GB2623748A

GB2623748A - A method of controlling airflow in a cabin using a neural network and a system of the same

Info

Publication number: GB2623748A
Application number: GB2215486.8A
Authority: GB
Inventors: S Mani Kishore Kurapati V; Dhariwal Shashank; Singh Nahar
Original assignee: Continental Automotive Technologies GmbH
Current assignee: Continental Automotive Technologies GmbH
Priority date: 2022-10-20
Filing date: 2022-10-20
Publication date: 2024-05-01
Also published as: GB202215486D0

Abstract

A method of controlling an airflow system for a cabin is disclosed. The method comprising: measuring, by way of an air quality measurement module, an index of quality of air and generating, by way of a processor, a set of modality parameters stored in a memory of the processor. The method further comprises predicting, by way of a neural network, a task in response to the index of quality of air measured by the air quality measurement module; and the set of modality parameters generated by the processor; and executing, by way of the processor, at least one target task corresponding to an adjustment of a window regulator. A system, a computer program product and a computer-readable medium having a computer program product stored thereon is also disclosed.

Description

A METHOD OF CONTROLLING AIRFLOW IN A CABIN USING A NEURAL NETWORK AND A SYSTEM OF THE SAME

TECHNICAL FIELD

This disclosure relates to an airflow method and using a neural network for controlling the same.

BACKGROUND

Apart from heating, ventilation and air-conditioning (HVAC) systems, windows function to allow ventilation of a cabin, for example a room, a building and even motor vehicles. Allowing natural air to flow into a confined area has many benefits.

One of the most direct influence is that allowing natural air to flow into a building or motor vehicle saves energy compared to switching on a HVAC system.

By way of an example, on a cool weather day, a person may want to enjoy natural flow of air entering a house or motor vehicle. On a personal level, ventilating a cabin through natural airflow such as a motor vehicle benefit people suffering from motion sickness. Yet on other occasions, one may wish to switch on the HVAC system instead, to keep traffic noise out.

As can be seen from the foregoing discussion, it is implicit that opening of windows is personalized according to an individual's preferences and changes in preferences based on above-mentioned factors.

The background description provided herein is for the purposes of generally presenting the context of the disclosure. Work of the presently named inventors, to the extent it is described in this background section, as well as aspects of the description that may not otherwise qualify as prior art at the time of filing, are neither expressly nor impliedly admitted as prior art against the present disclosure.

SUMMARY

A purpose of this disclosure is to ameliorate the problem of personalizing control of a window regulator to regulate airflow of a cabin by providing the subject-matter of the independent claims.

Further purposes of this disclosure are set out in the accompanying dependent claims.

The objective of this disclosure is solved by a method of controlling an airflow system for a cabin, the method comprising: measuring, by way of an air quality measurement module, an index of quality of air; generating, by way of a processor, a set of modality parameters stored in a memory of the processor; characterized by that: the method further comprises: predicting, by way of a neural network, a task in response to: the index of quality of air measured by the air quality measurement module; and the set of modality parameters generated by the processor; and executing, by way of the processor, at least one target task corresponding to 25 an adjustment of a window regulator.

An advantage of the above-described aspect of this disclosure yields a method of controlling an airflow system for a cabin through measurement of an index of quality of air and predicting a targeted task corresponding to an adjustment of a window regulator using a neural network, such that the adjustment of window regulator may be personalised according to a set of modality parameters. More advantageously, the adjustment of window regulator may save energy by allowing airflow to enter the cabin.

Preferred is a method as described above or as described above as being preferred, in which: the at least one target task corresponding to adjustment of a window regulator further comprises: executing, by way of the processor, a command for adjusting an opening of at least one window of the cabin to a pre-determined percentage; and executing, by way of the processor, a command for locking at least one window of the cabin.

The advantage of the above aspect of this disclosure is to yield a controlling an airflow system for a cabin by adjusting a widow regulator by opening or closing at least one window of the cabin to a pre-determined percentage to increase or decrease natural air flow entering the cabin and/or locking at least one of the windows for security reasons.

Preferred is a method as described above or as described above as being preferred, in which: the method further comprising: measuring, by way of the air quality measurement module, an ambient within the cabin; and measuring, by way of the air quality measurement module, an ambient outside of the cabin.

The advantage of the above aspect of this disclosure is to measure a temperature of an air quality index within the cabin and measure a temperature of an air quality index outside of the cabin. More advantageously, the temperature measured may be stored in provided to a neural network, as a type of modality data for predicting at least one corresponding target task.

Preferred is a method as described above or as described above as being preferred, in which: the method further comprising: retrieving, by way of the processor, the set of modality parameters stored in the memory of the processor; comparing, by way of the neural network, a set of occurring parameters received against the set of modality parameters retrieved; and predicting, by way of the neural network, the task in relation to the adjustment of the window regulator The advantage of the above aspect of this disclosure is to yield a sequence of steps for the neural network to compare data from the set of modality parameters stored in memory and comparing the retrieving the set of modality parameters provided to a set of occurring parameters, i.e., real-time measurement of information, to assist the neural network to Preferred is a method as described above or as described above as being preferred, in which: the set of modality parameters comprises: a set of data relating to a preference of a user; a set of data relating to an identification of a user; a set of data relating to a cabin; a set of data relating to at least one historical event, or combination thereof.

The advantage of the above aspect of this disclosure is to utilize modality parameters to train a neural network, such that the information may be stored in memory of a processor and used for predicting a task for adjustment of the window regulator. More advantageously, the set of modality parameters may be selected for personalizing the method of controlling airflow, and for security reasons.

Preferred is a method as described above or as described above as being preferred, in which: the set of data relating to a preference of a user comprises: a parameter in relation to temperature; a parameter in relation to a level of noise tolerance; or combination thereof.

The advantage of the above aspect of this disclosure is to yield a personalized method of controlling airflow by utilizing a parameter of a measurement of temperature and a parameter of a measurement of a level of noise tolerance of a user. The average parameter may be determined based upon a record of a collection of information from historical events that are provided to the neural network.

Preferred is a method as described above or as described above as being preferred, in which: the set of data relating to an identification of a user comprises: at least one biometric feature of a user.

The advantage of the above aspect of this disclosure is to yield a method of controlling an airflow entering a cabin by controlling at least one window based upon an identification of a user, of which the means for identifying the user is at least one biometric feature of the user.

Preferred is a method as described above or as described above as being preferred, in which: the set of data relating to a cabin comprises: a state of a cabin; a location of a cabin, or combination thereof.

The advantage of the above aspect of this disclosure is to provide a set of modality parameters to a neural network, of which the set of modality parameters include a set of data relating to a state of a cabin and/or a location of the cabin. More advantageously, this set of data is applicable to motor vehicles, for example whether the motor vehicle is in a driving state or a parking state, and a location of the motor vehicle if it is in a parking state.

Preferred is a method as described above or as described above as being preferred, in which: the set of occurring parameters received by the neural network in relation to the identification of a user comprises: a state of a cabin; a biometric feature recognition of a user; a video signal streaming from at least one image sensing device, or combination thereof.

The advantage of the above aspect of this disclosure is to transmit a set of occurring parameters, more specifically information relating to events occurring in real time, to a neural network for identifying a user. The information may include a state of the cabin, a biometric feature recognition of a user, and/or a video signal streaming from at least one image sensing device.

Preferred is a method as described above or as described above as being preferred, in which: the set of occurring parameters received by the neural network in relation to the preference of the user comprise: a level of noise within the cabin; an ambient within the cabin; a weather data received by the neural network, or combination thereof.

The advantage of the above aspect of this disclosure is to this disclosure is to transmit a set of occurring parameters, more specifically information relating to events occurring in real time, to a neural network for defining a preference of a user.

The information may include a level of noise within the cabin, a ambient within the cabin and/or a weather data received by the neural network.

The objective of this disclosure is solved by an airflow system comprising: an air quality measurement module operable to measure an index of quality of air; and a processor operable to generate a set of modality parameters; characterized in that: the airflow system further comprises a neural network, the neural network is operable to predict a task in response to: the index of quality of air measured; and the set of modality parameters generated; and in response to the task predicted, the processor is operable to execute at least one target task corresponding to an adjustment of a window regulator.

An advantage of the above-described aspect of this disclosure yields an airflow system for a cabin through measurement of an index of quality of air and predicting a targeted task corresponding to an adjustment of a window regulator using a neural network, such that the adjustment of window regulator may be personalised according to a set of modality parameters.

Preferred is a system as described above or as described above as being preferred, in which: the at least one target task corresponding to the adjustment of a window regulator comprises: a command to adjust an opening of at least one window to a pre-determined percentage; and a command to lock at least one window.

The advantage of the above aspect of this disclosure is to yield an airflow system operable to automate adjustment of at least one window in response to prediction generated by a neural network of an airflow system. The prediction may be according to amongst others, a user's preference. The types of adjustments may include opening the at least one window to a pre-determined percentage, or locking the at least one window.

Preferred is a system as described above or as described above as being preferred, in which: the neural network is a deep-based neural network.

The advantage of the above aspect of this disclosure is to yield an airflow system incorporating a neural network to predict and execute an adjustment of at least one window of a cabin, of which the neural network is a deep neural network.

The objective of this disclosure is solved by a computer program product comprising instructions to cause an airflow system having a neural network as disclosed herein to execute the method of controlling an airflow system as disclosed herein.

An advantage of the above-described aspect of this disclosure yields a computer program product suitable for controlling an airflow system incorporated with a neural network to predict and execute an adjustment of at least one window of a cabin.

The objective of this disclosure is solved by a computer-readable medium having 25 stored thereon a computer program product as disclosed herein.

BRIEF DESCRIPTION OF DRAWINGS

The present disclosure will become more fully understood from the detailed 5 description and the accompanying drawings, wherein: FIG. 1 shows a flowchart of a method of controlling an airflow system in accordance with an aspect of this disclosure.

FIG. 2 shows a flowchart of an airflow system in accordance with an aspect of this

disclosure.

FIG. 3 shows a flowchart for an identification process in accordance with an aspect of this disclosure.

FIG. 4 shows a flowchart for a noise cancellation process in accordance with an aspect of this disclosure.

FIG. 5 shows a system block diagram of an airflow system in accordance with an

aspect of this disclosure.

FIG 6 shows a decision tree of a neural network in accordance with an aspect of this disclosure.

In various aspects described by reference to the above figures, like reference signs refer to like components in several perspective views and/or configurations.

DETAILED DESCRIPTION

It should be understood that like reference numerals identify corresponding or similar elements throughout the several drawings. It should be understood that although a particular component arrangement is disclosed and illustrated in these exemplary embodiments, other arrangements could also benefit from the teachings of this disclosure.

It is the intent of this disclosure to present a method of controlling an airflow system that incorporates a neural network to predict and execute an adjustment of a window regulator to achieve the personalization of airflow entering into a cabin. Further, it is an intent of this disclosure to present a method of controlling an airflow system that incorporates a neural network to predict and execute an adjustment of a window regulator, for energy savings.

Hereinafter, the term "cabin" refers to an area surrounded or closed off on all sides. In some aspect, a "cabin" may refer to a building, a room, an apartment, or a house. In some aspect, a "cabin" may refer to a cabin of a motor vehicle. In all aspects, the "cabin" includes at least one window in connection with a window regulator for adjusting the at least one window.

The term "occurring parameters" used in the context herein refers to a set of measurable or quantifiable numeric or characteristics which are happening in real-time. By way of an example, in aspects of this disclosure, a cabin may refer to a motor vehicle and 'an occurring parameter' related to a motor vehicle may refer to a location of the motor vehicle at the point of time when data is measured.

The term "ambient" used in the context herein refers to an immediate surrounding, especially of environmental conditions. Therefore, the word "ambient" or "ambient condition" use in the context of a cabin or motor vehicle described herein shall include by way of example, other parameters related to environmental conditions including temperature, humidity, arid, speed of heating, ventilation, and conditioning (HVAC) fan, but not limited thereto.

FIG. 1 of the accompanying drawings which shows a flowchart illustrating a method of controlling an airflow system for a cabin in accordance with an aspect.

At step 102, an air quality measurement module measures an index of quality of air. At a next step 104, a processor generates a set of modality parameters stored in a memory of the processor. The processor may be in connection to a trained neural network.

At step 106, the neural network predicts a task in response to the index of quality of air measured by the air quality measurement module at step 102, and the set of modality parameters generated by the neural network at step 104. In some aspect, the air quality index of the ambient within a cabin measured may be stored in memory and generated at a later stage as a set of modality data, at step 106.

At step 108, the processor executes at least one target task corresponding to an adjustment of a window regulator. The at least one target task corresponding to the adjustment of a window regulator includes executing a command for adjusting an opening of the at least one window of the cabin to a predetermined percentage. By way of example, the adjustment of an opening of the window may be 10%, 30%, 50% and so on, as determined by system design specification. Further, the at least one target task corresponding to the adjustment of a window regulator includes a command for locking at least one window of the cabin.

Method for determinino command FIG. 2 shows a flowchart of method 200 for determining a command for adjustment of a window regulator in accordance with a preferred aspects.

At step 202, a processor of an airflow system initiates a query to determine an air quality index of an ambient within a cabin. The air quality index of an ambient within a cabin may be measured by an air quality measurement module at step 204. In some aspect, the air quality index of the ambient within a cabin measured may be stored in memory and generated at a later stage as a set of modality data classified as a historical event, at step 206. The processor of the airflow system is in communication with a neural network. The neural network may be a deep neural network.

At step 208, the processor determines an air quality index of a cabin. If the air quality index of an ambient of the cabin is determined to be good or excellent. The processor may execute at least one target task corresponding to an adjustment of a window regulator, to disable the airflow system at step 210 At step 212, the processor determines an air quality index outside of a cabin. If the air quality index of the ambient of the cabin is determined to be low or hazardous, the processor execute at least one target task corresponding to an adjustment of the window regulator, to close at least one window of the cabin, at step 214. The term "hazardous" may refer to a state considered to be risky or dangerous. Henceforth, in the context used herein, the term "hazardous" refers to an unhealthy level of air quality which may be deemed detrimental to a person's health.

At step 216, weather data based on global positioning system (GPS) is transmitted over-the-air. For example, the weather data may be retrievable from cloud and transmitted over-the-air to the cabin.

At step 218, the processor may initiate a query to determine an air quality index within the ambient of a cabin to determine whether the airflow system shall remain disabled, or whether the at least one window of the cabin shall remain closed. In the event the processor determines the air quality index outside the cabin is good or excellent, in particular it falls under a hazardous level, the processor execute at least one target task corresponding to an adjustment of a window regulator, to open at least one window of the cabin at step 220.

User Identification Process Referring now to FIG. 3 which shows a flowchart 300 illustrating a method of identifying and authenticating a user, in accordance with a preferred aspects.

At step 302, a set of data relating to a cabin is transmitted to the processor of the airflow system. The set of data includes a state of the cabin and a location of the cabin. An example of a state of the cabin may be whether the cabin is on the move. In some aspect where the cabin is a motor vehicle, a state of the cabin may refer to whether an engine of the motor vehicle is on or off. The location of a cabin may refer to a place, location or vicinity where the motor vehicle is parked.

At step 304, the processor of the airflow system initiates a query to verify a state of the cabin, to determine whether at least one of the access of the cabin is in a locked or unlock status. In response to the processor determining the state of the cabin is unlocked, in a next step at 306, the processor executes a command to a driver's monitoring system or a cabin monitoring system of the cabin to capture one or more images of a cabin. The images captured are transmitted to the processor, to analyse and determine whether there is at least one occupant in the cabin. In response to a determination there is at least one occupant present within the cabin, the processor determines at step 310, whether the at least one occupant present in the cabin is an authorized person. The verification of whether the at least one occupant is an authorized occupant may include an additional verification process. By way of an example, images of authorized occupant(s) may be stored and retrieved from a memory of the processor to determine whether the at least one occupant identified in the cabin is an authorized occupant.

In the event that the at least one occupant within the cabin is determined to be an 20 unauthorized person, the processor executes a command to the at least one window regulator to lock the cabin and issue a notification to alert an owner of the cabin, at step 310.

In another aspects, at step 304, the processor of the airflow system initiates a query to verify a state of the cabin, to determine whether at least one of the access of the cabin is in a locked or unlock status. A user attempting to access the cabin from outside of the cabin is authenticated as an authorized occupant and allowed access at step 312. At step 306, the processor executes a command to the driver monitoring system or cabin monitoring system to capture images within the cabin.

The images captured are transmitted to the processor to be analyzed and determine whether the at least one occupant in the cabin is an authorized user.

In the event the processor determines the at least one occupant in the cabin is an authorized user, the processor retrieves a set of data relating to a preference of a user at step 314. The set of data relating to a preference of a user may include a parameter in relation to an ambient condition within the cabin, for example a temperature, and a parameter in relation to a level of noise tolerance. The parameter in relation to temperature may be an average of a measurement of temperature based on historical events. For example, the parameter may be an average of an operating temperature of a heating, ventilation and air conditioning (HVAC) system of the cabin over a period of time. The parameter in relation to an ambient condition may further include historical events obtained in relation to temperature measurement of ambient outside of the cabin. The parameter may further include weather data obtained through GPS, transmitted to the airflow system. The parameter in relation to a level of noise tolerance may include historical events of date and time of the user opening at least one window of the cabin. This may be achieved by enabling noise cancelation based on user preferences, location information, drive duration, terrain etc. In this manner, cabin having independent music zone/ music bubble features can be advanced with natural airflow personalization system. The system enables the noise cancellation for the noise created with window open state. Once the user profile or use preference is determined, at step 314, the processor of the airflow system executes airflow command for adjusting an opening of at least one window of the cabin to a pre-determined percentage, which corresponds to user's preference.

In all of the above aspects, the airflow system applies an artificial intelligence (Al) approach, which includes both statistical machine learning (ML) approaches.

Suitable types of approaches include decision trees, ensemble models, k-nearest neighbour models, Bayesian networks and recent deep neural networks to cover multiple aspects of proposed prediction task with regard to the set of data in relation to use preference.

The airflow system monitors the air quality of the cabin and ride duration, when the external air quality is low/ hazardous, the airflow system disables natural airflow by closing the windows. The airflow system logs the user preferences and manually overrides the natural airflow settings. The logs are analyzed to improve the personalization, and the airflow system updates the user profile when manual overrides increase. The airflow system also tracks the noise levels created when the at least one window is in an open state.

Noise Cancellation Process Referring now to FIG. 4 which shows a flowchart 400 illustrating a method of noise cancellation in accordance with a preferred aspects.

At step 402, the processor executes at least one target task corresponding to an adjustment of a window regulator, such that a command for adjusting an opening of at least one window of the cabin to a pre-determined percentage. In some aspect, the step 402 may be a manual adjustment done by a user or occupant.

At step 404, a noise level of a cabin is monitored This monitoring of noise level may be achieve by a microphone array installed within a cabin.

In a next step 406, the processor executes a command for noise cancellation in 20 accordance with a user's preference. The user's preference may include a set of modality parameters comprising a set of data relating to a preference of a user in relation to noise level outside of the cabin.

At step 408, the processor executes a command for adjusting the window regulator 25 such that the at least one window of the cabin opens to a personalized percentage, in accordance with a user's preference.

At step 410, a noise level within the cabin is measured. The processor determines whether the noise level is acceptable by the user. In an aspect of this disclosure, the noise level measured at step 410 is an acceptable level in accordance with user's preference. The noise cancellation process thus remains at step 404, for monitoring a noise level within the cabin. In another aspect of this disclosure, the noise level measured at step 410 is at an unacceptable level in accordance with user's preference. In a next step 412, the processor executes a noise cancellation action. The noise cancellation action may be an activation of a vent, for example an air vent, for noise cancellation.

Airflow System FIG. 5 shows a system block diagram of an airflow system 500 in accordance with an aspect of this disclosure.

The airflow system 500 may include a processor 502 having a neural network 504.

The processor 502 is in communication with an air quality measurement module 506, a window regulator 508, a noise monitoring device 510, an air quality index sensor 512, a means for receiving weather data through Global Positioning System (GPS) 514 and a means for providing human-machine interface (HMI) data 516. The neural network 504 enables predication of airflow personalization feature as discussed herein. Suitable types of neural network 504 includes random forest and gradian boost.

For clarity and brevity, the following technical papers sufficiently describes the respective types of tree predictors and neural network which may be suitable for implementing an airflow system as disclosed, without departing from the scope of

inventive concept of this disclosure:

* Random Forest written by Leo Breiman, published by SpringerLink, October 2001 * Random Forest written by Leo Breiman, Statistics Department of University of California dated January 2001 * Gradient Boosting for classificatio published on sklearn.ensemble.GradientBoostingClassifier documentation scikit-learn 1.1.2 * Gradient Boosting for regression published on sklearn.ensemble.GradientBoostingRegressor documentation scikit-learn 1.1.2 The airflow system 500 further includes an air quality measurement module 506 operable to measure an index of air quality 512. In an aspect, the index of air quality 512 may comprises an index of air quality 512 of an ambient within a cabin. In another aspect, the index of air quality 512 may comprises an index of air quality 512 of an ambient outside of a cabin.

The airflow system 500 further includes a window regulator 508 operable to adjust at least a window of a vehicle. In an aspect, the adjustment of a window regulator 508 is executed by at least one a command from the processor 502, for adjusting an opening of at least one window of the cabin to a predetermined percentage. This The airflow system 500 further includes a noise monitoring device 510 for monitoring noise level. In an aspect, the noise monitoring device 510 monitors a noise level within a cabin. In another aspect, the noise monitoring device 510 monitors a noise level from outside the cabin. A suitable type of noise monitoring device 510 may be a microphone.

The processor 502 is operable to generate a set of modality parameters stored in memory, of which the neural network is operable to predict a task in response to an index of quality of air 512 measured by the air quality measurement module 506. In an aspect, at least one target task corresponding to adjustment of a window regulator 508 comprises a command for a command for adjusting an opening of at least one window of the cabin to a pre-determined percentage. In an aspect, at least one target task corresponding to adjustment of a window regulator 508 comprises a command for locking at least one window of the cabin.

The at least one target task corresponding to adjustment of a window regulator 508 may be determined by the processor 502, in response to comparing a set of modality data stored in memory of the processor 502 against a set of occurring 30 parameters received by the airflow system 500.

The set of modality parameters may include a set of data relating to a preference of a user; a set of data relating to an identification of a user; a set of data relating to a cabin; and/or a set of data relating to at least one historical event. The aforesaid set of data relating to at least one historical event assist the processor 502 to predict whether an adjustment of the window regulator 508 is required by a user.

The set of data relating to a preference of a user may comprises: a parameter in relation to temperature; and/or a parameter in relation to a level of noise tolerance. The aforesaid set of data may be used to determine an average threshold or tolerance of a user's sensitivity to temperature changes and/or noise level changes to assist the processor 502 to predict whether an adjustment of the window regulator 508 is required by a user.

The set of data relating to an identification of a user may comprises at least one biometric feature of a user. Examples of biometric feature of a user may include an image of an iris, an image of a face of the use or fingerprint of a user. The set of data relating a cabin may comprises a state of a cabin and/or location of a cabin.

The set of relating to a cabin may refer to a state of a cabin and/or a location of a cabin. In an aspect, in the context of a motor vehicle, a state of a cabin may refer to whether engine of motor vehicle is turn ON or turn OFF, or alternatively whether the motor vehicle is on the move or stationary. In the same context of a motor vehicle, the location of the cabin may refer to a location where the motor vehicle is travelling or parked.

The set of occurring parameters received by the neural network 504 in relation to the identification of a user may include a state of a cabin, a biometric of feature recognition of a user; and/or a video signal streaming from at least one image sensing device.

The set of occurring parameters received by the neural network 504 in relation to the preference of the user may comprises a level of noise within the cabin; an ambient within the cabin and/or a weather data received by the neural network 504.

In an aspect, the airflow system 500 captures a state of the airflow within a cabin and a state of the at least one window interactions from the user and considers it as feedback for estimating the model performance drift. An automatic model update with data collected from user interactions may be implemented to tune the model performance. Thermal comfort is possible with multiple combinations of parameters, instead of a singular combination. Imagine the same graph for 32 features (32-dimensional space).

The system 500 captures the user interactions as feedback and fine-tunes the neural network model 504 based on model drift analysis. The sensitivity of the user towards thermal changes is a very personal parameter that cannot be quantified, but the neural network model 504 is trained to learn the preferences of a user.

FIG. 6 shows an exemplary decision tree 600 for classification of a user's preference related parameters for predicting personalization features of an airflow system 500 as disclosed herein.

Like gradient boosting, random forest is also an algorithm for both classification and regression problems in neural networks. The two types of algorithms may be explained using random forests, which is represented by a bunch of decision trees.

In this exemplary aspect, an input 602 in relation to a preference of a user is provided to the decision tree 600. As shown in FIG. 6, the decision tree may include a first tree Ti. a second tree T2, up to nth number of decision tree, TN and the input 602 may be split to different trees, for example a first tree Ti, a second tree 12, to execute a prediction based upon a user's parameter in relation to temperature and a user's parameter in relation to a level of noise tolerance. Each node of the first tree Ti and the second tree T2 are asked questions in relation to the user's parameter and the questions get more specific as the tree gets to a second level, thus executing a prediction. Pi out the first tree Ti and a prediction, P2 out the second tree T2. The process increases the predictiveness of the neural network 504, by classifying the set of modality parameters.

In an aspect, a random forest type of neural network is used. The estimators may be 300, with zero (0) random state.

In an aspect, a gradient boost type of neural network is used. The estimators may be 300, with one (1) random state and a maximum dept of one (1).

Once the prediction, Pl, the prediction, P2 until prediction, PN have been executed, an average of all predictions, Zp is obtained. A final predic,tion, PF is concluded to predict a command for personalization of the airflow system 500 according to user's preference.

Thus, it can be seen that an airflow system and a method of controlling an airflow system for a cabin having the advantages of adjusting a window regulator of a cabin using a set of modality parameters has been provided. More advantageously, the airflow system is operable to adjust at least one window of the cabin according to a user's preference. More advantageously, the airflow system includes a user authentication process to allow access to the cabin and execute adjustment of a window regulator according to the user identified. While exemplary aspects have been presented in the foregoing detailed description of the disclosure, it should be appreciated that a vast number of variation exist.

The foregoing description shall be interpreted as illustrative and not be limited thereto. One of ordinary skill in the art would understand that certain modifications may come within the scope of this disclosure.

List of Reference Signs Flowchart (airflow control) 102 Measuring an index of quality of air 104 Generating a set of modality parameters by a processor 106 Predicting a task in response to input data from step 102, step 104 108 Executing at least one target task corresponding to an adjustment of a window regulator Flowchart (airflow system) 202 Processor with neural network / decision 204 Air quality measurement module 206 Classification of a historical event 208 Determines air quality index in cabin 210 Disabling of airflow system 212 Determines air quality index outside cabin 214 Adjustment of window regulator -close window 216 Receiving weather data OTA from GPS 218 Determine air quality index within ambient of cabin 220 Adjustment of window regulator -open window 300 Flowchart (user identification) 302 Set of data relating to a cabin 304 Verify a state of cabin 306 Executes command for driver monitoring / cabin monitoring to capture images 308 Determine whether at least one occupant detected is an authorized user 310 Lock and alert owner 312 Authentication of user identification 314 Executes airflow command to adjust opening of at least one window to a pre-determined percentage 400 Flowchart (noise cancellation) 402 Adjustment of window regulator 404 Monitoring a noise level of a cabin 406 Noise cancellation according to user's preference 408 Adjustment of window regulator 410 Monitoring a noise level of a cabin 412 Noise cancellation action 500 Airflow system 502 Processor 504 Neural network 506 Air quality measurement module 508 Window regulator 510 Noise monitoring device 512 Air quality index 514 Weather data 516 HMI data

Claims

Patent claims 1 A method (100) of controlling an airflow system for a cabin, the method comprising: measuring (102), by way of an air quality measurement module, an index of quality of air; generating (104), by way of a processor, a set of modality parameters stored in a memory of the processor; characterized by that: the method (100) further comprises: predicting (106), by way of a neural network, a task in response to: the index of quality of air measured by the air quality measurement module; and the set of modality parameters generated by the processor; and executing (108), by way of the processor, at least one target task 20 corresponding to an adjustment of a window regulator.
2 The method (100) according to claim 1, characterized by that the at least one target task corresponding to adjustment of a window regulator further comprises: executing, by way of the processor, a command for adjusting an opening of at least one window of the cabin to a pre-determined percentage; and executing, by way of the processor, a command for locking at least one window of the cabin.
3. The method according to claims 1-2, characterized by that the method further comprising: measuring, by way of the air quality measurement module, an ambient within the cabin; and measuring, by way of the air quality measurement module, an ambient outside of the cabin.
4 The method according to claims 1-3, characterized by that the method further comprising: retrieving, by way of the processor, the set of modality parameters stored in the memory of the processor; comparing, by way of the neural network, a set of occurring parameters received against the set of modality parameters retrieved; and predicting, by way of the neural network, the task in relation to the adjustment of the window regulator.
The method according to claims 1 -4, characterized by that the set of modality parameters comprises: a set of data relating to a preference of a user; a set of data relating to an identification of a user; a set of data relating to a cabin; a set of data relating to at least one historical event, or combination thereof.
6 The method according to claims 1 -5, characterized by that the set of data relating to a preference of a user comprises: a parameter in relation to temperature; a parameter in relation to a level of noise tolerance; or combination thereof.
7. The method according to claims 1 -5, characterized by that the set of data relating to an identification of a user comprises: at least one biometric feature of a user.
8 The method according to claims 1 -5, characterized by that the set of data relating to a cabin comprises: a state of a cabin; a location of a cabin, or combination thereof.
9 The method according to claim 4, characterized by that the set of occurring parameters received by the neural network in relation to the identification of a user comprises: a state of a cabin; a biometric feature recognition of a user; a video signal streaming from at least one image sensing device, or combination thereof.
10 The method according to claim 4, characterized by that the set of occurring parameters received by the neural network in relation to the preference of the user comprise: a level of noise within the cabin; an ambient within the cabin; a weather data received by the neural network, or combination thereof.
11.An airflow system comprising: an air quality measurement module operable to measure an index of quality of air; and a processor operable to generate a set of modality parameters; characterized in that: the airflow system further comprises a neural network, the neural network is operable to predict a task in response to: the index of quality of air measured; and the set of modality parameters generated; and in response to the task predicted, the processor is operable to execute at least one target task corresponding to an adjustment of a window regulator.
12. The system according to claim 11, characterized in that the at least one target task corresponding to the adjustment of a window regulator comprises: a command to adjust an opening of at least one window to a pre-determined percentage; and a command to lock at least one window.
13. The system according to claims 11 -12, characterized in that the neural network is a deep-based neural network.
14.A computer program product comprising instructions to cause the system of claims 11 to 13 to execute the steps of 1 to 10.
15.A computer-readable medium having stored thereon the computer program product of claim 14.