Decision-making matrix is a set of rules and conditions under that Logic Manager runs the specified macros when certain events occur.
To manage the decision rules you may use eva lm (lm-cmd) console applications, LM API functions or an appropriate LM EI interface section.
Rule configuration is stored in runtime/lm_dmatrix_rule.d/ folder.
Event analysis algorithms
Event means any change of the item state. The events are analyzed and processed in the following way:
A specific ID is assigned to each event by which you can monitor its processing in the logs.
The system takes the list of all rules sorted by their priority (the lower the value, the higher the priority) and description.
Each rule is analyzed separately, in the order of priority
Firstly, it’s checked whether the rule corresponds to the type, group, ID and property of the item that sent the event
Then, the system is verifying whether the current item state matches the rule conditions and the previous one is out of its range
For example, there is a temperature sensor; the condition 25 <= x <= 28 is specified in the rule; it will match only once - as soon as the transmitted temperature reaches 25-28 degrees. The rule will match again if the temperature exceeds this range and returns back.
When the controller is just started, the previous state is unknown or usually outdated. In this case, the system acts according to the configuration property of the rule for_initial: if it’s set to skip, the rule is ignored, if only or any, it’s verified and the rule matches if the current state matches the range.
If for_initial is set to only in the rule configuration, the rule is being checked only once, after the controller is started and the initial states of the items are received.
Logic variables always have an initial value stored in the local base, that’s why for_initial should always be any for them to let the rules work correctly, unless you really know what you do.
If the rule matches, a macro (if specified) with the specified arguments is executed.
If chill-out_time > 0 in the configuration, the rule is ignored after the match for the specified time. If rule matched during chill-out, macro is executed once, after chill-out period ends.
Rules can be created with either LM API create_rule function or with EVA shell.
To configure rule you may specify condition and action, or you may set rule parameters one-by-one after the rule is created.
To specify condition and action during rule creation, use the following format. Note that controller doesn’t check is condition item and/or macro exists on the moment of rule creation:
rule create if <condition> [then <action>]
Example, start unit unit:ventilation/v1 (call start macro function) if value of sensor:env/temp is more than 25:
rule create if sensor:env/temp.value > 25 then @start('unit:ventilation/v1')
Another example. Run macro macro1 if value of lvar lvar:tests/lvar1 is more than 25 but less than 35:
rule create if 35 > lvar:tests/lvar1.value > 25 then macro1()
Check only 3rd bit of value:
rule create if sensor:env/plc_state.b3 == 1 then macro1()
New rule is always created as “disabled” and you must enable it with “rule enable” CLI command or call LM API function set_rule_prop, setting enabled=True.
Unmodifiable rule parameters:
id rule id, always generated automatically when it is created
chillout_ends_in a virtual parameter specifying for how long (in seconds) the rule is ignored, if chillout_time is set
break_after_exec if True and the rule matches, further rules for the event are ignored
chillout_time the rule is ignored for a specified time (in seconds) after match
condition “virtual” parameter which allows get/set rule condition in the readable format (e.g. 25 < x <= 28)
description rule description
enabled if True, rule is enabled (new rules are disabled by default)
for_initial can be skip, only or any (default is skip). Indicates the rule processing algorithm when the server is started and the initial item states are received
for_item_group the rule matches only for a specific group of items (# or null - for all groups)
for_item_id the rule matches only for a specific item (# or null - for all items), may contain the mask *id, id* or *id*, i.e. *.temperature
for_item_type the rule matches only for a specific type of items (# or null - for all types)
for_oid “virtual” parameter which allows get/set rule condition in the readable format (e.g. sensor:group1/#.value)
for_prop the state property of the item (status or value) the rule is checking. For unit state, nstatus and nvalue properties may be additionally used.
for_prop_bit if set to a number (0-X), only state Xth bit is compared. Obviously, the compassion condition should be set either == 0 or == 1.
in_range_max matches when x < value
in_range_max_eq matches when x <= value (in_range_max should be specified)
in_range_min matches when x > value
in_range_min_eq matches when x >= value (in_range_min should be specified)
macro macro that is executed when the rule conditions match
macro_args arguments the macro is executed with
macro_kwargs keyword arguments the macro is executed with
priority the rule priority (integer; the lower the value, the higher the priority, 100 by default)
Tips for rule configuration
to set “x == value” condition via lm_api: if the value is numeric, use “value <= x <= value”. If the value is string, you may set only in_range_min_eq
if you set a field for_expire (with any value, i.e. Y) in a rule change request, the system automatically sets the rule to for_prop = status, x <= -1, which means the rule match when the item state is expired. This is useful to configure the rule to check for the lvar timers expiration as well as checking for units and sensors error states
if you set a field for_set (with any value, i.e. Y) in a rule change request, the system automatically sets the rule to for_prop = status, x == 1, which means the rule match when the item state is set. This is useful to configure the rule to check for the lvar timers reset as well as working with a logical flags
if the rule has no in_range_min and in_range_max conditions, it will match each time when the item changes its status (for_prop == status) or value (for_prop == value)
If rule option for_initial is set to any or only, it is possible to cache previous item state to prevent false rule triggering. This may cause additional system overload.
To enable state cache, set “plc/cache-remote-state” field in config/lm/main registry key to desired cache time-to-live in seconds (e.g. 604800 = time-to-live 1 week) and restart the controller.
Or use “feature” command:
eva feature setup lm_state_cache ttl=604800