adap 0.0.1

This is python package used detect different types of anomalies in compressors' sensors time series data.

Anomaly Detection and Prediction (ADAP) Package

This is a simple example package. You can use GitHub-flavored Markdown to write your content.

Anomaly type	Type code	Description	Example in the literature and/or alternative terminology	Importance ranking (with respect to time series visualization in this case study)	Potential end-user and application
Impossible values	A	Values 6666 are highly impossible in compressor sensor data and consider as out of range value of the sensors.	-	Important, but it can be deleted easily using simple rule	-
Constant values within the acceptable region for long time period	B	Constant values within the acceptable region for long time period (more than one day)	-	Important can be detected using rate of change	-
Large sudden spike	C	The spikes are larger than the surrounding data points but less than 6666. Example one or two values much higher than previous or next values	-	At any point in the time series, used 3 times standard deviation as threshold	-
Drift	D	Gradual change in values in positive to zero	-	High priority (Emergency shutdown, ESD)	-
Sudden zero	E	Value suddenly shifts in values in positive to zero	-	High priority (Emergency shutdown, ESD)	-

Table 1: Load current related anomaly types

Anomaly type	Type code	Description	Example in the literature and/or alternative terminology	Importance ranking (with respect to time series visualization in this case study)	Potential end-user and application
Impossible values	A	Values 6666 are highly impossible in compressor sensor data and consider as out of range value of the sensors.	-	Important, but it can be deleted easily using simple rule	-
Constant values within the acceptable region for long time period	B	Constant values within the acceptable region for long time period (more than one day)	-	Important can be detected using rate of change	-
Large sudden spike	C	The spikes are larger than the surrounding data points but less than 6666. Example one or two values much higher than previous or next values	-	At any point in the time series, used 3 times standard deviation as threshold	-
Drift	D	Gradual change in values in positive to zero	-	High priority (Emergency shutdown, ESD)	-
Sudden zero	E	Value suddenly shifts in values in positive to zero	-	High priority (Emergency shutdown, ESD)	-
Step up (sudden shift)	F	Increase the mean while keeping the variance constant	-	High priority (Emergency shutdown, ESD)	-
Step down (sudden shift)	G	Decrease the mean while keeping the variance constant	-	Normal, indicate motor flips from OFF to ON	-

Table 2: Suction pressure related anomaly types

Anomaly type	Type code	Description	Example in the literature and/or alternative terminology	Importance ranking (with respect to time series visualization in this case study)	Potential end-user and application
Impossible values	A	Values 6666 are highly impossible in compressor sensor data and consider as out of range value of the sensors.	-	Important, but it can be deleted easily using simple rule	-
Constant values within the acceptable region for long time period	B	Constant values within the acceptable region for long time period (more than one day)	-	Important can be detected using rate of change	-
Large sudden spike	C	The spikes are larger than the surrounding data points but less than 6666. Example one or two values much higher than previous or next values	-	At any point in the time series, used 3 times standard deviation as threshold	-
Drift	D	Gradual change in values in positive to zero	-	High priority (Emergency shutdown, ESD)	-
Sudden zero	E	Value suddenly shifts in values in positive to zero	-	High priority (Emergency shutdown, ESD)	-
Step up (sudden shift)	F	Increase the mean while keeping the variance constant	-	Normal, indicate motor flips from OFF to ON	-
Step down (sudden shift)	G	Decrease the mean while keeping the variance constant	-	High priority (Emergency shutdown, ESD)	-

Table 2: Discharge pressure related anomaly types