I'm working on a new web service in Java: energydashws
See the enhancement tag in the Issues list for features I'm planning on adding when I have time.
In the winter of 2012, we received an electric bill for over $700. Granted that was a pretty bad winter (and we heat a sunroom that has mostly glass walls with a small electric heater that's running all the time), but it showed me that I honestly had no real grasp of the amount of power we were using, so I started doing some research on power monitors.
There were a few that existed that had relatively advanced software for drawing charts and calculating stats, but they were pretty expensive, so I thought that maybe I could build something to do what I wanted instead. I had been wanting to learn about web app development for a while, so I decided that I would look into creating my own.
The basic building blocks for the hardware that provides data for energydash are a transmitter and receiver pair called an Envi kit. The kit includes two clamps that go around the main power lines that come into the breaker box, a transmitter that the clamps plug into, and a receiver that pairs with the transmitter and displays power usage information.
In addition to displaying information, the receiver also has an RJ-45 serial port on the back. That serial port transmits an xml message every 6 seconds with the data it received from the transmitter. The Current Cost folks have graciously provided an XML description document that aided me in gleaning meaningful data from the messages. I also purchased an RJ-45 to USB serial adapter that thankfully has Linux drivers.
The XML messages look like this:
<msg><src>CC128-v0.15</src><dsb>00013</dsb><time>00:10:07</time><tmprF>68.7</tmprF><sensor>0</sensor><id>00077</id><type>1</type><ch1><watts>00006</watts></ch1><ch2><watts>00269</watts></ch2><ch3><watts>00328</watts></ch3></msg>
In a more readable format:
<msg>
<src>CC128-v0.15</src>
<dsb>00013</dsb>
<time>00:10:07</time>
<tmprF>68.7</tmprF>
<sensor>0</sensor>
<id>00077</id>
<type>1</type>
<ch1>
<watts>00006</watts>
</ch1>
<h2>
<watts>00269</watts>
</ch2>
<ch3>
<watts>00328</watts>
</ch3>
</msg>
- src: The firmware version running on the receiver.
- dsb: Days since birth - If you know when you started, you can backtrack from this value to figure out the actual day of the reading, but envir_collector.py creates a timestamp at the time of reading. That takes care of two problems. The first is that at least my receiver seems to lose time, so as long as the system running envir_collector.py is using NTP, the time and date should be correct.
- time: Timestamp according to the receiver.
- tmprF: The temperature in degrees Fahrenheit from a sensor on the receiver.
- sensor: I believe this value is referring to the mains sensors.
- id: Radio ID from the sensor.
- type: Sensor type, 1 is electricity (not sure what other types there are).
- ch[1-3]: The mains sensors, readings are in watts. Mine happens to have 3 although only 2 of mine are in use.
There are also historical messages sent every few hours that include reading history, but I'm ignoring those since I'm storing the real-time readings. It's on my "to do" list to read those as well and store them in the database.
The envir_collector.py
module is a multi-threaded process with the following threads:
- Collector: Listens on the USB/serial line for transmissions from the receiver. Each line is added to a
Queue
to be processed. - Writer: Pulls work items from the
Queue
, parses the XML and creates anEnvirMsg
object that generates a document to save in theenvir_reading
collection in the database.
This gives us documents in the envir_reading
collection every 6 seconds as long as the link between the transmitter and receiver are working well.
Example:
> db.envir_reading.findOne()
{
"_id" : ObjectId("51b1456172ea814c7fb8bf09"),
"reading_timestamp" : ISODate("2013-06-07T02:28:49.432Z"),
"receiver_days_since_birth" : 19,
"receiver_time" : "22:22:00",
"ch1_watts" : 7,
"ch2_watts" : 323,
"ch3_watts" : 328,
"total_watts" : 658,
"temp_f" : 69.2
}
The energydash_statsd.py
daemon is a python application that uses the documents in the envir_reading
collection to build collections with coarser granularity to build charts with less data points. For exmaple, the first chart below the current readings covers the last 24 hours. For each chart type, the stats are stored in a relevant collection in the database. The stats are updated every 60 seconds and energydash_statsd.py
keeps track of where it left off by saving a document in the bookmarks
collection.
As mentioned before, the database used for this app is MongoDB. It has the following collections:
- bookmarks: Stores bookmarks to mark where a reader or writer left off. Mostly used in
energydash_statsd.py
but also used byenvir_collector.py
to store the last document it added toenvir_reading
:
{ "_id" : "envir_reading", "timestamp" : ISODate("2013-07-06T22:29:11.160Z") }
{ "_id" : "hours", "timestamp" : ISODate("2013-07-06T22:00:00Z") }
{ "_id" : "seconds", "usage" : 895, "timestamp" : ISODate("2013-07-06T22:29:53.884Z"), "tempf" : 74.6 }
- envir_reading: Stores a document for every 6-second reading that comes from the EnviR receiver.
{
"_id" : ObjectId("51b1456172ea814c7fb8bf09"),
"reading_timestamp" : ISODate("2013-06-07T02:28:49.432Z"),
"receiver_days_since_birth" : 19,
"receiver_time" : "22:22:00",
"ch1_watts" : 7,
"ch2_watts" : 323,
"ch3_watts" : 328,
"total_watts" : 658,
"temp_f" : 69.2
}
- hours: Stores hour-granularity averages of the documents in
envir_reading
:
{
"_id" : ISODate("2013-06-07T02:00:00Z"),
"count" : 286,
"average_usage" : 624,
"average_tempf" : 68.8779720279721,
"timestamps" : [
ISODate("2013-06-07T02:28:49.432Z"),
ISODate("2013-06-07T02:28:55.473Z"),
...
]
}
- hours_in_day: Stores documents with averages for each of the 24 hours in all days:
{
"_id" : "23",
"count" : 29,
"average_usage" : 842,
"average_tempf" : 67.8939769930134,
"timestamps" : [
ISODate("2013-06-07T03:00:00Z"),
ISODate("2013-06-08T03:00:00Z"),
...
],
"timezone" : "America/New_York"
}
- hours_per_dow: Stores documents with averages for each hour of each day of the week:
{
"_id" : "Wed",
"hours" : {
"20" : {
"average_usage" : 1270,
"count" : 4,
"average_tempf" : 72.23073396282791,
"timestamps" : [ ]
},
"21" : {
"average_usage" : 1125,
"count" : 4,
"average_tempf" : 71.1739317983943,
"timestamps" : [ ]
},
...
},
"timezone" : "America/New_York"
}
The energydash_app.py
module is a Flask-based web application. It pulls data from MongoDB and displays two main sections in the application:
- Current: The top section shows the most recent reading on the left and a line graph of readings for the last 24 hours on the right.
- Stats/Charts: The bottom, larger section presents tabs that have stats and charts for different time frames.
The charts are all rendered using jquery/flot which uses the HTML5 canvas.