Failure predictor detection device, failure predictor detection method, training device, and trained learning model generation method

A failure predictor detection device comprising processing circuitry:

• to acquire estimation data and comparison data of targeted equipment for failure predictor detection in a targeted time period for the failure predictor detection;
• to calculate an estimated value of the comparison data during a normal operation from the estimation data using a learning model;
• to detect a failure predictor of the equipment based on comparison results at multiple times between the estimated values and measured values shown by the comparison data;
• and
• to select, from among the acquired estimation data and the acquired comparison data, estimation data and comparison data in a time period during which a variance value is smaller than a predetermined selection threshold and to output the selected estimation data and the selected comparison data to the processing circuitry, the variance value being a value of variance in the data in the time period,
• wherein the processing circuitry calculates the estimated value using the selected estimation data and detects the failure predictor of the equipment or an environmental change around the equipment based on the comparison results at multiple times between the measured values shown by the selected comparison data and the estimated values, and
• wherein the equipment is an air conditioner, and the processing circuitry detects the failure predictor of the air conditioner based on the estimation data and the comparison data, the comparison data being compressor frequency, the estimation data being at least one of a time of day, a power consumption, an outdoor temperature, a room temperature, a set temperature, an operating condition, a refrigerant temperature, values from various thermistors, a fan speed, and values from various pressure sensors in the air conditioner
• wherein the processing circuit calculates a degree of deviation between the estimated value and the measured value, and
• wherein the processing circuitry detects a short-term environmental change around the equipment when a time period during which the degree of deviation exceeds the predetermined threshold is shorter than a predetermined time period and detects the failure predictor of the equipment or a long-term environmental change around the equipment when a time period during which the degree of deviation exceeds a predetermined threshold is equal to or longer than the predetermined time period.