# Available data Read this page before you ask the assistant a question. It tells you which SolarSENS data the assistant can read, at which level, and at which time resolution. If your question needs data that is not on this page, the assistant cannot answer it from SolarSENS. The data is the same for O\&M users and for internal operations users. Each person sees only the Plants that their SolarSENS account can access. ## Levels and resolutions | Level | Data | Resolution | Time range in one question | Tool | | --------- | ---------------------------------------------------------------------------------------------------------------- | --------------------------------------- | -------------------------------------------------------------------------------------- | --------------------------------------------------------------------- | | Portfolio | Operation Overview table of each Plant | One row for each Plant for one day | One day | `get_portfolio_dashboard` (`site_overview`) | | Portfolio | Yield for the last 7 days | Daily | 7 days | `get_portfolio_dashboard` (`last_7_days_yield`) | | Portfolio | Device status and string health summary | Current status | Now | `get_portfolio_dashboard` (`devices_status_summary`, `string_health`) | | Portfolio | Alarm status, as on the Operation Overview screen | Alarms in the time range | Up to 92 days | `get_portfolio_dashboard` (`alarm_status`) | | Plant | Yield, PR, availability, irradiation | Daily | Up to 366 days | `get_plant_daily_performance` | | Plant | Stored DataHub Plant metric values | 5 minutes, as stored | Up to 31 days, but 1,000 rows allow about 3 days for one metric or 1 day for 3 metrics | `query_plant_5min` | | Plant | Other stored daily values, including curtailment, soiling, revenue, temperature-corrected PR, and monthly values | Daily, as stored | Up to 366 days, maximum 1,000 rows | `query_plant_daily` | | Plant | Power, yield, and performance charts | 1, 5, 15, 30, or 60 minutes | One day | `get_plant_chart` | | Plant | Live production and device communication status | Current status | Now | `get_plant_dashboard` | | Plant | Disconnected strings | Current status | Now | `get_plant_dashboard` (`disconnected_strings`) | | Plant | Active alarms, alarm history, and details | Current alarm state and lifecycle times | The selected range filters alarm creation time; ten alarms per page | `list_alarms`, `get_alarm` | | Device | Device IDs, names, types, parents, labels, and attributes | Not time-based | Not applicable | `list_devices` | | Device | Named telemetry (for example `Active Power`) | 1, 5, 15, 30, or 60 minutes | One day, one device, up to 10 metrics | `query_device_telemetry` | | Inverter | Daily yield, availability, derating loss, operation time, and unavailability, overspill, and soiling loss | Daily | Up to 366 days, all inverters or selected inverters | `query_inverter_daily` | | String | String analytics, as on the String Analytics screen (for example average current and normalized power) | One record for each string for each day | One day for each question | `get_string_daily_analytics` | | String | Stored current (`Idc`) and voltage (`Udc`) of each string | 5, 15, 30, or 60 minutes | Up to 7 days, from approximately the last 72 days | `query_string_5min` | ## How many data points the assistant uses The resolution sets how strong a conclusion can be. Ask the assistant to tell you which data it used. * **Daily values** give one data point for each Plant, device, or string for each day. Use them to find candidates for an investigation. They do not show when a fault started during the day. * **Plant time series** give one row for each metric at each time. One answer can use a maximum of 1,000 rows. Select the metric keys from the tool schema. The assistant reads the values as stored. The retention period is unknown. The daily refusal limit estimates `ceil(span / 24 hours) + 1` samples for each metric. A request can fail this limit even if fewer rows are stored. * **Chart and telemetry values** give up to 288 data points for each metric for each day at 5-minute resolution. Use them to see the shape of a day. * **String analytics** are processed daily values. For a Plant with 114 strings, one day gives 114 records. * **String current and voltage** give one record for each string at each time step. For a Plant with 114 strings, one day at 5 minutes gives approximately 16,500 records. One answer can use a maximum of 1,000 records. Thus, start with the strings of one inverter at 15 or 60 minutes. Then ask for 5 minutes for a short time range. If a question needs more than 1,000 records, the assistant tells you to make the question smaller. ## Example questions that work Replace the text in square brackets with your own details. ``` Show the stored 5-minute power and irradiance at [Plant name] from [start time] to [end time] on [date]. Tell me which values are missing. ``` ``` Show daily yield, PR, and availability at [Plant name] from [start date] to [end date]. State the timezone and tell me which days have missing values. ``` ``` Show [Plant name] power and irradiance on [date] at 5-minute intervals. ``` ``` Analyze the string analytics of [Plant name] for [date]. List the weakest strings and tell me how many records you used. ``` ``` Compare the string analytics of [Plant name] for [date], [date], and [date]. Tell me which weak strings stay weak on all three days. ``` For a comparison over many days, ask for each day by name. The assistant reads string analytics one day at a time. ``` Show the current and voltage of the strings of [inverter] at [Plant name] on [date] at 15-minute resolution. Tell me which strings have low current. Then show those strings at 5-minute resolution from [start time] to [end time]. ``` ``` Show daily yield, availability, and soiling loss of each inverter at [Plant name] from [start date] to [end date]. Tell me which inverter has the lowest yield. ``` ## Data that is not available yet The assistant cannot read the data in this list. Do not ask for it, or the assistant can only tell you that the data is not available. * String current and voltage older than approximately 72 days. * Inverter loss values at a resolution of less than one day. * Work orders. * Month and year charts, heat maps, the loss waterfall, and insights charts. * DataHub metric values for devices other than the stored string and inverter values listed above. Every tool only reads data. The assistant cannot acknowledge an alarm, change a setting, or control a Plant or a device. ## Read the answer carefully * A missing value is not zero. If the assistant reports missing values, the data for that time did not arrive. For device telemetry, an empty series can also mean a wrong metric key. The assistant states the key that it used. * Daily and processed values show candidates for an investigation. They do not prove a hardware fault. Make sure that a site check or higher-resolution data confirms the fault. * Use the Plant-local date. The assistant uses the time zone of the Plant. A clock time with no zone is Plant time. If you name a zone, for example SG (UTC+8) or VN (UTC+7), the assistant changes the time to Plant time and states the two times. * Device telemetry covers one Plant-local day, today or a past day.