# US 2020/0401092 A1 · Environment controller and method for predicting co2 level variations based on sound level measurements

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2020/0401092 A1
- Document ID: US20200401092A1
- Status: Published application
- Kind: A1
- Application number: 16/445,729
- Filed: 2019-06-19
- Published: 2020-12-24
- Earliest claimed: 2019-06-19
- Applicants: Distech Controls Inc.
- Inventors: Francois GERVAIS
- Web page: https://patentgrep.com/patent/US20200401092A1

## CPC

- G05B 15/02: electric
- G05B 19/042: using digital processors (G05B19/05 takes precedence)
- G06N 3/04: Architecture, e.g. interconnection topology
- G06N 3/045: Combinations of networks
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/09: Supervised learning
- G05B 13/027: {using neural networks only}
- G05B 13/048: {using a predictor}
- G05B 2219/25011: Domotique, I-O bus, home automation, building automation
- G05B 2219/2642: Domotique, domestic, home control, automation, smart house
- G06N 3/084: Backpropagation, e.g. using gradient descent
- G05B: CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS (systems for controlling or regulating non-electric variables G05D; systems for regulating electric or magnetic variables G05F; control devices or systems insofar as characterised by mechanical features only G05G)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

Method and environment controller predicting CO2 level variations based on sound level measurements. The environment controller determines N consecutive sets of frequency domain sound level measurements. Each set of frequency domain sound level measurements comprises a given number M of sound level amplitudes at the corresponding given number M of frequencies. The environment controller executes a neural network inference engine using a predictive model for inferring one or more output based on inputs. The inputs comprise the N consecutive sets of frequency domain sound level measurements. The one or more output comprises a predicted variation of a CO2 level. For example, the environment controller receives a plurality of consecutive time domain sound level measurements from a sound sensor and generates the N consecutive sets of frequency domain sound level measurements based on the plurality of consecutive time domain sound level measurements (for instance by using a Fast Fourier Transform algorithm).

## Claims

### Claim 1

An environment controller comprising:
at least one communication interface;
memory for storing a predictive model; and
a processing unit for:
determining N consecutive sets of frequency domain sound level measurements, each set of frequency domain sound level measurements comprising a given number M of sound level amplitudes at the corresponding given number M of frequencies, N and M being integers; and
executing a neural network inference engine using the predictive model for inferring one or more output based on inputs, the inputs comprising the N consecutive sets of frequency domain sound level measurements, the one or more output comprising a predicted variation of a Carbon Dioxide (CO2) level.

### Claim 2 (depends on 1)

The environment controller of claim 1, wherein determining the N consecutive sets of frequency domain sound level measurements comprises receiving the N consecutive sets of frequency domain sound level measurements via the at least one communication interface.

### Claim 3 (depends on 2)

The environment controller of claim 2, wherein the N consecutive sets of frequency domain sound level measurements are received from a sound sensor.

### Claim 4 (depends on 1)

The environment controller of claim 1, wherein determining the N consecutive sets of frequency domain sound level measurements comprises:
receiving a plurality of consecutive time domain sound level measurements via the at least one communication interface; and
generating the N consecutive sets of frequency domain sound level measurements based on the plurality of consecutive time domain sound level measurements.

### Claim 5 (depends on 4)

The environment controller of claim 4, wherein the plurality of consecutive time domain sound level measurements is received from a sound sensor.

### Claim 6 (depends on 4)

The environment controller of claim 4, wherein the generation of the N consecutive sets of frequency domain sound level measurements based on the plurality of consecutive time domain sound level measurements uses a Fast Fourier Transform algorithm.

### Claim 7 (depends on 1)

The environment controller of claim 1, wherein the processing unit further determines N consecutive CO2 level measurements corresponding to the N consecutive sets of frequency domain sound level measurements and the inputs further include the N consecutive CO2 level measurements.

### Claim 8 (depends on 7)

The environment controller of claim 7, wherein the determination of the N consecutive CO2 level measurements is based on data received via the at least one communication interface from a CO2 sensor.

### Claim 9 (depends on 1)

The environment controller of claim 1, wherein the processing unit further determines a CO2 level measurement corresponding to the N consecutive sets of frequency domain sound level measurements and the inputs further include the CO2 level measurement.

### Claim 10 (depends on 9)

The environment controller of claim 9, wherein the determination of the CO2 level measurement is based on data received via the at least one communication interface from a CO2 sensor.

### Claim 11 (depends on 1)

The environment controller of claim 1, wherein the one or more output further comprises a predicted variation of temperature.

### Claim 12 (depends on 11)

The environment controller of claim 11, wherein the processing unit further determines N consecutive temperature measurements corresponding to the N consecutive sets of frequency domain sound level measurements and the inputs further include the N consecutive temperature measurements.

### Claim 13 (depends on 12)

The environment controller of claim 12, wherein the determination of the N consecutive temperature measurements is based on data received via the at least one communication interface from a temperature sensor.

### Claim 14 (depends on 11)

The environment controller of claim 11, wherein the processing unit further determines a temperature measurement corresponding to the N consecutive sets of frequency domain sound level measurements and the inputs further include the temperature measurement.

### Claim 15 (depends on 14)

The environment controller of claim 14, wherein the determination of the temperature measurement is based on data received via the at least one communication interface from a temperature sensor.

### Claim 16 (depends on 1)

The environment controller of claim 1, wherein the processing unit further generates at least one command for controlling at least one controlled appliance and transmits the at least one command to the at least one controlled appliance via the at least one communication interface, the generation of the at least one command being based at least on the predicted variation of the CO2 level.

### Claim 17 (depends on 16)

The environment controller of claim 16, wherein the at least one controlled appliance comprises a Variable Air Volume (VAV) appliance.

### Claim 18 (depends on 1)

The environment controller of claim 1, wherein each sound level amplitude at the corresponding frequency consists of a sound pressure at the given frequency, a sound pressure level at the given frequency or a sound power at the given frequency.

### Claim 19 (depends on 1)

The environment controller of claim 1, wherein the neural network inference engine implements a neural network comprising an input layer, followed by fully connected layers; the input layer comprising N*M neurons respectively receiving the sound level amplitudes; the predictive model comprising weights for the fully connected layers.

### Claim 20 (depends on 1)

The environment controller of claim 1, wherein the neural network inference engine implements a neural network comprising one input layer, followed by at least one one-dimensional convolutional layer, followed by fully connected layers; the input layer comprising M neurons respectively receiving a one-dimension matrix, each one-dimension matrix comprising N sound level amplitudes at a given frequency among the M frequencies, the at least one one-dimensional convolutional layer applying a one-dimensional convolution to each one-dimension matrix; the predictive model comprising weights for the fully connected layers and parameters for the at least one one-dimensional convolutional layer.

### Claim 21 (depends on 20)

The environment controller of claim 20, wherein the neural network further comprises at least one pooling layer.

### Claim 22 (depends on 1)

The environment controller of claim 1, wherein the neural network inference engine implements a neural network comprising an input layer, followed by at least one two-dimensional convolutional layer, followed by fully connected layers; the input layer comprising one neuron receiving a two-dimensions matrix comprising the N*M sound level amplitudes, the at least one two-dimensional convolutional layer applying a two-dimensional convolution to the two-dimensions matrix; the predictive model comprising weights for the fully connected layers and parameters for the at least one two-dimensional convolutional layer.

### Claim 23 (depends on 22)

The environment controller of claim 22, wherein the neural network further comprises at least one pooling layer.

### Claim 24

A method for predicting carbon dioxide (CO2) level variations based on sound level measurements, the method comprising:
storing a predictive model in a memory of a computing device;
determining by a processing unit of the computing device N consecutive sets of frequency domain sound level measurements, each set of frequency domain sound level measurements comprising a given number M of sound level amplitudes at the corresponding given number M of frequencies, N and M being integers; and
executing by the processing unit of the computing device a neural network inference engine using the predictive model for inferring one or more output based on inputs, the inputs comprising the N consecutive sets of frequency domain sound level measurements, the one or more output comprising a predicted variation of a CO2 level.

### Claim 25 (depends on 24)

The method of claim 24, wherein determining the N consecutive sets of frequency domain sound level measurements comprises receiving by the processing unit the N consecutive sets of frequency domain sound level measurements via a communication interface of the computing device.

### Claim 26 (depends on 25)

The method of claim 25, wherein the N consecutive sets of frequency domain sound level measurements are received from a sound sensor.

### Claim 27 (depends on 24)

The method of claim 24, wherein determining the N consecutive sets of frequency domain sound level measurements comprises:
receiving by the processing unit a plurality of consecutive time domain sound level measurements via a communication interface of the computing device; and
generating by the processing unit the N consecutive sets of frequency domain sound level measurements based on the plurality of consecutive time domain sound level measurements.

### Claim 28 (depends on 27)

The method of claim 27, wherein the plurality of consecutive time domain sound level measurements is received from a sound sensor.

### Claim 29 (depends on 27)

The method of claim 27, wherein the generation of the N consecutive sets of frequency domain sound level measurements based on the plurality of consecutive time domain sound level measurements uses a Fast Fourier Transform algorithm.

### Claim 30 (depends on 24)

The method of claim 24, further comprising determining by the processing unit N consecutive CO2 level measurements corresponding to the N consecutive sets of frequency domain sound level measurements and the inputs further include the N consecutive CO2 level measurements.

### Claim 31 (depends on 30)

The method of claim 30, wherein the determination of the N consecutive CO2 level measurements is based on data received via a communication interface of the computing device from a CO2 sensor.

### Claim 32 (depends on 24)

The method of claim 24, further comprising determining by the processing unit a CO2 level measurement corresponding to the N consecutive sets of frequency domain sound level measurements and the inputs further include the CO2 level measurement.

### Claim 33 (depends on 32)

The method of claim 32, wherein the determination of the CO2 level measurement is based on data received via a communication interface of the computing device from a CO2 sensor.

### Claim 34 (depends on 24)

The method of claim 24, wherein the one or more output further comprises a predicted variation of temperature.

### Claim 35 (depends on 34)

The method of claim 34, further comprising determining by the processing unit N consecutive temperature measurements corresponding to the N consecutive sets of frequency domain sound level measurements and the inputs further include the N consecutive temperature measurements.

### Claim 36 (depends on 35)

The method of claim 35, wherein the determination of the N consecutive temperature measurements is based on data received via a communication interface of the computing device from a temperature sensor.

### Claim 37 (depends on 34)

The method of claim 34, further comprising determining by the processing unit a temperature measurement corresponding to the N consecutive sets of frequency domain sound level measurements and the inputs further include the temperature measurement.

### Claim 38 (depends on 37)

The method of claim 37, wherein the determination of the temperature measurement is based on data received via a communication interface of the computing device from a temperature sensor.

### Claim 39 (depends on 24)

The method of claim 24, further comprising generating by the processing unit at least one command for controlling at least one controlled appliance and transmitting by the processing unit the at least one command to the at least one controlled appliance via a communication interface of the computing device, the generation of the at least one command being based at least on the predicted variation of the CO2 level.

### Claim 40 (depends on 39)

The method of claim 39, wherein the at least one controlled appliance comprises a Variable Air Volume (VAV) appliance.

### Claim 41 (depends on 24)

The method of claim 24, wherein each sound level amplitude at the corresponding frequency consists of a sound pressure at the given frequency, a sound pressure level at the given frequency or a sound power at the given frequency.

### Claim 42 (depends on 24)

The method of claim 24, wherein the neural network inference engine implements a neural network comprising an input layer, followed by fully connected layers; the input layer comprising N*M neurons respectively receiving the sound level amplitudes; the predictive model comprising weights for the fully connected layers.

### Claim 43 (depends on 24)

The method of claim 24, wherein the neural network inference engine implements a neural network comprising one input layer, followed by at least one one-dimensional convolutional layer, followed by fully connected layers; the input layer comprising M neurons respectively receiving a one-dimension matrix, each one-dimension matrix comprising N sound level amplitudes at a given frequency among the M frequencies, the at least one one-dimensional convolutional layer applying a one-dimensional convolution to each one-dimension matrix; the predictive model comprising weights for the fully connected layers and parameters for the at least one one-dimensional convolutional layer.

### Claim 44 (depends on 43)

The method of claim 43, wherein the neural network further comprises at least one pooling layer.

### Claim 45 (depends on 24)

The method of claim 24, wherein the neural network inference engine implements a neural network comprising an input layer, followed by at least one two-dimensional convolutional layer, followed by fully connected layers; the input layer comprising one neuron receiving a two-dimensions matrix comprising the N*M sound level amplitudes, the at least one two-dimensional convolutional layer applying a two-dimensional convolution to the two-dimensions matrix; the predictive model comprising weights for the fully connected layers and parameters for the at least one two-dimensional convolutional layer.

### Claim 46 (depends on 45)

The method of claim 45, wherein the neural network further comprises at least one pooling layer.

### Claim 47

A non-transitory computer program product comprising instructions executable by a processing unit of a computing device, the execution of the instructions by the processing unit of the computing device providing for predicting carbon dioxide (CO2) level variations based on sound level measurements by:
storing by the processing unit a predictive model in a memory of a computing device;
determining by the processing unit N consecutive sets of frequency domain sound level measurements, each set of frequency domain sound level measurements comprising a given number M of sound level amplitudes at the corresponding given number M of frequencies, N and M being integers; and
executing by the processing unit a neural network inference engine using the predictive model for inferring one or more output based on inputs, the inputs comprising the N consecutive sets of frequency domain sound level measurements, the one or more output comprising a predicted variation of a CO2 level.