Intelligent air conditioner remote controller

Abstract

The application discloses an intelligent air conditioner remote controller, wherein the radar sensor assembly and the infrared sensor assembly respectively contain multiple radar sensors and infrared sensors in multiple directions; the main control MCU continuously executes a detection process, including: at each time, extracting heartbeat and breathing information from the reflection signals of the radar sensors in each detection area, extracting voltage information from the voltage signals of the infrared sensors, obtaining corresponding samples, determining the sample categories based on the threshold values of each information, and determining the personnel actual measurement results according to the sample categories of all detection areas at this time; training a deep learning model based on the samples at multiple times and obtaining the model prediction results at the next time, and adjusting the threshold values for the next detection process when the model prediction results are inconsistent with the corresponding personnel actual measurement results; the air conditioner control module controls the working state of the air conditioner according to the personnel actual measurement results; and the display screen displays the personnel actual measurement results and the air conditioner state information. The application can accurately detect whether there is a person in the area and automatically control the state of the air conditioner.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of intelligent remote controller, and particularly relates to an intelligent air conditioner remote controller. BACKGROUND

[0002] Air conditioner is an indispensable device in people's work and life. When using air conditioner, the air conditioner will still run empty due to human factors, which will cause a lot of power waste. Therefore, how to accurately detect the presence or absence of human body and intelligently control the opening or closing of air conditioner is a technical problem to be solved. SUMMARY

[0003] In order to solve the above problems existing in the prior art, the present application provides an intelligent air conditioner remote controller. The technical problem to be solved by the present application is realized by the following technical scheme:

[0004] An intelligent air conditioner remote controller, comprising: a radar sensor assembly, an infrared sensor assembly, a main control MCU, a display screen and an air conditioner control module; wherein,

[0005] The radar sensor assembly comprises radar sensors arranged at multiple directions of the intelligent air conditioner remote controller; any radar sensor is used for emitting radio waves to a corresponding detection area, receiving a reflection signal at each moment and sending it to the main control MCU;

[0006] The infrared sensor assembly comprises infrared sensors arranged at multiple directions of the intelligent air conditioner remote controller; any infrared sensor is used for detecting human body infrared radiation information in a corresponding detection area and outputting a voltage signal at each moment to the main control MCU;

[0007] The main control MCU is used for continuously executing a detection process, the detection process comprising: at each moment, extracting heartbeat information and breathing information from the reflection signal of each detection area, extracting voltage information from the voltage signal, obtaining a sample of the corresponding detection area, and determining the sample category based on the threshold value corresponding to each information to represent whether there is a person in the corresponding detection area, and determining the personnel measured result of the working area where the intelligent air conditioner remote controller is located according to the sample category of all detection areas at this moment; and based on all samples at multiple moments accumulated, training a deep learning model, using the trained deep learning model to obtain the model prediction result of whether there is a person in the working area at the next moment of the multiple moments, and when the model prediction result is inconsistent with the corresponding personnel measured result, adjusting the threshold value corresponding to each information for the next detection process;

[0008] The air conditioner control module is used for controlling the working state of the air conditioner according to the personnel measured result;

[0009] The display screen is used for displaying the personnel measured result and air conditioner state information.

[0010] In one embodiment of the present application, the plurality of orientations includes front, back, left and right orientations of the intelligent air conditioner remote controller.

[0011] In one embodiment of the present application, for any moment, heartbeat information and breathing information are extracted from the reflection signals of each detection area, including:

[0012] At this moment, for each detection area corresponding to an orientation, the reflection signals from the radar sensor of the detection area are preprocessed, and the heartbeat information and breathing information are extracted from the preprocessed signals.

[0013] In one embodiment of the present application, the preprocessing of the reflection signals from the radar sensor of the detection area includes:

[0014] The reflection signals from the radar sensor of the detection area are sequentially subjected to mean filtering denoising processing, linear interpolation processing and time-frequency conversion processing to obtain preprocessed signals.

[0015] In one embodiment of the present application, the extraction of the heartbeat information and the breathing information from the preprocessed signals includes:

[0016] In the frequency range of the preprocessed signals, the frequency corresponding to the peak point in the preset heartbeat frequency range is determined as the heartbeat information, and the frequency corresponding to the peak point in the preset breathing frequency range is determined as the breathing information.

[0017] In one embodiment of the present application, at each moment, the sample of any detection area includes the heartbeat information, the breathing information and the voltage information extracted from the detection area.

[0018] The determination of the sample category based on the threshold value corresponding to each information to represent whether there is a person in the corresponding detection area includes:

[0019] At this moment, for each detection area, it is judged whether the sample of the detection area satisfies: the corresponding heartbeat information is greater than the heartbeat threshold value, the corresponding breathing information is greater than the breathing threshold value, and the corresponding voltage information is greater than the voltage threshold value; if so, the sample category of the detection area is determined as a person; otherwise, the sample category of the detection area is determined as no person.

[0020] In one embodiment of the present application, the determination of the personnel measured result of the working area where the intelligent air conditioner remote controller is located according to the sample categories of all detection areas at this moment includes:

[0021] If the sample category of a detection area at the moment is a person, it is determined that the category result of the working area where the intelligent air conditioner remote controller is located at the moment is a person, otherwise, it is determined that the category result of the working area at the moment is no person.

[0022] If all category results within the preset time period up to the moment are a person, it is determined that the personnel measured result of the working area is a person, otherwise, it is determined that the personnel measured result of the working area is no person.

[0023] In an embodiment of the present application, the deep learning model is trained based on all samples of multiple moments, and the model prediction result of whether the working area has a person includes:

[0024] The samples of all detection areas at multiple moments are obtained by random sampling and returning, and a deep learning model is trained by using the multiple training sets and the category labels of each sample.

[0025] Input data of the next moment of the multiple moments is obtained, and the trained deep learning model is inputted to obtain a model prediction result, which is used to represent whether the working area where the intelligent air conditioner remote controller is located has a person.

[0026] In an embodiment of the present application, the air conditioner control module controls the working state of the air conditioner according to the personnel measured result, which includes:

[0027] If the personnel measured result received by the air conditioner control module within the first preset time period is that the working area has a person, a first type of binary instruction code is generated, and the first type of binary instruction code is modulated and sent to the air conditioner by using the infrared communication module built in the intelligent air conditioner remote controller to control the air conditioner to start.

[0028] If the personnel measured result received by the air conditioner control module within the second preset time period is that the working area has no person, a second type of binary instruction code is generated, and the second type of binary instruction code is modulated and sent to the air conditioner by using the infrared communication module to control the air conditioner to stop.

[0029] In an embodiment of the present application, the air conditioner state information includes working parameters such as switch state, working mode, temperature, wind power level, wind speed, wind direction, and the power consumption saved by the intelligent air conditioner remote controller for the air conditioner.

[0030] The beneficial effects of the present application are as follows:

[0031] The present application provides a deep learning-based intelligent air conditioner remote controller, which embeds radar sensors and infrared sensors in the remote controller, combines the detection of human presence to control the air conditioner, and the combination of the radar sensors and the infrared sensors can improve the accuracy of personnel detection, thereby improving the accuracy of air conditioner working state control. At the same time, the present application trains a random forest model to predict the human presence at the next moment, compares it with the actual detection result to dynamically adjust the threshold for judgment, and can improve the recognition rate of human presence. Compared with using radar sensors alone, the present application can reduce misjudgment; by automatically detecting the personnel activity in the room, the automatic on-off control of the air conditioner is realized, which can avoid energy waste and improve user experience; and the degree of automation is high, the user does not need to operate frequently, which can improve the energy use efficiency, reduce unnecessary power consumption, and the installation and use are convenient, which is compatible with existing air conditioner equipment and has practical value. BRIEF DESCRIPTION OF DRAWINGS

[0032] Figure 1 A structure diagram of the intelligent air conditioner remote controller provided by the embodiment of the present application is shown in the figure.

[0033] Figure 2 A working process diagram of the intelligent air conditioner remote controller provided by the embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0034] The present application will be further described in detail below in combination with specific embodiments, but the embodiments of the present application are not limited thereto.

[0035] The embodiment of the present application provides an intelligent air conditioner remote controller, as shown in the figure, which can include: Figure 1

[0036] It includes: a radar sensor assembly, an infrared sensor assembly, a main control MCU, a display screen and an air conditioner control module; wherein,

[0037] The radar sensor assembly includes radar sensors arranged at multiple orientations of the intelligent air conditioner remote controller; any radar sensor is used to emit radio waves to the corresponding detection area, receive the reflection signal at each moment and send it to the main control MCU;

[0038] The infrared sensor assembly includes infrared sensors arranged at multiple orientations of the intelligent air conditioner remote controller; any infrared sensor is used to detect the human infrared radiation information in the corresponding detection area and output the voltage signal at each moment to the main control MCU;

[0039] ​The master MCU is configured to continuously perform a detection process, which includes: at each time point, extracting heartbeat information and breathing information from the reflection signals of each detection area, extracting voltage information from the voltage signal, obtaining samples corresponding to the detection area, and determining the sample category based on the threshold corresponding to each information to represent whether there is a person in the detection area, and determining the personnel measurement result of the working area of the intelligent air conditioner remote controller based on the sample categories of all detection areas at this time point; and based on all samples at multiple time points, training a deep learning model, using the trained deep learning model to obtain the model prediction result of whether there is a person in the working area at the next time point of the multiple time points, and adjusting the threshold corresponding to each information for the next detection process when the model prediction result is inconsistent with the corresponding personnel measurement result.

[0040] The air conditioner control module is configured to control the working state of the air conditioner according to the personnel measurement result.

[0041] The display screen is configured to display the personnel measurement result and air conditioner state information.

[0042] The intelligent air conditioner remote controller of the embodiment of the present application can be adapted to any existing air conditioner, and can be another intelligent remote controller independent of the original remote controller of the air conditioner, which is used to detect whether there is a person in the working area of the air conditioner in real time, so as to control the working state of the air conditioner, such as controlling the opening or closing of the air conditioner.

[0043] Further, the intelligent air conditioner remote controller of the embodiment of the present application can also be realized by improving the original remote controller of the air conditioner, such as by integrating additional modules and corresponding configuration, so that the original remote controller of the air conditioner becomes the intelligent air conditioner remote controller. In this case, the original functions of the original remote controller of the air conditioner remain unchanged, wherein the radar sensor assembly and the infrared sensor assembly are additional integrated modules, and the master MCU, the display screen and the air conditioner control module can be realized by function upgrading through program writing configuration and the like on the corresponding components of the original remote controller of the air conditioner, and the specific process is not described here. Of course, the master MCU and the air conditioner control module can also be additional integrated modules, which are all reasonable. For the original functions of the original remote controller of the air conditioner, please refer to the relevant technical understanding, which is not described below.

[0044] In order to facilitate understanding of the scheme, the following describes each part of the intelligent air conditioner remote controller in the embodiment of the present application.

[0045] Radar sensor assembly:

[0046] In the embodiment of the present application, a radar sensor is arranged at each of multiple directions of the intelligent air conditioner remote controller, which together constitute a radar sensor assembly. Each direction corresponds to a detection area.

[0047] The number of the plurality of orientations can be set according to the requirement of detection accuracy, for example, in an optional implementation, the plurality of orientations include front, back, left and right four orientations of the intelligent air conditioner remote controller. Then, the radar sensor located at the front orientation can be at the midpoint position of the front side edge of the intelligent air conditioner remote controller, the radar sensor located at the back orientation can be at the midpoint position of the back side edge of the intelligent air conditioner remote controller, the radar sensor located at the left orientation can be at the midpoint position of the left side edge of the intelligent air conditioner remote controller, and the radar sensor located at the right orientation can be at the midpoint position of the right side edge of the intelligent air conditioner remote controller.

[0048] Each radar sensor is used for detecting the personnel condition in the corresponding detection area. Specifically, the radar sensor can adopt a 24GHz radar sensor, each radar sensor can cover a certain range of detection area, and the radar sensor continuously emits radio waves to the corresponding detection area, receives the reflection signal at each moment and sends it to the master MCU.

[0049] Wherein, the radar sensor emits radio waves with a frequency of f t . When the radio waves meet the moving target (such as the chest or heart of the human body), the frequency of the reflected wave will change due to the movement of the target. This frequency change Δf (Doppler shift) is related to the speed v of the target, which can be expressed as:

[0050]

[0051] Wherein, f t is the frequency of the radar sensor emission signal; v is the relative speed of the target, that is, the fluctuation speed of the chest caused by breathing and heartbeat; c is the speed of light.

[0052] The reflection signal received by the radar sensor at each moment can be represented as a time domain signal, which is a continuous signal for a period of time, and is represented as:

[0053] s(t)=Acos(2*π*(f t +Δf)t+φ);

[0054] Wherein, A is the amplitude; φ is the initial phase.

[0055] Infrared sensor assembly:

[0056] In the embodiment of the application, one infrared sensor is arranged in each of the plurality of orientations corresponding to the radar sensors, to form an infrared sensor assembly. Each orientation corresponds to a detection area.

[0057] In the embodiment of the application, for each orientation, the positions of the radar sensor and the infrared sensor can be reasonably set to ensure that both of them can achieve good detection on the same detection area.

[0058] The infrared sensor can adopt a pyroelectric infrared sensor to detect the presence of a human body by detecting infrared radiation emitted by the human body. The surface temperature of a human body is usually around 37°C, which emits infrared radiation with a peak wavelength of about 10 microns. When a person enters the detection area of the infrared sensor, the infrared radiation of the human body causes a temperature change in the pyroelectric material inside the infrared sensor, thereby generating an electric charge. The pyroelectric formula is as follows:

[0059] Q = p * S * ΔT;

[0060] Where Q is the generated electric charge; p is the pyroelectric coefficient (related to the material); S is the effective area of the infrared sensor; and ΔT is the temperature change. The electric charge is converted into a voltage signal by a circuit and transmitted to the host MCU for processing.

[0061] Host MCU:

[0062] The host MCU continuously performs detection processes, and the following describes any detection process.

[0063] At any given time, heartbeat information and breathing information are extracted from the reflection signals of each detection area, including:

[0064] At this time, for each detection area corresponding to a direction, the reflection signals from the radar sensor of the detection area are preprocessed, and heartbeat information and breathing information are extracted from the preprocessed signals.

[0065] In an optional implementation, the preprocessing of the reflection signals from the radar sensor of the detection area includes:

[0066] The reflection signals from the radar sensor of the detection area are sequentially subjected to mean filter denoising processing, linear interpolation processing, and time-frequency conversion processing to obtain preprocessed signals.

[0067] Specifically, mean filter denoising processing smoothes data by calculating the average value of all points in the neighborhood through mean filter, thereby reducing the influence of noise.

[0068] The reflection signals from the radar sensor are s(t), and the data after mean filter denoising processing is y(t). The window size is 2*k+1, then:

[0069]

[0070] Where s(t) represents the reflection signal received at time t. Mean filter denoising processing is described in related technical understanding and will not be described in detail here.

[0071] Then a linear interpolation method is used to insert a data point between two adjacent points, and the formula is as follows:

[0072]

[0073] Wherein, y'(t) is the value corresponding to the interpolation point, t is the independent variable value of the interpolation point, t1 and t2 are the independent variable values of the adjacent two points.

[0074] Then the interpolation point set is combined with the original data set into a data set z(t). Through interpolation, the data quantity can be expanded, thereby increasing the training set scale of subsequent model training, which is beneficial to improve the training accuracy.

[0075] For z(t), the time-frequency conversion processing is Fourier transform, so that the time domain signal z(t) is converted into a frequency domain signal Z(f), which is expressed as:

[0076] Z(f) = F{z(t)};

[0077] Wherein, F{} represents Fourier transform.

[0078] In an optional embodiment, the heartbeat information and the breathing information are extracted from the preprocessed signal, including:

[0079] In the frequency range of the preprocessed signal, the frequency corresponding to the peak point in the preset heartbeat frequency range is determined as the heartbeat information, and the frequency corresponding to the peak point in the preset breathing frequency range is determined as the breathing information.

[0080] Specifically, Z(f) is the frequency domain representation of z(t) at each time, the heartbeat frequency is usually 1-2Hz, and this range is the preset heartbeat frequency range. The breathing frequency is usually 0.1-0.5Hz, and this range is the preset breathing frequency range.

[0081] The peak value in the preset heartbeat frequency range in the frequency range of the preprocessed signal can be found to obtain the heartbeat information, and the peak value in the preset breathing frequency range can be found to obtain the breathing information.

[0082] For the infrared signal, mainly consider the voltage change of the infrared sensor, the voltage and temperature change relationship is as follows:

[0083] V(t) = δ * ΔT(t);

[0084] Wherein, V(t) is the voltage information extracted from the voltage signal at time t, δ is the sensitivity coefficient of the infrared sensor, and ΔT(t) is the temperature change at time t. When the human body exists, the temperature change ΔT(t) will produce significant voltage fluctuation.

[0085] In the embodiment of the present application, at each moment, the sample of any detection area includes the heartbeat information, the breathing information and the voltage information extracted from the detection area. Therefore, it can be understood that for the four orientations of the above example, at each moment, each orientation obtains a sample, and a total of four samples are obtained.

[0086] For the master MCU, the threshold value corresponding to the extracted information is used to determine the sample category to represent whether there is a person in the corresponding detection area, including:

[0087] At this moment, for each detection area, it is determined whether the sample of the detection area satisfies: the corresponding heartbeat information is greater than the heartbeat threshold value, the corresponding breathing information is greater than the breathing threshold value, and the corresponding voltage information is greater than the voltage threshold value; if it is satisfied, it is determined that the sample category of the detection area is a person; otherwise, it is determined that the sample category of the detection area is no person.

[0088] The heartbeat threshold value, the breathing threshold value and the voltage threshold value are determined by a large amount of data and pre-experiment. The heartbeat threshold value can be represented by H th , the breathing threshold value can be represented by B th , and the voltage threshold value can be represented by V th . Only when the heartbeat information, the breathing information and the voltage information corresponding to the detection area are all greater than the corresponding threshold value, it is determined that there is a person in the detection area, and then the sample category of the detection area is determined to be a person.

[0089] For each sample, the category of each information in the sample is the sample category after the sample category is determined.

[0090] As described above, at each moment, each detection area can obtain a sample and determine the sample category. Therefore, for the moment, the personnel measurement result of the working area of the intelligent air conditioner remote controller can be determined according to the sample categories of all detection areas at the moment, and the process includes:

[0091] If the sample category of a detection area at the moment is a person, it is determined that the category result of the working area of the intelligent air conditioner remote controller at the moment is a person, otherwise, it is determined that the category result of the working area at the moment is no person.

[0092] If all category results within the preset time period up to the moment are a person, it is determined that the personnel measurement result of the working area is a person, otherwise, it is determined that the personnel measurement result of the working area is no person.

[0093] Therefore, through the above determination, the personnel measurement result of the working area can be obtained at each moment, and the master MCU will send it to the air conditioner control module and the display screen.

[0094] Because each time the host MCU will obtain the samples of each detection area and the sample category, after accumulating multiple times, the host MCU can train a deep learning model based on all samples of multiple times, and use the trained deep learning model to obtain the next time of the multiple times, whether the working area has a model prediction result of a person.

[0095] Specifically, the process can include the following steps:

[0096] 1) For the accumulated multiple times of all detection area samples, a plurality of training sets are constructed by random sampling with replacement, and a deep learning model is trained using the plurality of training sets and the category labels of each sample therein;

[0097] Among them, the deep learning model includes a random forest deep learning model, which is described in detail below.

[0098] First, the accumulated multiple times of all detection area samples can constitute an original data set, wherein each sample is the heartbeat information, respiratory information and voltage information extracted from a detection area at a time, and carries the corresponding category, i.e. sample category.

[0099] The sample information extracted from the radar signal and the infrared signal is grouped according to a set of I sample data sets, i.e. three kinds of information are grouped, and J training sets are generated by random sampling with replacement, and the size of each training set is also I. The probability of each sample being selected in one sampling is 1 / I, and the probability of not being selected is (I-1) / I. After I times of sampling, the probability of a sample not being selected is:

[0100]

[0101] At each split node of each tree of the random forest deep learning model, not all features (the category of the feature is three, including respiratory information, heartbeat information and voltage information) are considered, but a part of the features are randomly selected, and the randomly selected features are used to extract a subset from the data set (each feature information carries the corresponding category) for training. For example, it is possible to randomly select "heartbeat information" and "voltage information" to consider the split. Then let each tree grow as much as possible until there are no more features available for splitting. When all trees are constructed, they can be used to predict the next time result (whether the working area has a person), and the prediction results of all trees are integrated, such as 60 trees predicting "yes" and 40 trees predicting "no", and the final prediction result is "yes".

[0102] In the embodiment of the present application, the random forest deep learning model predicts by constructing multiple decision trees, each tree is trained using a subset of data and a subset of features, and the number of trees, maximum depth, minimum sample size and other parameters are adjusted through cross-validation and grid search method to find the optimal parameter combination, which can improve the accuracy of judgment.

[0103] In the embodiment of the present application, the training process of the random forest deep learning model can be understood in combination with the existing random forest deep learning model training process.

[0104] 2) Obtain the input data of the next moment of the plurality of moments, and input into the trained deep learning model to obtain the model prediction result, the model prediction result is used to represent whether there is a person in the working area of the intelligent air conditioner remote controller;

[0105] The input data includes heartbeat information, breathing information and voltage information extracted from all detection areas at the next moment.

[0106] The model prediction result of the next moment is compared with the actual measurement result of the person at the next moment, and when the two are inconsistent, the threshold values corresponding to the heartbeat information, breathing information and voltage information are adjusted to determine whether there is a person in the detection area in the next detection process. At the same time, the actual measurement result of the person at the next moment is also added to the original data set to continue to be used for prediction.

[0107] Air conditioner control module:

[0108] The air conditioner control module controls the working state of the air conditioner according to the actual measurement result of the person, and the process can include:

[0109] If the air conditioner control module receives the actual measurement result of the person in the working area within a first preset time, a first type of binary instruction code is generated, and after the first type of binary instruction code is modulated, the infrared communication module built-in the intelligent air conditioner remote controller is used to send the air conditioner to control the air conditioner to start;

[0110] If the air conditioner control module receives the actual measurement result of the person in the working area within a second preset time, a second type of binary instruction code is generated, and after the second type of binary instruction code is modulated, the infrared communication module built-in the intelligent air conditioner remote controller is used to send the air conditioner to control the air conditioner to start.

[0111] Specifically, since false detection is avoided while power consumption caused by frequent switching of the air conditioner is reduced, the embodiment of the present application sets a first preset time length for turning on the air conditioner and a second preset time length for turning off the air conditioner, and only controls the air conditioner to be turned on when the actual measurement results of the personnel all indicate that there is a person in the working area within the first preset time length, and only controls the air conditioner to be turned on when the actual measurement results of the personnel all indicate that there is no person in the working area within the second preset time length. The first preset time length and the second preset time length can be set according to actual needs, for example, the first preset time length can be 10 minutes and the second preset time length can be 30 minutes.

[0112] The binary instruction code is composed of binary data, and the modulation of the second type of binary instruction code can be pulse width modulation of the carrier signal and the binary instruction code to obtain a modulated signal. This process can be understood in related technologies and will not be described here.

[0113] In the embodiment of the present application, the infrared communication module built-in the intelligent air conditioner remote controller is designed to be compatible with air conditioners of various manufacturers, so that the purpose of controlling the working state by communicating with all air conditioners on the market can be achieved.

[0114] Specifically, the infrared communication module can be an infrared light-emitting diode, which emits the modulated signal towards the air conditioner, so that the infrared receiver of the air conditioner can receive it to realize turning on or off of the air conditioner.

[0115] Display screen:

[0116] The display screen can be a liquid crystal screen, on which the actual measurement results of the personnel and the air conditioner state information can be displayed. The air conditioner state information can include working parameters such as on-off state, working mode, temperature, wind force level, wind speed, wind direction, and the power saved by the intelligent air conditioner remote controller for the air conditioner.

[0117] Among them, the working parameters can be various parameters usually displayed on the display screen of the original remote controller of the air conditioner. If the intelligent air conditioner remote controller is independent of the original remote controller of the air conditioner, it can communicate with the air conditioner through the infrared communication module to obtain the above-mentioned working parameters. If the intelligent air conditioner remote controller is realized by integrating and improving the original remote controller of the air conditioner, its display screen can also obtain the above-mentioned working parameters. The specific process can be understood in related technologies and will not be described here.

[0118] Compared with the parameters displayed on the display screen of the traditional remote controller, the embodiment of the present application can display the power saved by the intelligent air conditioner remote controller for the air conditioner. This part of data can be obtained by multiplying the unit power consumption of the air conditioner in the on state and the off time of the air conditioner under the control of the intelligent air conditioner remote controller.

[0119] The working process of the intelligent air conditioner remote controller of the embodiment of the present application can be understood in Figure 2It is understood that the tagged data refers to determining the sample category, and the parameter optimization refers to adjusting the threshold value, and the remaining parts are no longer described one by one.

[0120] The application provides an intelligent air conditioner remote controller based on deep learning, which embeds a radar sensor and an infrared sensor in the remote controller, combines the detection of the presence of a human body to control the air conditioner, and the combination of the radar sensor and the infrared sensor can improve the accuracy of personnel detection, thereby improving the accuracy of the control of the working state of the air conditioner. At the same time, the application trains a random forest model to predict the presence of a human being at the next moment, compares it with the actual detection result to dynamically adjust the threshold value for judgment, and can improve the recognition rate of the presence of a human body. Compared with the radar sensor alone, the application can reduce the misjudgment; by automatically detecting the activity of the personnel in the room, the automatic on-off control of the air conditioner is realized, energy waste can be avoided, and user experience can be improved; and the degree of automation is high, the user does not need to operate frequently, the energy use efficiency can be improved, unnecessary power consumption can be reduced, and the application is convenient to install and use, can be compatible with existing air conditioner equipment, and has practical value.

[0121] It should be noted that in the description of the application, it should be understood that the terms "center", "longitudinal", "transverse", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the application and simplifying the description, and therefore cannot be understood as indicating or implying that the devices or elements indicated must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the application.

[0122] In addition, the terms "first", "second" are only for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more of the features. In the description of the application, the meaning of "a plurality of" is two or more, unless otherwise specifically limited.

[0123] In the description of the specification, reference to "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" means that a particular feature, structure, material, or characteristic being described is included in at least one embodiment or example of the application. The appearances of the phrases "in one embodiment", "in some embodiments", "an example", "a specific example", or "some examples" in various places in the specification are not necessarily referring to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples. Also, the terminology used in the description is for the purpose of describing particular embodiments only and is not intended to be limiting. It is also possible in the present application that additional or

[0124] The above descriptions are only the preferred embodiment of the application, not intended to limit the protection scope of the application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the application shall be included in the protection scope of the application.

Claims

1. A smart air conditioner remote control, characterized in that, include: Radar sensor assembly, infrared sensor assembly, main control MCU, display screen, and air conditioning control module; among which, The radar sensor assembly includes radar sensors disposed at multiple locations on the smart air conditioner remote control; each radar sensor is used to transmit radio waves to the corresponding detection area, receive the reflected signal at each moment, and send it to the main control MCU; wherein, the multiple locations include the front, back, left, and right locations of the smart air conditioner remote control; The infrared sensor assembly includes infrared sensors installed at multiple locations on the smart air conditioner remote control; each infrared sensor is used to detect human infrared radiation information within the corresponding detection area and outputs a voltage signal at each moment to the main control MCU. The main control MCU is used to continuously execute the detection process, which includes: at each moment, extracting heartbeat and breathing information from the reflected signals of each detection area, extracting voltage information from the voltage signal, obtaining samples of the corresponding detection area, and determining the sample category based on the thresholds corresponding to each extracted information to characterize whether there is a person in the detection area, and determining the actual personnel measurement result of the working area where the smart air conditioner remote control is located based on the sample categories of all detection areas at that moment; and training a deep learning model based on all samples accumulated over multiple moments, using the trained deep learning model to obtain the model prediction result of whether there is a person in the working area at the next moment, and adjusting the thresholds corresponding to each information for the next detection process when the prediction result is inconsistent with the actual personnel measurement result; wherein, at each moment, the sample of any detection area includes the heartbeat, breathing, and voltage information extracted from that detection area; the air conditioner control module is used to control the working state of the air conditioner according to the actual personnel measurement result; The display screen is used to display the personnel's actual measurement results and air conditioning status information; The step of determining the sample category based on the threshold corresponding to each extracted piece of information to characterize whether there are people in the detection area includes: At this moment, for each detection area, it is determined whether the sample in that detection area meets the following conditions: the corresponding heartbeat information is greater than the heartbeat threshold, the corresponding respiratory information is greater than the respiratory threshold, and the corresponding voltage information is greater than the voltage threshold. If these conditions are met, the sample category of that detection area is determined to be occupied; otherwise, the sample category of that detection area is determined to be unoccupied. The step of determining the actual measurement results of personnel in the working area where the smart air conditioner remote control is located based on the sample categories of all detection areas at that moment includes: If at any given moment a sample in a detection area is classified as "occupied," then the working area where the smart air conditioner remote control is located is determined to be classified as "occupied" at that moment; otherwise, the working area is determined to be classified as "unoccupied" at that moment. If all categories show "person" within the preset time frame up to that moment, then the actual personnel measurement result for the work area is determined to be "person"; otherwise, the actual personnel measurement result for the work area is determined to be "person"; The process of training a deep learning model based on all samples accumulated over multiple time points, and using the trained deep learning model to obtain the model prediction result for whether there are people in the working area at the next time point, includes: For samples from all detected regions obtained at multiple time points, multiple training sets are constructed by random sampling with replacement. A deep learning model is trained using the multiple training sets and the class labels of each sample therein. The deep learning model includes a random forest deep learning model. The input data for the next moment of the multiple moments is obtained and fed into the trained deep learning model to obtain the model prediction result. The model prediction result is used to characterize whether there is a person in the working area where the smart air conditioner remote control is located. The input data includes the heartbeat information, breathing information and voltage information extracted from all detection areas at the next moment.

2. The intelligent air conditioner remote control according to claim 1, characterized in that, For any given moment, extract heartbeat and respiratory information from the reflected signals of each detection area, including: At that moment, for each detection area corresponding to each direction, the reflected signal from the radar sensor in that detection area is preprocessed, and heartbeat and breathing information are extracted from the preprocessed signal.

3. The intelligent air conditioner remote control according to claim 2, characterized in that, The preprocessing of the reflected signals from the radar sensors in the detection area includes: The reflected signals from the radar sensors in the detection area are sequentially subjected to mean filtering for noise reduction, linear interpolation, and time-frequency conversion to obtain the preprocessed signal.

4. The intelligent air conditioner remote control according to claim 3, characterized in that, The extraction of heartbeat and respiratory information from the preprocessed signal includes: Within the frequency range of the preprocessed signal, the frequency corresponding to the peak point within the preset heart rate frequency range is determined as heart rate information, and the frequency corresponding to the peak point within the preset respiratory frequency range is determined as respiratory information.

5. The intelligent air conditioner remote control according to claim 1, characterized in that, The air conditioning control module controls the operating status of the air conditioner based on the measured results of the personnel, including: If the air conditioning control module receives the actual measurement results of personnel within a first preset time period and all of them are that there are people in the work area, it generates a first type of binary instruction code, modulates the first type of binary instruction code, and sends it to the air conditioner through the infrared communication module built into the smart air conditioner remote control to control the air conditioner to turn on. If the air conditioning control module receives personnel measurement results within a second preset time period, indicating that the work area is unoccupied, it generates a second type of binary instruction code, modulates the second type of binary instruction code, and sends it to the air conditioner via the infrared communication module to control the air conditioner to turn off.

6. The intelligent air conditioner remote control according to claim 1 or 5, characterized in that, The air conditioner status information includes: operating parameters including on / off status, working mode, temperature, fan speed, wind speed, and wind direction, as well as the power saving of the air conditioner by the smart air conditioner remote control.

Images