Configure algorithm instances

I-V curve diagnosis

Perform pattern analysis on string I-V curves by configuring arrays to detect faults in photovoltaic (PV) plant strings. This allows for simple and efficient Operations and Maintenance (O&M) and reduces costs. For detailed instructions, see I-V curve diagnosis.

Create an array

1. On the Algorithm Instances tab, find the instance for which the Algorithm Template is PV Intelligent O&M - IV Curve Diagnosis, and then click View in the Actions column.

2. On the instance details page, click Create Array.

3. In the panel that appears on the right, configure the array information as prompted.

   1. Set the Basic Information, and then click Next.

      Parameter	Description
      Array name	Can contain digits, letters, and underscores (_). Cannot start with a digit. Maximum length is 32 characters.
      Array description	Optional. Enter a description for the array.

   2. Select the Brand and Model, and then click Next.

      You can select a Module Brand and Module Model. The following performance metrics are available for detection:

      Note

      If the module brands and models provided by IoT Platform do not meet your business needs, you can submit a ticket to specify your required brands and models.

      Jinko Solar

      JA Solar

      LONGi Green Energy

      Astronergy

      Trina Solar

      

4. In the Module Count section, set the Number of Modules in Parallel (an integer from 1 to 6) and Number of Modules in Series (an integer from 1 to 10). You can preview the array diagram.

   

5. Click Finish.

6. (Optional) To modify array information, find the target array in the array list, and click Edit in the Actions column. Then, you can modify the Basic Information, Brand and Model, and Module Count.

Configure IV detection

1. In the array list, find the target array and click IV Detection in the Actions column.

2. In the panel that appears on the right, configure the detection data. This includes Ambient Light Intensity, Environmental Temperature, Collection Time, and Upload IV Data.

   The valid ranges and units for Ambient Light Intensity, Environmental Temperature, and Collection Time are shown in the input field prompts.

   

3. Click Start Detection.

4. After the detection is complete, a dialog box appears with the detection results. You can view the Diagnosis Status and the corresponding Probability.

   Click Expand I-V Curve Graph to view the detected I-V curve.

   

If the algorithm returns an error, the detection fails. Correct the input parameters based on the error message and repeat the preceding IV detection steps.

The error response parameters are as follows:

Optional: View detection records

1. On the Instance Details page, click Detection Records in the Actions column of the target array.

2. Specify the Detection Time. Click View Results in the Actions column of the target record to view its detection results.

   

Power time series curve diagnosis

Data Service provides power time series curve diagnosis for single-site detection and multi-site ranking. Depending on the algorithm instance that you purchased, you can upload power time series data to detect the efficiency of photovoltaic power generation.

For more information about single-site detection and multi-site ranking for power time series curve diagnosis, see Usage instructions and scenarios for power time series curve diagnosis.

Time series anomaly detection

Data Service provides two types of anomaly detection: anomaly points and anomaly intervals. You can upload device time series data based on the detection type that you select to detect anomalies in the time series. For detailed instructions, see Time series anomaly detection.

1. On the Algorithm Instances tab, find the Time Series Anomaly Detection algorithm instance, and then click View in the Actions column.

2. In the Detection List section, click Time Series Anomaly Detection.

3. In the Time Series Anomaly Detection panel that appears on the right, configure the data to be detected.

   Parameter	Description
   Anomaly detection type	Options: Anomaly point: Detects anomaly points in device time series data and highlights anomalous data points. Anomaly interval: Detects anomaly intervals in device time series data. Adjacent intervals with significant data fluctuations are displayed in different colored blocks.
   Detection sensitivity level	This parameter appears only when Anomaly detection type is set to Anomaly point. Options: High, Medium (default), Low. For the same batch of time series data, the number of detected anomaly points decreases as sensitivity level decreases from high to low. Higher sensitivity levels detect more anomaly points.
   Window size	This parameter appears only when Anomaly detection type is set to Anomaly interval. Sets the number of consecutive time series points selected in one sliding window during detection. The algorithm compares the shape differences between the first and second halves of the window to determine if a change point exists within the window. Larger windows include more data points, making interval statistical features more pronounced. Note A large window may overlook local small-interval anomalies. Set the window size based on your specific metric data and business scenario. Window size must be an even number between 20 and 2000. Default is 200 if not set.
   Penalty value	This parameter appears only when Anomaly detection type is set to Anomaly interval. Sets the sensitivity for detecting anomalous fluctuation intervals: A smaller penalty value increases detection sensitivity and detects more change points. A larger penalty value decreases detection sensitivity and detects fewer change points. Penalty value range is 0 to 10000. Default is 10 if not set.
   Upload time series anomaly detection data	Click Download Sample Data to get a CSV file. The file contains the following fields: value: Data to be detected. dates: Timestamps of the data to be detected, in YYYY-MM-DD HH:mm:ss format. We recommend a 5-minute interval between timestamps. Important All fields above are required and must meet the specified requirements. Detection fails if any field is empty or incorrect. Uploaded file size cannot exceed 50 KB. After successful parsing of the uploaded time series anomaly detection data, the time series anomaly curve graph appears.

4. Click Start Detection.

5. After the detection is complete, a detection results dialog box appears. You can view the highlighted anomaly data.

   Click the zoom-in icon to expand the time series anomaly detection graph and view anomaly data with the corresponding timestamps.

   

6. (Optional) On the Instance Details page, specify the Detection Time . Click View Results in the Actions column of the target record to view its detection results.