Elevator noise detection system

Abstract

This invention provides an elevator noise detection system that can efficiently detect abnormal noises in elevators, which are constantly changing and operate under many different conditions. [Solution] The elevator abnormal noise detection system 10 is installed in the elevator car 3 and includes a measuring unit that collects operating data of the operating state of the elevator car 3, a learning unit that generates a trained predictive model of the first operating data based on the first operating data in the first operating state of the elevator, a first calculation unit that obtains predictive data by taking the second operating data in the second operating state of the elevator as input to the first trained predictive model, and a second calculation unit that calculates an abnormality score based on the degree of discrepancy between the second operating data and the predictive data.

Claims

1. A measuring unit installed in the elevator car to collect operational data on the operating status of the elevator car, A learning unit generates a trained predictive model of the first operating data based on first operating data continuously measured over the entire operating pattern, which includes all possible operating states of the elevator car's position and speed, in the first operating state of the elevator. A first calculation unit obtains prediction data by taking the second operating data in the second operating state of the elevator as input to the trained prediction model, A second calculation unit calculates an anomaly score based on the degree of discrepancy between the second operating data and the predicted data, An elevator noise detection system equipped with [specific feature].

2. The elevator noise detection system according to claim 1, wherein the first operating state is the normal operating state.

3. An elevator noise detection system according to claim 1, wherein a trained predictive model of the operating data is generated by continuously measuring the operating pattern multiple times.

4. An elevator noise detection system according to claim 1, which generates a trained predictive model and calculates an anomaly score using driving data in which the position of the elevator car in the aforementioned driving pattern is converted into a one-hot vector.

5. The elevator noise detection system according to claim 1, wherein the neural network used to generate the trained predictive model is a recurrent neural network.

6. The elevator noise detection system according to claim 5, wherein the recurrent neural network is an LSTM neural network.

7. The elevator noise detection system according to claim 1, wherein the neural network used to generate the trained predictive model is a Transformer neural network.

8. An elevator noise detection system according to any one of claims 1 to 7, wherein the aforementioned operating data includes acoustic data.

9. An elevator noise detection system according to claim 8, which divides the acoustic data into intervals of a certain amount of time, converts the acoustic data divided into intervals of a certain amount of time into frequency data, generates a trained predictive model using the frequency data, and detects abnormal noise or a precursor to the abnormal noise.

10. The elevator noise detection system according to claim 8, wherein the aforementioned operating data includes speed data.

11. An elevator noise detection system according to claim 1, comprising an output unit that outputs the calculated score.

Images