David's Astronomy - Notes 2019-09-07 (S00691)

Session Aims & Highlights

- Observing Result
- Night Summary Plot
- Session Event Log

Operational Issues

- Critical Issues (0), Major Issues (1), Minor Issues (6), Continuous Improvement (16)

Investigation - Slow Schedule Building in AstroPlan

2019-09-09

Pegasus UPB - Software Update (1.4.17)

New Wind Sensor (WMR800X)

Images from 2019-09-07 >>

Timings on Observatory Computer

235 targets, N=200, G=40

  Genetic Modelling Loop
    Create Initial Population     00:04:18         1    258.45         -         -
    Run Generation                00:00:49        39      1.27      0.20      3.03

  Total Time
    Create Schedule (Total)       00:05:35         1    335.67         -         -

  Specific Activites
    CreatePlan                    00:04:18       200      1.29      0.98      3.45
    Choose Target                 00:00:16  18764292    < 0.01      0.00      0.02
    Calc. Fitness                 00:03:58  12393047    < 0.01      0.00      0.08

Nearby 4 minutes is taken up in calculating fitness (Cal Fitness) during initial population build.

Series of tests on the Development Computer.
This has a faster CPU and the times taken on different activites are accordingly lower.
'Choose Target' called over 17.3 million times, Calc Fitness called 11.5 million times.

Test 1: 236 targets, N=200, G=40

  Genetic Modelling Loop
    Create Initial Population     00:02:55         1    175.03         -         -
    Run Generation                00:01:22        39      2.12      1.08      2.61

  Total Time
    Create Schedule (Total)       00:04:28         1    268.19         -         -

  Specific Activites
    CreatePlan                    00:02:52       200      0.86      0.83      1.04
    Choose Target                 00:00:10  17296065    < 0.01      0.00      0.00
    Calc. Fitness                 00:02:20  11527680    < 0.01      0.00      0.02

Test 2: 236 targets, N=200, G=40, with WriteLogFile routine deactivated

  Genetic Modelling Loop
    Create Initial Population     00:01:04         1     64.77         -         -
    Run Generation                00:01:20        39      2.06      1.69      2.29

  Total Time
    Create Schedule (Total)       00:02:35         1    155.82         -         -

  Specific Activites
    CreatePlan                    00:01:02       200      0.31      0.30      0.35
    Choose Target                 00:00:10  17333562    < 0.01      0.00      0.00
    Calc. Fitness                 00:00:32  11543797    < 0.01      0.00      0.01

The second test saw a 2 minutes saving in total time required to create the Schedule.
Most of this saving saving was in Calc.Fitness calls during the creation of the initial populations.

2019-09-17, V.1.10.2 Check
Test 3: 222 targets, N=200, G=40, with WriteLogFile routine deactivated during GA fitness
108 targets disabled

  Genetic Modelling Loop
    Create Initial Population     00:01:04         1     64.74         -         -
    Run Generation                00:01:12        39      1.85      0.61      2.09

  Total Time
    Create Schedule (Total)       00:02:25         1    145.64         -         -

  Specific Activites
    CreatePlan                    00:01:02       200      0.31      0.30      0.35
    Choose Target                 00:00:12  18968330    < 0.01      0.00      0.00
    Calc. Fitness                 00:00:31  12002536    < 0.01      0.00      0.01

2019-09-17, V.1.10.2 Check
Test 4: 222 targets, N=200, G=40, with WriteLogFile routine deactivated during GA fitness. Modified GetPlan (optimised)
197 targets disabled

  Genetic Modelling Loop
    Create Initial Population     00:00:07         1      7.94         -         -
    Run Generation                00:00:44        39      1.14      0.34      1.53

  Total Time
    Create Schedule (Total)       00:01:01         1     61.39         -         -

  Specific Activites
    CreatePlan                    00:00:00       200    < 0.01      0.00      0.01
    Choose Target                 00:00:00    160959    < 0.01      0.00      0.00
    Calc. Fitness                 00:00:00     36443    < 0.01      0.00      0.02

2019-09-09

Pegasus UPB - Software Update (1.4.17)

UPB Software on Observatory and Development Computer was updated to the latest version (1.4.17, Sep/19) today, 2019-09-09.

Again it had been hoped that the 1.4.17 update would include a fix to the problem that UPB Env Humidity values fall to a constant 90% when the real humidity exceeds 97%, but it doesn't.

A fix to allow the UPB Software & ASCOM UPB Switch Drivers to work together (promised for End August) is also still awaited.

[ Pegasus Astro Support Page: https://pegasusastro.com/support/ ]

[ The long awaited fix to allow proper workable access and operation of UPB Switches via ASCOM was eventually delievered in Oct 2019
(See Pegasus UPB - ASCOM Switch Driver Update (2.0), 2019-10-01 ) ]

New Wind Sensor (WMR800X)

A replacement Wind Sensor for the WMR80 Oregon Scientific Weather Station was installed today, 2019-09-09.

The old unit had been having increased problems with sensing wind direction. Originally it could detect 8 directions, then 6-7 directions and recently it seems that only 1 or 2 directions are being reported, mostly N (000°).

The new anemonmeter unit (WGR800X) was purchased from Weather Spares (https://www.weatherspares.co.uk/) . The new unit detects 16 directions and after a moment of hestitation about whether it would work ok with my older base station, eventually I got signals through and confirmed that its reading ok.

New Wind Sensor (assembled)		New Wind Sensor (in Position)


Direction Data from new Wind Sensor Unit 36 Hour Plot shows the raw data (detecting 16 directions, with 22.5 deg resolution)


5 day plot shows the data as rolling average (Compare new data with data from old sensor, which has recently been ouputing only N and W)


Later 5 day plot showing subtle wind direction changes & trends