Re: Stupid home non-automation product

[Marc_F_Hult <MFHult@xxxxxxxxxxxxxxxxxxxxx>]

On Wed, 07 Mar 2007 18:05:39 GMT, "Jeff Volp" <JeffVolp@xxxxxxx> wrote in
message  <T9DHh.117033$5j1.38997@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx>:

>"Marc_F_Hult" <MFHult@xxxxxxxxxxxxxxxxxxxxx> wrote in message
>news:gfptu2hlpvs1ebti25dafoepao715rvndt@xxxxxxxxxx

>>
>> Sure. This is a relatively simple rule of the form:
>>
>> WHILE [Time period}
>> IF [computed value] > [constant] THEN [Boolean result]
>>
>> One has been able to point-and-click through this sort of computed
>> rule-making using windows-based CyberHouse for a decade.
>>
>> But there is very little "historical data" and no "statistics" involved.
>>
>> And the rule might not be smart enough to avoid violating the "do no
>> harm' maxim.  For example, if the positive delta T is because there is
>> a fire in the garage, will the system know not to literally fan the
>> flames?
>
>A little more complex than that because it does use the max temperature
>from  the prior day.  Just about every summer day here is a carbon copy of
>the day  before, so that works well to anticipate high for the day.  No
>sense burning  electricity if it isn't going to be at least in the 90's.
>
>The fan will switch on one hour before sunrise only if the prior day was
>hot, and the outside temperature is at least 8 degrees cooler than the
>garage at that point. The fan shuts off if the delta T drops below 4
>degrees  or the garage starts to warm up.  Again, no sense burning
>electricity if it  isn't helping.  While not intended to deal with a fire,
>I think that  condition is pretty much covered.  I suppose adding a
>high-temperature  lockout for redundancy wouldn't hurt.
>
>Jeff

This is a neat application that works well because you provide the
intelligence, analysis and the solution algorithm and is a good example of
"intelligent computing in home automation"

But it also illustrates my question of: " Artificial Intelligence  (AI) :
Where art thou 50 years after you were given a name (1956) and [the gift
of] language (LISP in 1958)?" because the approach you use could have could
have been implemented in 1957 with mechanical devices such as timers,
relays and thermostats (hence my use of the word "stagnation").

Yours is a 'hand-coded' solution to the problem which uses analog input,
conditional statements and computed binary thresholds (TempYesterday >
TempHot AND TempOutside => TempGarage + 8 )  combined with measured
feedback (TempOutside - TempGarage <4).

Your "word-problem solution" is beyond the capability of the vast majority
of homeowners.  (Having taught environmental science at one of the US's
top-ranked colleges, my experience is that solution of this sort of math
"word problem' is also beyond the capabilities of even exceptionally
motivated, multi-talented and intelligent individuals.  It jist is; the
'why' is complicated.)

And once the word problem is adequately translated into a rule set, your
solution requires programming an obscure machine (Ocelot) in an obscure
computer language (C-Max) -- something that even most participants of this
newsgroup probably couldn't do. (Hence my frustration with the
oft-repeated, exceptionally bad advice to buy an Ocelot that was proffered
by some regulars in this newsgroup to folks that can't program.)

So as a practical matter, your approach is still out of reach for most
homeowners and even most/many HA enthusiasts. (AI: Where art thou?)

An 'intelligent' home automation system c/would use:

1)  A predictive, mechanistic model based on
	a) system properties (eg thermal conductivity of walls, heat
     	   capacity of garage contents etc)
	b) physical processes (eg diffusion and convection equations
	    for heat transport) and
	c) boundaries (house temp = 20C)

2) An empirical approach  akin to an auto-tuning  PID
   (Proportional-Integral-Derivative) algorithm used for industrial
    temperature control.

And(or)

3) A stochastic (statistical) approach that could use local and regional
   data (eg variance in soil moisture; probability of rain in the next
   week).

The homeowner would input desired outcome ("Least Cost", "Most
Comfortable", "Most Safe") and a measure of certainty / error tolerance
( "Critical", "Important", "If Practical" etc).

The HA system would determine and make appropriate local measurements,
acquire values for constants, chose solution method and compute and
implement the solution.


... Marc
Marc_F_Hult
www.NeuralHome.org