This forum uses cookies
This forum makes use of cookies to store your login information if you are registered, and your last visit if you are not. Cookies are small text documents stored on your computer; the cookies set by this forum can only be used on this website and pose no security risk. Cookies on this forum also track the specific topics you have read and when you last read them. Please confirm that you accept these cookies being set.

the average value of tagged objects
#1
Hi,

Ok so I have room of lights.  These objects ( group addresses ) which are either switch ( boolean ) or value based ( scale ) they are tagged as "Liivng".  I want to show the average value of these lights.   All of the status objects have been tagged with the tag called "Living Feedback".  An event based script is placed on the "Living Feedback" tag.  I created a two virtual objects.  The first allows me to set the value of tag objects.  The second represents the feedback, the average value of the objects.  The problem is that there is always a varying number of lights and it can that only one light is changed or all of them are changed. The feedback value can jump up and down whilst every light returns its status.   I have tried to overcome this by not recalculating the every time by checking when the average value status object has just been updated.  Which can lead to the value not always being correct. Is there a better way to do this?

Thanks,

Roger
 
Code:
if (event.type=='groupwrite') then
 local groupstatus = grp.find('5/3/0')  -- Status address of living
 local delta = os.time()-groupstatus.updatetime
 if(delta>5) then
   local currentV = round(groupstatus.value,0)
   local sobj = grp.tag('Living Feedback')
   local sTotal = 0
   local counter = 0
   for key, obj in ipairs(sobj) do
     if(sobj[key].data==true)then
     elseif(sobj[key].data==false)then
     else
       sTotal= sTotal + sobj[key].data
       counter=counter+1
     end
   end
   local value= round((sTotal / counter),0)
   if(currentV~=value) then
     grp.update('5/3/0', value, dt.scale)
   end
 end
end
Reply
#2
Any suggestions?
Reply
#3
Hi Roger,

You could do it directly with a sql query avg(), 

Make a selection in table objects where tag contains TAG name with avg on
value.

This way you get back direct the avg of all your objects and send result to knx object 
when ~= old value or delay to keep bus traffic low.

BR,

Erwin

Hi Roger,

You could do it directly with a sql query avg(), 

Make a selection in table objects where tag contains TAG name with avg on
value.

This way you get back direct the avg of all your objects and send result to knx object 
when ~= old value or delay to keep bus traffic low.

BR,

Erwin
Reply
#4
You cannot do a SQL query on object values as only raw value is stored. The best approach would be to monitor the bus and calculate the average once there's no status telegram for several seconds. This example can be used a starting point:

http://forum.logicmachine.net/showthread...65#pid1265
Reply
#5
Hi,

That's my question.  How can we determine when to make the avg calculation?  Given effectively the avg status should only update once all the lights have given their feedback.  Because in a group of the tagged objects, all could change or just one.  The group can consist of 5 objects or it could be 100 objects.  Selecting a fixed amount of time doesn't work all the time.    

Thx,


Roger
Reply
#6
Hi Edgars,


I think we can but i don't know how to convert hex to dec in a sql query

If i use decimal it works perfect:

result = db:getall('SELECT AVG(id) FROM objects WHERE tagcache LIKE "%AVG%"')
log(result[1]['AVG(id)'])

but for hex yow would do some extra like:

result = db:getall('SELECT AVG(CONVERT(INT,datahex)) FROM objects WHERE tagcache LIKE "%AVG%"')
log(result[1]['AVG(datahex)'])

But above is not correct...

BR,

Erwin
Reply
#7
Each group consists of tagged objects so count does not matter, you have to monitor all objects anyhow. The idea is to have a timer for each group, where each status update resets/reloads this timer. Only once all status updates are received you can calculate and send the average. I'll try to prepare an example next week.
Reply
#8
Is there any update to this?
Reply
#9
Sorry for the delay, here's a universal resident script which can be used for Average / AND / OR calculation.

Code:
if not client then
  -- each group has 3 fields:
  -- tag - status object tag
  -- output - status output object
  -- mode - calculate mode (and/or/avg)
  groups = {
    { tag = 'my_group_and', output = '1/1/3', mode = 'and' },
    { tag = 'my_group_or', output = '1/1/4', mode = 'or' },
    { tag = 'my_group_avg', output = '1/1/5', mode = 'avg' },
  }

  -- time to wait between last telegram from any status in group and update
  updatedelay = 0.5

  -- object value cache
  values = {}

  -- object datatype cache
  datatypes = {}

  -- send update only when value changes
  grp.checkupdate = function(alias, value)
    if values[ alias ] ~= value then
      grp.update(alias, value)
    end
  end

  calc = {}

  -- AVERAGE value
  calc['avg'] = function(group)
    local result, count, value = 0, 0

    for _, address in ipairs(group.objects) do
      value = values[ address ]

      -- number must be in [0..100] range
      if type(value) == 'number' then
        result = result + value
        count = count + 1
      -- boolean true is 100%, false is 0%
      elseif type(value) == 'boolean' then
        if toboolean(value) then
          result = result + 100
        end

        count = count + 1
      end
    end

    if count > 0 then
      result = math.floor(result / count + 0.5)
      grp.checkupdate(group.output, result)
    end
  end

  -- AND gate
  calc['and'] = function(group)
    local result = true

    for _, address in ipairs(group.objects) do
      result = result and toboolean(values[ address ])
    end

    grp.checkupdate(group.output, result)
  end

  -- OR gate
  calc['or'] = function(group)
    local result = false

    for _, address in ipairs(group.objects) do
      result = result or toboolean(values[ address ])
    end

    grp.checkupdate(group.output, result)
  end

  -- prepare each group
  for _, group in ipairs(groups) do
    object = grp.find(group.output)

    -- cache output status object value and datatype
    values[ object.address ] = object.data
    datatypes[ object.address ] = object.datatype
    group.output = object.address

    -- group input status object list
    group.objects = {}

    -- find all status objects and cache values and datatypes
    objects = grp.tag(group.tag)
    for _, object in ipairs(objects) do
      values[ object.address ] = object.data
      datatypes[ object.address ] = object.datatype
      table.insert(group.objects, object.address)
    end

    -- force update on first run
    group.timer = 0

    -- calc function reference
    group.fn = calc[ group.mode ]
  end

  -- handle group writes
  function eventhandler(event)
    local dst, datatype

    dst = event.dst
    datatype = datatypes[ event.dst ]
    -- unknown object, stop
    if not datatype then
      return
    end

    values[ dst ] = dpt.decode(event.datahex, datatype)

    -- check if any group needs to be updated
    for _, group in ipairs(groups) do
      for _, address in ipairs(group.objects) do
        if address == dst then
          group.timer = updatedelay
        end
      end
    end
  end

  require('genohm-scada.eibdgm')
  client = eibdgm:new({ timeout = 0.25 })
  client:sethandler('groupwrite', eventhandler)
end

tsec, tusec = os.microtime()
client:step()
delta = os.udifftime(tsec, tusec)

-- check if any group has an active timer
for _, group in ipairs(groups) do
  timer = group.timer

  if timer then
    timer = timer - delta

    -- timer expired, run calc function
    if timer <= 0 then
      group.fn(group)
      timer = nil
    end

    group.timer = timer
  end
end
Reply
#10
Perfect solution adminWink

Do I must use 1 script for 1 central status e.g. Room and another script for another status? Or I can add in this script a few groups of statuses e.g. Room1, Room2, Room3, Floor1 etc...

I tried to add:

Code:
 groups = {
  { tag = 'my_group_and', output = '4/0/0', mode = 'and' },
  { tag = 'my_group_or', output = '4/0/1', mode = 'or' },
  { tag = 'my_group_avg', output = '4/0/2', mode = 'avg' },
 
  { tag = 'new_group_and', output = '4/0/20', mode = 'and' },
  { tag = 'new_group_or', output = '4/0/21', mode = 'or' },
  { tag = 'new_group_avg', output = '4/0/22', mode = 'avg' },
}



but new_group_xxx is not calculated.
Reply
#11
You should do full script reload after adding new group via disable / enable. You should also use only one script for all groups, otherwise there's unnecessary resource usage increase.
Reply
#12
(14.06.2016, 08:49)admin Wrote: You should do full script reload after adding new group via disable / enable. You should also use only one script for all groups, otherwise there's unnecessary resource usage increase.

You're rightSmile It works. Perfect solution.
Reply
#13
This is cool, thank you. Yes it works.
Reply
#14
HI, 
I've tried this script also for obtaining an average value of some temperature values obtained from a temperature sensor that send values at 5-10 second.
The script should work at each 10 min(scheduled script), and calculate the average value for the elapsed time from the last run
Maybe something I make wrong, or I just miss something.

13/2/9 - is the GA where the sensor sends the values
13/2/11 - should be the average value of received value.

This is script:
Code:
if not client then
 -- each group has 3 fields:
 -- tag - status object tag
 -- output - status output object
 -- mode - calculate mode (and/or/avg)
 groups = {
 
   { tag = '13/2/9', output = '13/2/11', mode = 'avg' },
 }

 -- time to wait between last telegram from any status in group and update
 updatedelay = 0.5

 -- object value cache
 values = {}

 -- object datatype cache
 datatypes = {}

 -- send update only when value changes
 grp.checkupdate = function(alias, value)
   if values[ alias ] ~= value then
     grp.update(alias, value)
   end
 end

 calc = {}

 -- AVERAGE value
 calc['avg'] = function(group)
   local result, count, value = 0, 0

   for _, address in ipairs(group.objects) do
     value = values[ address ]

     -- number must be in [0..100] range
     if type(value) == 'number' then
       result = result + value
       count = count + 1
     -- boolean true is 100%, false is 0%
     elseif type(value) == 'boolean' then
       if toboolean(value) then
         result = result + 100
       end

       count = count + 1
     end
   end

   if count > 0 then
     result = math.floor(result / count + 0.5)
     grp.checkupdate(group.output, result)
   end
 end

 -- AND gate
 calc['and'] = function(group)
   local result = true

   for _, address in ipairs(group.objects) do
     result = result and toboolean(values[ address ])
   end

   grp.checkupdate(group.output, result)
 end

 -- OR gate
 calc['or'] = function(group)
   local result = false

   for _, address in ipairs(group.objects) do
     result = result or toboolean(values[ address ])
   end

   grp.checkupdate(group.output, result)
 end

 -- prepare each group
 for _, group in ipairs(groups) do
   object = grp.find(group.output)

   -- cache output status object value and datatype
   values[ object.address ] = object.data
   datatypes[ object.address ] = object.datatype
   group.output = object.address

   -- group input status object list
   group.objects = {}

   -- find all status objects and cache values and datatypes
   objects = grp.tag(group.tag)
   for _, object in ipairs(objects) do
     values[ object.address ] = object.data
     datatypes[ object.address ] = object.datatype
     table.insert(group.objects, object.address)
   end

   -- force update on first run
   group.timer = 0

   -- calc function reference
   group.fn = calc[ group.mode ]
 end

 -- handle group writes
 function eventhandler(event)
   local dst, datatype

   dst = event.dst
   datatype = datatypes[ event.dst ]
   -- unknown object, stop
   if not datatype then
     return
   end

   values[ dst ] = dpt.decode(event.datahex, datatype)

   -- check if any group needs to be updated
   for _, group in ipairs(groups) do
     for _, address in ipairs(group.objects) do
       if address == dst then
         group.timer = updatedelay
       end
     end
   end
 end

 require('genohm-scada.eibdgm')
 client = eibdgm:new({ timeout = 0.25 })
 client:sethandler('groupwrite', eventhandler)
end

tsec, tusec = os.microtime()
client:step()
delta = os.udifftime(tsec, tusec)

-- check if any group has an active timer
for _, group in ipairs(groups) do
 timer = group.timer

 if timer then
   timer = timer - delta

   -- timer expired, run calc function
   if timer <= 0 then
     group.fn(group)
     timer = nil
   end

   group.timer = timer
 end
end
Reply
#15
No, this script cannot be used to calculate averages for a single group address because it does not store any historical info.

Since your sensor sends data periodically, you can map an event script to 13/2/9 which will store values and calculate average once 10 minutes pass.
Code:
-- storage key
key = 'temp_average'
-- time between average value send (in seconds)
maxtime = 10 * 60
-- resulting average value
result = '13/2/11'

time = os.time()

data = storage.get(key)
if not data then
  data = { values = {}, time = time }
end

value = event.getvalue()
table.insert(data.values, value)

delta = time - data.time
if delta >= maxtime or delta < 0 then
  storage.delete(key)

  avg = 0
  count = #data.values
  for _, value in ipairs(data.values) do
    avg = avg + value
  end

  grp.write(result, avg / count)
else
  storage.set(key, data)
end
Reply
#16
Thanks for the script. It works like a charm Wink
Reply
#17
Hello Admin,

I copied the script, added the tags and only used the 'and' function

groups = {
{ tag = 'GROEP_Route_Wasplaats', output = '2/1/70', mode = 'and' },
{ tag = 'GROEP_Route_Mudroom', output = '2/1/71', mode = 'and' }

}

I get an error in line 84 : Resident script:84: attempt to index global 'object' (a nil value)
stack traceback:

Assigned output GA's don't update

I do not have anything in the common functions

I truly hope you have any Ideas?

Regards,
Hendrik
Reply
#18
Have you created the output objects?
Reply
#19
Ha, I thought so but I missed one.

Thanks!
Reply


Forum Jump: