type	Description
Key	string	Unique state channel identifier
X	1xM double	Time points (monotonically increasing)
Y	1xM double	State values

---

ThresholdRule

Condition-value threshold pair.

Constructor

rule = ThresholdRule(struct('machine', 1), 50);
rule = ThresholdRule(struct('machine', 1, 'phase', 3), 30, 'Direction', 'lower', 'Label', 'Low');
rule = ThresholdRule(struct(), 50);  % empty condition = always active

Properties

Property	Type	Description
Condition	struct	State keys/values that activate this rule
Value	number	Threshold value
Direction	string	'upper' or 'lower'
Label	string	Display label
Color	RGB triplet	Color override (empty = use theme)

Methods

matches = rule.matchesState(stateStruct);

---

SensorRegistry

Predefined sensor catalog for quick access.

Static Methods

SensorRegistry.list();                                       % Print all available sensors
s = SensorRegistry.get('pressure');                          % Retrieve by key
sensors = SensorRegistry.getMultiple({'pressure', 'temperature'});  % Multiple at once

---

Resolution Algorithm

The resolve() method uses an efficient segment-based approach:

1. Find state-change timestamps from all channels
2. Evaluate conditions once per segment (typically 5-50 segments)
3. Batch rules sharing the same condition
4. SIMD-accelerated violation detection per segment

Complexity: O(S x R) where S = number of state segments and R = number of rules. For 10M data points with 20 segments and 5 rules, this evaluates 100 condition checks instead of 50M per-point evaluations.