Real time machine learning based predictive and preventive maintenance of vacuum pump

A machine learning architecture comprising:

• a) a vacuum pump including a motor and a blower;
• b) the motor being associated with a machine wearable motor sensor;
• c) the blower being associated with a machine wearable blower sensor;
• d) a communications network receiving data from the motor sensor and the blower sensor, the motor sensor data comprising motor temperature, motor vibration, and motor sound, and the blower sensor data comprising blower vibration, blower sound, and blower temperature;
• the communications network providing the motor sensor data and the blower sensor data to a machine learning module;
• e) the machine learning module classifying the motor sensor data and the blower sensor data as either vacuum state sensor data or as break state sensor data, with the motor sensor data and the blower sensor data being classified individually one by one and in combination, the break state sensor data being received when a rotor of a vacuum pump is malfunctioning;
• f) the machine learning module analyzing the vibration data in the vacuum state sensor data classification to detect an operating vacuum level;
• g) an alarm device connected to the machine learning module by the communications network for providing a visible alarm when the vacuum state sensor data of one of a motor or blower vibration sensor and a motor or blower temperature sensor exceeds a pre-defined safety range;
• h) the machine learning module classifying the vacuum break data into either a clean filter category and a clogged filter category, wherein a visible alarm is raised when sensor data under the clogged filter category is detected;
• and
• i) the machine learning module analyzing the blower sensor data and the motor sensor data based on machine learning by transferring vibrational energy from one axis to another to detect the extent of oldness of the oil in the blower bearings based on smoothness of rotation and to detect instability and wobbling of a rigid rotational axis as indicative of at least one of a low oil level and an overused oil structure.

Internet of Things (IoT) based system for overseeing process control and predictive maintenance of a network of vacuum pumps employing machine wearable sensors associated with the vacuum pumps comprising:

• a) a machine learning module associated with a database;
• b) a plurality of machine-wearable sensors, each of which is secured to a vacuum pump, each sensor being capable of transmitting captured data over a wireless communications network;
• c) a machine learning algorithm engine receiving temperature, vibration and sound data from the sensors via the wireless communications network, for processing the received data including by transferring vibrational energy data for one axis to another axis to recognize at least one pattern and a deviation and issuing control commands, responsive to the deviation, pertaining to the vacuum pump;
• and
• d) one or more control modules for receiving, over the communications network, the control commands and executing same.

A method for monitoring a vacuum pump, comprising:

• a) receiving via a communications network motor sensor data and blower sensor data from the vacuum pump, said blower sensor data comprising vibration data, temperature data and sound data;
• b) displaying on a mobile device sensor data received over the communications network;
• c) classifying the motor sensor data and the blower sensor data as either vacuum state data taken while the vacuum pump is operating or break state data taken while the vacuum pump is inoperative;
• d) analyzing the vacuum state data to detect an operating vacuum level;
• e) actuating a first human perceptible alarm when either of the vacuum state vibration data and the vacuum state temperature data exceeds respective predefined safety ranges;
• f) classifying vacuum break into clean filter and clogged filter categories;
• g) actuating a second human perceptible alarm if a data entry is made in the clogged filter category;
• h) analyzing the blower sensor data in association with the motor sensor data based on machine learning by transferring measured vibrational energy from one axis to another to detect the extend of oldness of the oil in the blower bearings based on smoothness of rotation and to detect instability and wobbling of rigid rotational axes denoting occurrence of low motor oil level and occurrence of low blower oil level;
• i) actuating a third human perceptible alarm upon detection of low motor oil level or low blower oil level;
• j) wherein the alarms are selected from the group comprising pressure state alarms, filter alarms, oil state alarms and blower alarms.

A method of operating a vacuum pump having a motor and a blower, consisting of:

• a) mounting at least one physical parameter sensor on the motor and one physical parameter sensor on the blower, the sensors producing data respecting sensed physical parameters;
• b) providing the sensed physical parameter data to a communications network;
• c) receiving, in an electronic device, sensor data from the motor sensor and from the blower sensor via the communications network;
• d) displaying, on the electronic device, sensor data received over the communications network;
• e) dividing the motor sensor data into vacuum state data and break state data;
• f) determining pump operating vacuum level from the vacuum state data;
• g) triggering a first audible alarm if the vacuum level exceeds a predefined safe range;
• h) dividing the break state data as either clean filter data or clogged filter data;
• i) applying at least one preselected criteria to the clean filter data to define a clean filter data pattern;
• j) applying at least one preselected criteria to the clogged filter data to define a clogged filter data pattern;
• k) dividing subsequent motor sensor data into vacuum state data or break state data;
• l) comparing the break state data to the clean filter data pattern to detect any deviation therefrom;
• m) comparing the break state data to the clogged filter data pattern to detect any deviation therefrom;
• n) providing a human sensory perceptible second alarm signal on the electronic device indicating deviation of the subsequently received motor sensor break state data from either the clean filter data pattern or from the clogged filter data pattern;
• o) changing vacuum pump oil and oil filter if the signal indicates deviation from the clean filter data pattern;
• p) halting operation of the vacuum pump if a third alarm signal indicates deviation from the clogged filter data pattern.

A method of operating a granular resin material drying machine having a vacuum pump with a motor and a blower, consisting of:

• a) mounting at least one physical parameter sensor on the motor and one on the blower, the sensors producing data respecting sensed physical parameters;
• b) providing the sensed physical parameter data to a communications network
• c) receiving, in a portable electronic device, via the communications network, sensor data from the motor sensor and from the blower sensor;
• d) displaying on the portable electronic device sensor data received via the communications network;
• e) categorizing the motor sensor data as either vacuum state data and break state data;
• f) determining operating vacuum level from the vacuum state data;
• g) triggering an alarm if the vacuum level exceeds a predefined safe range;
• h) categorizing the break state data as either clean filter data or clogged filter data;
• i) triggering an alarm if the break state data is indicative of a clogged filter;
• j) algorithmically processing the motor sensor data and the blower sensor data to detect a deficient motor oil level and defective motor oil structure if at least one preselected criteria is met;
• k) providing a human perceptible alarm signal if deficient motor oil level or deficient motor oil structure is detected;
• l) wherein the alarms are selected from the group comprising pressure state alarms, filter alarms, oil state alarms and blower alarms.