BioTestLab

The system is designed for inspecting transparent glass vials containing a lyophilized (tablet) preparation, sealed with a rubber stopper and an aluminum cap. Inspection is carried out for vials of the following types: R6 – 6 ml and R10 – 10 ml.

The system is integrated into vial rolling machine or labeling machine, which will capture and transmit images to the server. The system processes the images and issues commands to the rejection node. The system should be equipped with an intuitive interface for easy control. The decision-making function for rejection should be based on machine learning using neural network technology.

• The system operates stably, with constant monitoring of its performance
• The system identifies product defects according to 6 parameters with high precision and accuracy

Development and setup of the system, datasets creation and training of neural network model, testing.

The machine checks vials for the following defects/non-defects based on the following criteria:

• Cosmetic defects of the vial (cracks, scratches, defects in the shape of the vial, etc.);
• Visible mechanical inclusions;
• Completeness of vial dosing;
• Absence of lyophilizate in the vial;
• Presence of liquid in the vial;
• Quality of the formed lyophilized tablet (foam, cracks, etc.).

The machine performs an inspection of each vial in automatic mode and provides 100% rejection of detected defects onto a separate tray for defects. False rejection of non-defective vials is allowed but not more than 1% of the total number of vials in the batch.

• The operator interface displays current information about the machine’s operating status, rejection rates, self-diagnostic results, alarms, deviations from controlled parameters, malfunction detection, and notification of alerts.
• Data is collected in a production report, which includes the following information: the date and time of the start of the shift, the date and time of the start and end of the inspection, the product type and batch number, and the total number of inspected and rejected vials.
• The vial inspection system is designed to be operated and serviced by a single operator.

• Camera transmits images of inspected vials to the server for further processing.
• Rejection block moves vials with detected defects to a tray with a defect size of 450x300mm. At the beginning of the work, the operator sets the alarm limits, which will be determined by the number of vials in the series. Rejection occurs without damaging the vial for further evaluation by the operator.
• Server with software processes the image received from the camera, recognizes vials with defects, and sends a command to the rejection block to move the defect to the tray.
• Operator interface – a web interface on the server is used to display information and input settings. The operator interface displays information about the number of processed vials and the number of rejected ones. Access to the interface is provided through a local network and password protection.

• Server system unit or cloud environment
• 10-port PoE switch
• Uninterruptible power supply
• Camera
• Rejection block in a steel case.