pf-dre-core 1.0.11

Core library used by DRE and MMS Tools & Subsystems

Meter Management System Client

The python implementation of the pf-dre-database repo provides a client for all Data interactions required with the Meter Management System:

• Relational tables (Read Only)
• Timescale DB (Read/Write - No insertion or deletion)
  • JSON Schema
  • Narrow Data Format Schema

This python implementation is to be built and deployed to PyPI for use across all python subsystems of the Demand Response Engine.

Real-Time Input Data Format

When issuing calls to the MMS which require a time series based DataFrame to be passed the format of the schema should be followed with the following general rules.

• Timestamps are to be generated in string format following the ISO 8601 standard and to follow simple conventions should be kept in UTC format.
• Any columns within the data structure which are a JSON datatype are to be created in serialized string format, not as a pure python dictionary.

import json
# Correct Format
correct_jsonb_value = json.dumps({'A': 'Dictionary', 'B': 'to', 'C': 'Send'})
# Incorrect Format
incorrect_jsonb_value = {'A': 'Dictionary', 'B': 'to', 'C': 'Send'}

Example Data Frame for a narrow column format schema

measurement_date	device_id	device_metric_type_id	value
2020-01-01T12:00:00.000	1	P	1.0
2020-01-01T12:01:00.000	1	P	2.0
2020-01-01T12:00:00.000	1	Q	-1.0
2020-01-01T12:01:00.000	1	Q	-2.0
2020-01-01T12:00:00.000	2	P	10.0
2020-01-01T12:01:00.000	2	P	20.0
2020-01-01T12:00:00.000	2	Q	-10.0
2020-01-01T12:01:00.000	2	Q	-20.0
object (str)	int64	object (str)	float64

Example Data Frame for a JSON schema

measurement_date	device_id	metrics
2020-01-01T12:00:00.000	1	{"P": 1.0, "Q": -1.0, "S": 'NaN'}
2020-01-01T12:00:00.000	2	{"P": 2.0, "Q": -2.0}
2020-01-01T12:01:00.000	1	{"P": 10.0, "Q": -10.0}
2020-01-01T12:01:00.000	2	{"P": 20.0, "Q": -20.0}
object (str)	int64	object (str)

Real-Time Standardized Output DataFrame Format

When issuing calls to the MMS which return a time series DataFrame, the client, regardless of schema will be constructed to return in a standardized format. This makes the reading and manipulation of data consistent.

device_id	device_metric_type_id	measurement_date	value
1	P	2020-01-01T12:00:00.000	1001.0
		2020-01-01T12:01:00.000	1012.0
	Q	2020-01-01T12:00:00.000	12.132
		2020-01-01T12:01:00.000	-2.132
2	P	2020-01-01T12:00:00.000	2001.0
		2020-01-01T12:01:00.000	2012.0
	Q	2020-01-01T12:00:00.000	22.132
		2020-01-01T12:01:00.000	-3.132
int64	object (str)	object (str)	float64

The client also has the option of returing the data frame results in a raw, un-standardised format. In this case, the dataframe will be returned in the format of the underlying database schema without any alteration.

Forecast Input Data Format

When issuing calls to the MMS which require a forecast time series based DataFrame to be passed, the format of the schema should be followed with the following general rules.

• Timestamps are to be generated in string format following the ISO 8601 standard and to follow simple conventions should be kept in UTC format.
• Any columns within the data structure which are a JSON datatype are to be created in serialized string format, not as a pure python dictionary.

Example Data Frame for a narrow column format schema

received_date	device_id	device_metric_type_id	measurement_date	value
2020-01-01T12:00:00.000	1	P	2020-01-01T12:00:00.000	1.0
2020-01-01T12:00:00.000	1	P	2020-01-01T12:01:00.000	2.0
2020-01-01T12:00:00.000	1	P	2020-01-01T12:02:00.000	3.0
2020-01-01T12:00:00.000	1	Q	2020-01-01T12:00:00.000	-1.0
2020-01-01T12:00:00.000	1	Q	2020-01-01T12:01:00.000	-2.0
2020-01-01T12:00:00.000	1	Q	2020-01-01T12:02:00.000	-3.0
2020-01-01T12:01:00.000	1	P	2020-01-01T12:01:00.000	2.0
2020-01-01T12:01:00.000	1	P	2020-01-01T12:02:00.000	3.0
2020-01-01T12:01:00.000	1	P	2020-01-01T12:03:00.000	4.0
2020-01-01T12:01:00.000	1	Q	2020-01-01T12:01:00.000	-2.0
2020-01-01T12:01:00.000	1	Q	2020-01-01T12:02:00.000	-3.0
2020-01-01T12:01:00.000	1	Q	2020-01-01T12:03:00.000	-4.0
object (str)	int64	object (str)	object	float64

Example Forecast Data Frame for a JSON schema

received_date	device_id	metrics
2020-01-01T12:00:00.000	1	{ "P": {"2020-01-01T12:00:00+00:00": 1.0, "2020-01-01T12:01:00+00:00": 2.0, "2020-01-01T12:02:00+00:00": 3.0}, "Q": {"2020-01-01T12:00:00+00:00": -1.0, "2020-01-01T12:01:00+00:00": -2.0, "2020-01-01T12:02:00+00:00": -3.0, "2020-01-01T12:03:00+00:00": 'NaN'}}
2020-01-01T12:01:00.000	1	{ "P": {"2020-01-01T12:01:00+00:00": 2.0, "2020-01-01T12:02:00+00:00": 3.0, "2020-01-01T12:03:00+00:00": 4.0}, "Q": {"2020-01-01T12:01:00+00:00": -2.0, "2020-01-01T12:02:00+00:00": -3.0, "2020-01-01T12:03:00+00:00": -4.0}}
object (str)	int64	object

Forecast Standardized Output DataFrame Format

When issuing calls to the MMS which return a time series DataFrame, the client, regardless of schema will be constructed to return in a standardized format. This makes the reading and manipulation of data consistent.

received_date	device_id	device_metric_type_id	measurement_date	value
2020-01-01T12:00:00.000	1	P	2020-01-01T12:00:00.000	1.0
			2020-01-01T12:01:00.000	2.0
			2020-01-01T12:02:00.000	3.0
		Q	2020-01-01T12:00:00.000	-1.0
			2020-01-01T12:01:00.000	-2.0
			2020-01-01T12:02:00.000	-3.0
2020-01-01T12:01:00.000	1	P	2020-01-01T12:01:00.000	2.0
			2020-01-01T12:02:00.000	3.0
			2020-01-01T12:03:00.000	4.0
		Q	2020-01-01T12:01:00.000	-2.0
			2020-01-01T12:02:00.000	-3.0
			2020-01-01T12:03:00.000	-4.0
object (str)	int64	object (str)	object (str)	float64

The client also has the option of returning the data frame results in a raw, un-standardised format by initializing the client with the argument: standardized=False In this case, the data frame will be returned in the format of the underlying database schema without any alteration.

Accessing Standardized DataFrame Contents

Real-time DataFrames python # Accessing Device Metric Timeseries time_series = df.loc[(1, 'P'), 'value'] time_series_times = df.loc[(1, 'P'), 'value'].index.values.tolist()) time_series_values = df.loc[(1, 'P'), 'value'].values.tolist()) # Accessing Most Recent Device Metric # (Note the standardized format is sorted in ascending order) latest_P = df.loc[(1, 'P'), 'value'].iloc[0] Forecast DataFrames

Prerequisites

• Python 3.7.0+

Setup

The following environment variables are required in order to make use of the client.

• PGDATABASE: The name of the MMS Database instance.
• PGUSER: MMS Database user.
• PGPASSWORD: MMS Database password.
• PGHOST: MMS Database host.
• PGPORT: MMS Database port (read/write permissions required).

These can be set through updating the file at ./static/local-ssm-values.json Note that there