IP Library Granted Patent US 12694646
Granted Patent B2
US 12694646 · App. 18/197,281 · Granted Jul 28, 2026

State detection apparatus

Inventors: Takashi Oshima (Tokyo, JP); Keisuke Yamamoto (Tokyo, JP); Goichi Ono (Tokyo, JP)
Assignee: Hitachi, Ltd.
G06V10/764G06T5/10G06T2207/20052G06T2207/20056G06T2207/20084
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12694646
App. No.
18/197,281
Granted
Jul 28, 2026
Kind
B2
Abstract

The time-series signal of the sensor is transformed to the spectral intensity by fast Fourier transform (FFT) or the like, and the one-dimensional data of the spectral intensity is generated. A pseudo image is generated, for example, by repeatedly arranging the one-dimensional data in the vertical direction, or by arranging the one-dimensional data for a plurality of sensors in the vertical direction. The state of the facility is identified by analyzing the pseudo image with an image analysis unit such as a convolutional neural network.

Claims (62)

1 . A state detection apparatus comprising:

a plurality of sensors; and

a processing apparatus coupled with the plurality of sensors and configured to execute data processing operations comprising

transforming data of a plurality of digitized time-series signals each associated with one of the plurality of sensors into data relating to spectral intensity comprising a spectral intensity vector for each of the digitized time-series signals;

quantizing each spectral intensity vector into a quantized spectral intensity vector;

generating a pseudo image based on the data relating to the spectral intensity by generating a two-dimensional image by arranging each quantized spectral intensity vector corresponding to one of the plurality of sensors as one or more rows in a vertical direction; and

analyzing the pseudo image and outputting a classification result of a state of a facility based on the analysis of the pseudo image,

wherein a width in pixels of the generated pseudo image is equal to a number of dimensions of the quantized spectral intensity vector,

wherein a vertical size of the pseudo image in a number of pixels is equal to a product of the number of sensors and a number of the one or more rows, and

wherein a pixel value of the pseudo image is equivalent to a frequency spectral intensity of the digitized time-series signal of the sensor such that the pseudo image is independent of a phase of the digitized time-series signal.

2 . The state detection apparatus according to claim 1 , wherein

the processing apparatus performs the transformation by discrete Fourier transform or fast Fourier transform.

3 . The state detection apparatus according to claim 2 , wherein

the processing apparatus analyzes the pseudo image by a convolutional neural network.

4 . The state detection apparatus according to claim 3 , wherein

the processing apparatus performs a process of transforming a value corresponding to the spectral intensity into a non-linear quantization, in the transformation.

5 . The state detection apparatus according to claim 4 , wherein

the processing apparatus performs a process of replacing a part of the value of the data relating to the spectral intensity, in the transformation.

6 . The state detection apparatus according to claim 4 , wherein

the plurality of sensors is included in a plurality of sensor groups each including different types of sensors,

the plurality of sensor groups is disposed at different inspection points, and

the processing apparatus generates a plurality of pseudo images, wherein

when generating the plurality of pseudo images, data relating to the spectral intensities of the sensors of the same type is arranged in a same area among the plurality of pseudo images to be generated.

7 . The state detection apparatus according to claim 4 , wherein

the plurality of sensors is included in a plurality of sensor groups each including different types of sensors,

the plurality of sensor groups is disposed at different inspection points, and

the processing apparatus generates a plurality of pseudo images, wherein

when generating the plurality of pseudo images, data relating to the spectral intensities of the sensors located at the same inspection point is arranged in a same area among the plurality of pseudo images to be generated.

8 . The state detection apparatus according to claim 3 , wherein

the processing apparatus sets a time width or period of discrete Fourier transform or fast Fourier transform based on an output of the convolutional neural network.

9 . The state detection apparatus according to claim 3 , further comprising:

a camera for acquiring a camera image; and

an input apparatus for a user to use, wherein

the processing apparatus processes the camera image or the pseudo image selected by the user with the convolutional neural network.

10 . The state detection apparatus according to claim 1 , wherein

the processing apparatus analyzes the pseudo image by a convolutional neural network.

11 . The state detection apparatus according to claim 1 , wherein

the processing apparatus executes the transformation using at least one of discrete cosine transform, discrete sine transform, discrete Fourier transform that is slower than fast Fourier transform, and Walsh-Hadamard transform.

12 . The state detection apparatus according to claim 1 , wherein

the processing apparatus executes the transformation using a combination of at least two of discrete cosine transform, discrete sine transform, discrete Fourier transform, fast Fourier transform, and Walsh-Hadamard transform.

13 . The state detection apparatus according to claim 1 , further comprising:

an allocation table that stores an allocation of orthogonal transforms used for the transformation, wherein

the processing apparatus executes the transformation using at least one of discrete cosine transform, discrete sine transform, discrete Fourier transform, fast Fourier transform, and Walsh-Hadamard transform, based on the allocation table.

14 . The state detection apparatus according to claim 13 , wherein

the processing apparatus updates the allocation table based on information input during on-chip training.

15 . A system using the state detection apparatus according to claim 1 , the system comprising:

a computer for determining a combination of an orthogonal transform used for the transformation and an image analysis processing algorithm based on detection accuracy and resource information of the state detection apparatus, and for configuring the state detection apparatus according to the result of the determination, or for sending an instruction to configure the state detection apparatus according to the result of the determination.

16 . A state detection apparatus comprising:

a sensor; and

a processing apparatus coupled with the sensor and configured to execute data processing operations comprising

transforming data of the digitized time-series signals of the sensor into data relating to spectral intensity comprising a spectral intensity vector for each of the digitized time-series signals;

generating a pseudo image based on the data relating to the spectral intensity by generating a two-dimensional image by arranging each spectral intensity vector as one or more rows in a vertical direction; and

analyzing the pseudo image and outputting a classification result of a state of a facility based on the analysis of the pseudo image,

wherein a width in pixels of the generated pseudo image is equal to a number of dimensions of the spectral intensity vector,

wherein a vertical size of the pseudo image in a number of pixels is equal to a product of the sensor and a number of the one or more rows, and

wherein a pixel value of the pseudo image is equivalent to a frequency spectral intensity of the digitized time-series signal of the sensor such that the pseudo image is independent of a phase of the digitized time-series signal.

17 . The state detection apparatus according to claim 16 , wherein

the processing apparatus performs the transformation by discrete Fourier transform or fast Fourier transform.

18 . The state detection apparatus according to claim 17 , wherein

the processing apparatus analyzes the pseudo image by a convolutional neural network.

19 . The state detection apparatus according to claim 16 , wherein

the processing apparatus analyzes the pseudo image by a convolutional neural network.