mirror of
https://github.com/Akkudoktor-EOS/EOS.git
synced 2025-10-29 13:56:21 +00:00
b397b5d43e6d59ccaead983493d5ac9b02b8c8b2
8 Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
Bobby Noelte
|
b397b5d43e |
fix: automatic optimization (#596)
This fix implements the long term goal to have the EOS server run optimization (or
energy management) on regular intervals automatically. Thus clients can request
the current energy management plan at any time and it is updated on regular
intervals without interaction by the client.
This fix started out to "only" make automatic optimization (or energy management)
runs working. It turned out there are several endpoints that in some way
update predictions or run the optimization. To lock against such concurrent attempts
the code had to be refactored to allow control of execution. During refactoring it
became clear that some classes and files are named without a proper reference
to their usage. Thus not only refactoring but also renaming became necessary.
The names are still not the best, but I hope they are more intuitive.
The fix includes several bug fixes that are not directly related to the automatic optimization
but are necessary to keep EOS running properly to do the automatic optimization and
to test and document the changes.
This is a breaking change as the configuration structure changed once again and
the server API was also enhanced and streamlined. The server API that is used by
Andreas and Jörg in their videos has not changed.
* fix: automatic optimization
Allow optimization to automatically run on configured intervals gathering all
optimization parameters from configuration and predictions. The automatic run
can be configured to only run prediction updates skipping the optimization.
Extend documentaion to also cover automatic optimization. Lock automatic runs
against runs initiated by the /optimize or other endpoints. Provide new
endpoints to retrieve the energy management plan and the genetic solution
of the latest automatic optimization run. Offload energy management to thread
pool executor to keep the app more responsive during the CPU heavy optimization
run.
* fix: EOS servers recognize environment variables on startup
Force initialisation of EOS configuration on server startup to assure
all sources of EOS configuration are properly set up and read. Adapt
server tests and configuration tests to also test for environment
variable configuration.
* fix: Remove 0.0.0.0 to localhost translation under Windows
EOS imposed a 0.0.0.0 to localhost translation under Windows for
convenience. This caused some trouble in user configurations. Now, as the
default IP address configuration is 127.0.0.1, the user is responsible
for to set up the correct Windows compliant IP address.
* fix: allow names for hosts additional to IP addresses
* fix: access pydantic model fields by class
Access by instance is deprecated.
* fix: down sampling key_to_array
* fix: make cache clear endpoint clear all cache files
Make /v1/admin/cache/clear clear all cache files. Before it only cleared
expired cache files by default. Add new endpoint /v1/admin/clear-expired
to only clear expired cache files.
* fix: timezonefinder returns Europe/Paris instead of Europe/Berlin
timezonefinder 8.10 got more inaccurate for timezones in europe as there is
a common timezone. Use new package tzfpy instead which is still returning
Europe/Berlin if you are in Germany. tzfpy also claims to be faster than
timezonefinder.
* fix: provider settings configuration
Provider configuration used to be a union holding the settings for several
providers. Pydantic union handling does not always find the correct type
for a provider setting. This led to exceptions in specific configurations.
Now provider settings are explicit comfiguration items for each possible
provider. This is a breaking change as the configuration structure was
changed.
* fix: ClearOutside weather prediction irradiance calculation
Pvlib needs a pandas time index. Convert time index.
* fix: test config file priority
Do not use config_eos fixture as this fixture already creates a config file.
* fix: optimization sample request documentation
Provide all data in documentation of optimization sample request.
* fix: gitlint blocking pip dependency resolution
Replace gitlint by commitizen. Gitlint is not actively maintained anymore.
Gitlint dependencies blocked pip from dependency resolution.
* fix: sync pre-commit config to actual dependency requirements
.pre-commit-config.yaml was out of sync, also requirements-dev.txt.
* fix: missing babel in requirements.txt
Add babel to requirements.txt
* feat: setup default device configuration for automatic optimization
In case the parameters for automatic optimization are not fully defined a
default configuration is setup to allow the automatic energy management
run. The default configuration may help the user to correctly define
the device configuration.
* feat: allow configuration of genetic algorithm parameters
The genetic algorithm parameters for number of individuals, number of
generations, the seed and penalty function parameters are now avaliable
as configuration options.
* feat: allow configuration of home appliance time windows
The time windows a home appliance is allowed to run are now configurable
by the configuration (for /v1 API) and also by the home appliance parameters
(for the classic /optimize API). If there is no such configuration the
time window defaults to optimization hours, which was the standard before
the change. Documentation on how to configure time windows is added.
* feat: standardize mesaurement keys for battery/ ev SoC measurements
The standardized measurement keys to report battery SoC to the device
simulations can now be retrieved from the device configuration as a
read-only config option.
* feat: feed in tariff prediction
Add feed in tarif predictions needed for automatic optimization. The feed in
tariff can be retrieved as fixed feed in tarif or can be imported. Also add
tests for the different feed in tariff providers. Extend documentation to
cover the feed in tariff providers.
* feat: add energy management plan based on S2 standard instructions
EOS can generate an energy management plan as a list of simple instructions.
May be retrieved by the /v1/energy-management/plan endpoint. The instructions
loosely follow the S2 energy management standard.
* feat: make measurement keys configurable by EOS configuration.
The fixed measurement keys are replaced by configurable measurement keys.
* feat: make pendulum DateTime, Date, Duration types usable for pydantic models
Use pydantic_extra_types.pendulum_dt to get pydantic pendulum types. Types are
added to the datetimeutil utility. Remove custom made pendulum adaptations
from EOS pydantic module. Make EOS modules use the pydantic pendulum types
managed by the datetimeutil module instead of the core pendulum types.
* feat: Add Time, TimeWindow, TimeWindowSequence and to_time to datetimeutil.
The time windows are are added to support home appliance time window
configuration. All time classes are also pydantic models. Time is the base
class for time definition derived from pendulum.Time.
* feat: Extend DataRecord by configurable field like data.
Configurable field like data was added to support the configuration of
measurement records.
* feat: Add additional information to health information
Version information is added to the health endpoints of eos and eosDash.
The start time of the last optimization and the latest run time of the energy
management is added to the EOS health information.
* feat: add pydantic merge model tests
* feat: add plan tab to EOSdash
The plan tab displays the current energy management instructions.
* feat: add predictions tab to EOSdash
The predictions tab displays the current predictions.
* feat: add cache management to EOSdash admin tab
The admin tab is extended by a section for cache management. It allows to
clear the cache.
* feat: add about tab to EOSdash
The about tab resembles the former hello tab and provides extra information.
* feat: Adapt changelog and prepare for release management
Release management using commitizen is added. The changelog file is adapted and
teh changelog and a description for release management is added in the
documentation.
* feat(doc): Improve install and devlopment documentation
Provide a more concise installation description in Readme.md and add extra
installation page and development page to documentation.
* chore: Use memory cache for interpolation instead of dict in inverter
Decorate calculate_self_consumption() with @cachemethod_until_update to cache
results in memory during an energy management/ optimization run. Replacement
of dict type caching in inverter is now possible because all optimization
runs are properly locked and the memory cache CacheUntilUpdateStore is properly
cleared at the start of any energy management/ optimization operation.
* chore: refactor genetic
Refactor the genetic algorithm modules for enhanced module structure and better
readability. Removed unnecessary and overcomplex devices singleton. Also
split devices configuration from genetic algorithm parameters to allow further
development independently from genetic algorithm parameter format. Move
charge rates configuration for electric vehicles from optimization to devices
configuration to allow to have different charge rates for different cars in
the future.
* chore: Rename memory cache to CacheEnergyManagementStore
The name better resembles the task of the cache to chache function and method
results for an energy management run. Also the decorator functions are renamed
accordingly: cachemethod_energy_management, cache_energy_management
* chore: use class properties for config/ems/prediction mixin classes
* chore: skip debug logs from mathplotlib
Mathplotlib is very noisy in debug mode.
* chore: automatically sync bokeh js to bokeh python package
bokeh was updated to 3.8.0, make JS CDN automatically follow the package version.
* chore: rename hello.py to about.py
Make hello.py the adapted EOSdash about page.
* chore: remove demo page from EOSdash
As no the plan and prediction pages are working without configuration, the demo
page is no longer necessary
* chore: split test_server.py for system test
Split test_server.py to create explicit test_system.py for system tests.
* chore: move doc utils to generate_config_md.py
The doc utils are only used in scripts/generate_config_md.py. Move it there to
attribute for strong cohesion.
* chore: improve pydantic merge model documentation
* chore: remove pendulum warning from readme
* chore: remove GitHub discussions from contributing documentation
Github discussions is to be replaced by Akkudoktor.net.
* chore(release): bump version to 0.1.0+dev for development
* build(deps): bump fastapi[standard] from 0.115.14 to 0.117.1
bump fastapi and make coverage version (for pytest-cov) explicit to avoid pip break.
* build(deps): bump uvicorn from 0.36.0 to 0.37.0
BREAKING CHANGE: EOS configuration changed. V1 API changed.
- The available_charge_rates_percent configuration is removed from optimization.
Use the new charge_rate configuration for the electric vehicle
- Optimization configuration parameter hours renamed to horizon_hours
- Device configuration now has to provide the number of devices and device
properties per device.
- Specific prediction provider configuration to be provided by explicit
configuration item (no union for all providers).
- Measurement keys to be provided as a list.
- New feed in tariff providers have to be configured.
- /v1/measurement/loadxxx endpoints are removed. Use generic mesaurement endpoints.
- /v1/admin/cache/clear now clears all cache files. Use
/v1/admin/cache/clear-expired to only clear all expired cache files.
Signed-off-by: Bobby Noelte <b0661n0e17e@gmail.com>
|
||
|
Bobby Noelte
|
80bfe4d0f0 |
Improve caching. (#431)
* Move the caching module to core. Add an in memory cache that for caching function and method results during an energy management run (optimization run). Two decorators are provided for methods and functions. * Improve the file cache store by load and save functions. Make EOS load the cache file store on startup and save it on shutdown. Add a cyclic task that cleans the cache file store from outdated cache files. * Improve startup of EOSdash by EOS Make EOS starting EOSdash adhere to path configuration given in EOS. The whole environment from EOS is now passed to EOSdash. Should also prevent test errors due to unwanted/ wrong config file creation. Both servers now provide a health endpoint that can be used to detect whether the server is running. This is also used for testing now. * Improve startup of EOS EOS now has got an energy management task that runs shortly after startup. It tries to execute energy management runs with predictions newly fetched or initialized from cached data on first run. * Improve shutdown of EOS EOS has now a shutdown task that shuts EOS down gracefully with some time delay to allow REST API requests for shutdwon or restart to be fully serviced. * Improve EMS Add energy management task for repeated energy management controlled by startup delay and interval configuration parameters. Translate EnergieManagementSystem to english EnergyManagement. * Add administration endpoints - endpoints to control caching from REST API. - endpoints to control server restart (will not work on Windows) and shutdown from REST API * Improve doc generation Use "\n" linenend convention also on Windows when generating doc files. Replace Windows specific 127.0.0.1 address by standard 0.0.0.0. * Improve test support (to be able to test caching) - Add system test option to pytest for running tests with "real" resources - Add new test fixture to start server for test class and test function - Make kill signal adapt to Windows/ Linux - Use consistently "\n" for lineends when writing text files in doc test - Fix test_logging under Windows - Fix conftest config_default_dirs test fixture under Windows From @Lasall * Improve Windows support - Use 127.0.0.1 as default config host (model defaults) and addionally redirect 0.0.0.0 to localhost on Windows (because default config file still has 0.0.0.0). - Update install/startup instructions as package installation is required atm. Signed-off-by: Bobby Noelte <b0661n0e17e@gmail.com> |
||
Dominique Lasserre
|
c1dd31528b | Config: Move lat/long/timezone from prediction to general | ||
|
Dominique Lasserre
|
af5e4a753a | PVForecast: planes as nested config (list) | ||
|
Dominique Lasserre
|
3257dac92b | Rename settings variables (remove prefixes) | ||
|
Dominique Lasserre
|
be26457563 |
Nested config, devices registry
* All config now nested.
- Use default config from model field default values. If providers
should be enabled by default, non-empty default config file could
be provided again.
- Environment variable support with EOS_ prefix and __ between levels,
e.g. EOS_SERVER__EOS_SERVER_PORT=8503 where all values are case
insensitive.
For more information see:
https://docs.pydantic.dev/latest/concepts/pydantic_settings/#parsing-environment-variable-values
- Use devices as registry for configured devices. DeviceBase as base
class with for now just initializion support (in the future expand
to operations during optimization).
- Strip down ConfigEOS to the only configuration instance. Reload
from file or reset to defaults is possible.
* Fix multi-initialization of derived SingletonMixin classes.
|
||
|
Bobby Noelte
|
830af85fca |
Fix2 config and predictions revamp. (#281)
measurement:
- Add new measurement class to hold real world measurements.
- Handles load meter readings, grid import and export meter readings.
- Aggregates load meter readings aka. measurements to total load.
- Can import measurements from files, pandas datetime series,
pandas datetime dataframes, simple daetime arrays and
programmatically.
- Maybe expanded to other measurement values.
- Should be used for load prediction adaptions by real world
measurements.
core/coreabc:
- Add mixin class to access measurements
core/pydantic:
- Add pydantic models for pandas datetime series and dataframes.
- Add pydantic models for simple datetime array
core/dataabc:
- Provide DataImport mixin class for generic import handling.
Imports from JSON string and files. Imports from pandas datetime dataframes
and simple datetime arrays. Signature of import method changed to
allow import datetimes to be given programmatically and by data content.
- Use pydantic models for datetime series, dataframes, arrays
- Validate generic imports by pydantic models
- Provide new attributes min_datetime and max_datetime for DataSequence.
- Add parameter dropna to drop NAN/ None values when creating lists, pandas series
or numpy array from DataSequence.
config/config:
- Add common settings for the measurement module.
predictions/elecpriceakkudoktor:
- Use mean values of last 7 days to fill prediction values not provided by
akkudoktor.net (only provides 24 values).
prediction/loadabc:
- Extend the generic prediction keys by 'load_total_adjusted' for load predictions
that adjust the predicted total load by measured load values.
prediction/loadakkudoktor:
- Extend the Akkudoktor load prediction by load adjustment using measured load
values.
prediction/load_aggregator:
- Module removed. Load aggregation is now handled by the measurement module.
prediction/load_corrector:
- Module removed. Load correction (aka. adjustment of load prediction by
measured load energy) is handled by the LoadAkkudoktor prediction and
the generic 'load_mean_adjusted' prediction key.
prediction/load_forecast:
- Module removed. Functionality now completely handled by the LoadAkkudoktor
prediction.
utils/cacheutil:
- Use pydantic.
- Fix potential bug in ttl (time to live) duration handling.
utils/datetimeutil:
- Added missing handling of pendulum.DateTime and pendulum.Duration instances
as input. Handled before as datetime.datetime and datetime.timedelta.
utils/visualize:
- Move main to generate_example_report() for better testing support.
server/server:
- Added new configuration option server_fastapi_startup_server_fasthtml
to make startup of FastHTML server by FastAPI server conditional.
server/fastapi_server:
- Add APIs for measurements
- Improve APIs to provide or take pandas datetime series and
datetime dataframes controlled by pydantic model.
- Improve APIs to provide or take simple datetime data arrays
controlled by pydantic model.
- Move fastAPI server API to v1 for new APIs.
- Update pre v1 endpoints to use new prediction and measurement capabilities.
- Only start FastHTML server if 'server_fastapi_startup_server_fasthtml'
config option is set.
tests:
- Adapt import tests to changed import method signature
- Adapt server test to use the v1 API
- Extend the dataabc test to test for array generation from data
with several data interval scenarios.
- Extend the datetimeutil test to also test for correct handling
of to_datetime() providing now().
- Adapt LoadAkkudoktor test for new adjustment calculation.
- Adapt visualization test to use example report function instead of visualize.py
run as process.
- Removed test_load_aggregator. Functionality is now tested in test_measurement.
- Added tests for measurement module
docs:
- Remove sphinxcontrib-openapi as it prevents build of documentation.
"site-packages/sphinxcontrib/openapi/openapi31.py", line 305, in _get_type_from_schema
for t in schema["anyOf"]: KeyError: 'anyOf'"
Signed-off-by: Bobby Noelte <b0661n0e17e@gmail.com>
|
||
|
Bobby Noelte
|
31bd2de18b |
Fix config and prediction revamp. (#259)
Extend single_test_optimization.py to be able to use real world data from new prediction classes. - .venv/bin/python single_test_optimization.py --real_world --verbose Can also be run with profiling "--profile". Add single_test_prediction.py to fetch predictions from remote prediction providers - .venv/bin/python single_test_prediction.py --verbose --provider-id PVForecastAkkudoktor | more - .venv/bin/python single_test_prediction.py --verbose --provider-id LoadAkkudoktor | more - .venv/bin/python single_test_prediction.py --verbose --provider-id ElecPriceAkkudoktor | more - .venv/bin/python single_test_prediction.py --verbose --provider-id BrightSky | more - .venv/bin/python single_test_prediction.py --verbose --provider-id ClearOutside | more Can also be run with profiling "--profile". single_test_optimization.py is an example on how to retrieve prediction data for optimization and use it to set up the optimization parameters. Changes: - load: Only one load provider at a time (vs. 5 before) Bug fixes: - prediction: only use providers that are enabled to retrieve or set data. - prediction: fix pre pendulum format strings - dataabc: Prevent error when resampling data with no datasets. Signed-off-by: Bobby Noelte <b0661n0e17e@gmail.com> |