Air conditioner control method, device, air conditioner and storage medium

Abstract

The present invention discloses an air conditioner control method, device, air conditioner, and storage medium. The method comprises: collecting a user's sleep information through radar detection technology, and predicting and outputting the sleep information using a neural network model to obtain the user's sleep state; collecting indoor noise information through a multimodal noise sensing system, and performing voiceprint analysis on the noise information to obtain a voiceprint analysis result; generating a control instruction based on the sleep state and the voiceprint analysis result, and controlling the operation of the air conditioner based on the control instruction. The present invention uses millimeter-wave radar technology and a neural network model to accurately determine the user's sleep state, and simultaneously performs voiceprint analysis on the noise information through a multimodal noise sensing system. This allows for sleep state monitoring and adaptive control of air conditioner noise, thereby improving the performance of the air conditioner.

Description

Technical Field

[0001] The present invention relates to the technical field of air conditioners, and in particular to an air conditioner control method and device, an air conditioner, and a storage medium. Background Art

[0002] In existing technologies, smart air conditioning systems typically monitor sleep status using simple sensors (such as infrared or pressure sensors) and adjust noise levels based on fixed thresholds. For example, when ambient noise exceeds a certain threshold, the air conditioner switches to low-noise mode. However, existing smart air conditioning systems have at least the following shortcomings:

[0003] (1) Insufficient sleep monitoring accuracy. Existing smart air conditioning systems mostly use infrared or pressure sensors to monitor sleep status. This method cannot accurately distinguish between sleeping and awake states, which results in the inability of smart air conditioning systems to accurately control the user's actual sleep status;

[0004] (2) Limited noise control capabilities. Existing intelligent air conditioning systems’ noise control is usually based on fixed thresholds or simple ambient noise detection. They lack the ability to separate the air conditioning operating noise from the ambient noise and cannot achieve dynamic adjustment, resulting in unsatisfactory noise reduction effects.

[0005] (3) Single user interaction function. The user interface functions of existing intelligent air conditioning systems are limited. Users can only make simple settings and cannot adjust the noise control logic according to personal preferences, making it difficult to meet diverse needs.

[0006] (4) Poor real-time control. The existing intelligent air-conditioning system has a slow response speed and cannot quickly switch to sleep mode, which affects the user experience;

[0007] Sleep quality is influenced by numerous factors, with noise undoubtedly being one of the most critical. Air conditioners inevitably generate a certain amount of noise during operation, which can affect a user's sleep quality. Therefore, how to implement sleep status monitoring and adaptively control air conditioner noise to improve the user experience is a challenge facing those skilled in the art. Summary of the Invention

[0008] Embodiments of the present invention provide an air conditioner control method and device, an air conditioner, and a storage medium, aiming to improve the use effect of the air conditioner.

[0009] In a first aspect, an embodiment of the present invention provides an air conditioner control method, comprising:

[0010] Collecting the user's sleep information through radar detection technology, and using a neural network model to predict and output the sleep information to obtain the user's sleep status;

[0011] Collecting indoor noise information through a multimodal noise perception system, and performing voiceprint analysis on the noise information to obtain a voiceprint analysis result;

[0012] A control instruction is generated in combination with the sleep state and the voiceprint analysis result, and the operation of the air conditioner is controlled based on the control instruction.

[0013] In a second aspect, an embodiment of the present invention provides an air conditioner control device, comprising:

[0014] A first collecting unit is configured to collect the user's sleep information through radar detection technology, and predict and output the sleep information using a neural network model to obtain the user's sleep state;

[0015] a second collecting unit, configured to collect indoor noise information through a multimodal noise sensing system, and perform voiceprint analysis on the noise information to obtain a voiceprint analysis result;

[0016] An operation control unit is used to generate a control instruction in combination with the sleep state and the voiceprint analysis result, and to control the operation of the air conditioner based on the control instruction.

[0017] In a third aspect, an embodiment of the present invention provides an air conditioner, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the air conditioner control method as described in the first aspect when executing the computer program.

[0018] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the air conditioner control method as described in the first aspect is implemented.

[0019] Embodiments of the present invention provide an air conditioner control method, device, air conditioner, and storage medium. The method comprises: collecting a user's sleep information through radar detection technology, and predicting and outputting the sleep information using a neural network model to obtain the user's sleep state; collecting indoor noise information through a multimodal noise sensing system, and performing voiceprint analysis on the noise information to obtain a voiceprint analysis result; generating control instructions based on the sleep state and voiceprint analysis result, and controlling the operation of the air conditioner based on the control instructions. Embodiments of the present invention use millimeter-wave radar technology and a neural network model to accurately determine the user's sleep state, while simultaneously performing voiceprint analysis on the noise information through a multimodal noise sensing system. This allows for sleep state monitoring and adaptive control of air conditioner noise, thereby improving the performance of the air conditioner. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0021] Figure 1 A schematic flow chart of an air conditioner control method provided by an embodiment of the present invention;

[0022] Figure 2 A schematic diagram of a sub-flow chart of step S101 in an air conditioner control method provided by an embodiment of the present invention;

[0023] Figure 3 A schematic diagram of a sub-flow chart of step S102 in an air conditioner control method provided by an embodiment of the present invention;

[0024] Figure 4 A schematic diagram of a sub-flow chart of step S103 in an air conditioner control method provided by an embodiment of the present invention;

[0025] Figure 5 A schematic diagram of a control method for an air conditioner according to an embodiment of the present invention;

[0026] Figure 6 A communication diagram of an air conditioner control method provided by an embodiment of the present invention;

[0027] Figure 7 A neural network architecture diagram of an air conditioner control method provided by an embodiment of the present invention;

[0028] Figure 8 A schematic block diagram of an air conditioner control device provided by an embodiment of the present invention;

[0029] Figure 9 A first sub-schematic block diagram of an air conditioner control device provided by an embodiment of the present invention;

[0030] Figure 10 A second sub-schematic block diagram of an air conditioner control device provided by an embodiment of the present invention;

[0031] Figure 11 A third sub-schematic block diagram of an air conditioner control device provided by an embodiment of the present invention;

[0032] Figure 12 A schematic block diagram of an air conditioner provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0033] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0034] It will be understood that when used in this specification and the appended claims, the terms “comprises” and “comprising” indicate the presence of described features, integers, steps, operations, elements and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof.

[0035] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the present invention. As used in the specification and appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms unless the context clearly indicates otherwise.

[0036] It should be further understood that the term "and / or" used in the present description and appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0037] See below Figure 1 An embodiment of the present invention provides an air conditioner control method, which specifically includes: steps S101 to S103.

[0038] Step S101: collecting user's sleep information through radar detection technology, and using a neural network model to predict and output the sleep information to obtain the user's sleep state;

[0039] Step S102: collecting indoor noise information through a multimodal noise sensing system, and performing voiceprint analysis on the noise information to obtain a voiceprint analysis result;

[0040] Step S103: Generate a control instruction based on the sleep state and the voiceprint analysis result, and control the operation of the air conditioner based on the control instruction.

[0041] In this embodiment, combined with Figure 5 First, radar detection technology, such as millimeter-wave radar technology, is used to send pulses to the human body and receive echo pulses reflected by the human body to collect the user's sleep information. Then, the user's sleep state is predicted through a neural network model. Secondly, a multimodal noise perception system (i.e. Figure 5The noise module in the sleep state analyzes the noise in the room and obtains the voiceprint analysis results. It then generates control instructions based on the sleep state and the voiceprint analysis results to control the operation of the air conditioner.

[0042] This embodiment uses radar detection technology, such as millimeter-wave radar technology and a neural network model, to accurately determine the user's sleep state. At the same time, a multimodal noise perception system is used to perform voiceprint analysis and processing on noise information. This can achieve sleep state monitoring and adaptive control of air conditioner noise, thereby improving the use effect of the air conditioner.

[0043] In actual application scenarios, a 60GHz millimeter-wave radar module can be used to collect sleep information. This module can monitor the body's subtle movements and breathing signals in real time to determine the user's sleep state. The millimeter-wave radar module determines the user's sleep state by monitoring subtle movements (such as breathing and body movement). When the user falls asleep, the radar module detects a significant decrease in body movement frequency and breathing rate. Specifically, the sleep state is divided into two modes: "falling asleep" and "awake":

[0044] Sleeping mode: low body movement frequency and steady breathing;

[0045] Awake mode: high body movement rate and unstable breathing rate.

[0046] The signals collected by the millimeter-wave radar module mainly include the human body's breathing rate, body movement frequency, and body movement amplitude. The following characteristic parameters can be selected (including but not limited to):

[0047] Respiratory rate (unit: breaths / minute);

[0048] Body movement frequency (unit: times / minute);

[0049] Body motion amplitude (unit: mm);

[0050] Distance (unit: centimeters).

[0051] The above characteristics can effectively reflect the human body's sleep state: breathing rate and body movement amplitude are higher when awake, but will decrease significantly after falling asleep. Figure 6 The millimeter-wave radar module can be connected to the air-conditioning control mainboard through serial communication, thereby realizing real-time data transmission and processing. The pin definition of the millimeter-wave radar module is shown in Table 1:

[0052] Table 1

[0053]

[0054] In one embodiment, if Figure 2 As shown, step S101 includes steps S201 to S203.

[0055] Step S201: inputting the sleep information into the input layer of a multi-layer perceptron model;

[0056] Step S202: performing feature processing on the sleep information using the hidden layer of the multi-layer perceptron model to obtain sleep features corresponding to the sleep information;

[0057] Step S203: Utilize the output layer of the multi-layer perceptron model to perform probability distribution prediction on the sleep characteristics, and output the sleep state based on the result of the probability distribution prediction.

[0058] In this embodiment, a multi-layer perceptron (MLP) model is used to predict the sleep state corresponding to the sleep information. In practical applications, the TensorFlow platform can be used for training. Figure 7 , the multi-layer perceptron (MLP) model structure in this embodiment is as follows:

[0059] Input layer: contains three input features, namely breathing frequency, body movement frequency and body movement amplitude;

[0060] Hidden layer: There are 3 layers, each with 128, 64 and 32 neurons respectively, and the activation function of the hidden layer is ReLU;

[0061] Output layer: Contains 2 neurons, corresponding to the "sleeping" and "awake" states respectively, and uses the softmax function to convert the output into a probability distribution.

[0062] The specific processing process is as follows: if the input feature vector is [0.375, 0.25, 0.2], the input vector of the input layer [0.375, 0.25, 0.2] is directly passed in, with a dimension of 3. In the hidden layer, the input of the first hidden layer (128 neurons) is processed by 128 neurons. Each neuron performs a weighted summation on the input and adds a bias. The activation function ReLU sets negative values to 0 and outputs 128 values. The first hidden layer is used to extract nonlinear relationships between features, such as the combination of low body movement frequency and high breathing smoothness; the second hidden layer (64 neurons) is input from the output of the first hidden layer to 64 neurons, and continues to perform weighted summation and add a bias. It is processed by the activation function ReLU and outputs 64 values. The second hidden layer is used to further abstract features and may identify more complex sleep patterns; the input of the third hidden layer (32 neurons) is the output of the second hidden layer, which is input to 32 neurons, and is weighted summed and added with a bias. It is processed by the activation function ReLU and outputs 32 values. The third hidden layer is used to compress features and prepare for classification. The input to the output layer (2 neurons) is the output of the third hidden layer, which is passed to 2 neurons. Unnormalized scores are calculated, for example, [2.2, -1.3]. The softmax function is used to convert these scores into probabilities. For example, the probability of the first item is 0.95 (calculated as e^2.2 / (e^2.2 + e^-1.3)), and the probability of the second item is 0.05 (calculated as e^-1.3 / (e^2.2 + e^-1.3)). The output probability distribution is [0.95, 0.05], representing the probabilities of "falling asleep" and "being awake," respectively. For example, the input is [0.375, 0.25, 0.2]. After processing through three hidden layers, the characteristic pattern is extracted, and the output layer score is [2.2, -1.3]. After the softmax calculation, the probability distribution is [0.95, 0.05].

[0063] Furthermore, based on the output probability distribution, the user's sleep state can be determined. For example, if the confidence score of the probability of falling asleep is greater than a preset first confidence threshold (e.g., 0.85), the user is considered "asleep." If the confidence score of the probability of being awake is greater than a preset second confidence threshold (e.g., 0.5) and the duration exceeds 30 minutes, the user is considered "awake." Setting thresholds ensures classification reliability and avoids false positives when the probability approaches 0.5. For example, if the output probability distribution is [0.95, 0.05], the probability of falling asleep (0.95) is greater than 0.85, resulting in a "sleep" state. If the output probability distribution is [0.3, 0.7], the probability of being awake (0.7) is greater than 0.5, resulting in a "awake" state.

[0064] In one embodiment, if Figure 3 As shown, step S102 includes: steps S301 to S305.

[0065] Step S301: Collect indoor sound signals through a microphone array and convert the sound signals into time domain waveforms;

[0066] Step S302: collecting a mechanical vibration signal of the air conditioner through a vibration sensor, and outputting three-axis acceleration data based on the mechanical vibration signal;

[0067] Step S303: performing a fast Fourier transform on the time domain waveform to obtain corresponding first spectrum information, and performing a fast Fourier transform on the three-axis acceleration data to obtain corresponding second spectrum information;

[0068] Step S304: Calculate the spectral centroids of the first spectrum information and the second spectrum information respectively, and extract pulse characteristics;

[0069] Step S305: performing noise separation on the noise information based on the calculation result of the spectrum center of gravity and the extraction result of the pulse characteristics to obtain the operating noise of the air conditioner and the indoor environmental noise; wherein the environmental noise includes broadband random noise and high-frequency pulse noise.

[0070] In this embodiment, the multimodal noise perception system specifically includes a microphone array, a vibration sensor, and a voiceprint analysis module. The technical indicators and functional descriptions thereof can be shown in Table 2 below:

[0071] Table 2

[0072]

[0073] As shown in Table 2, the microphone array uses a four-channel digital MEMS microphone (signal-to-noise ratio (SNR) ≥ 65dB) to capture airborne acoustic signals. Using spatial sound field reconstruction technology, it can locate noise sources and distinguish whether the noise originates from the air conditioner itself or elsewhere in the environment. The vibration sensor, a triaxial accelerometer (±2g range), can be installed near the air conditioner's compressor or fan to detect mechanical vibrations generated during operation. These vibration signals are closely related to noise; for example, low-frequency vibrations from the compressor can be converted into airborne noise. The voiceprint analysis module, based on an embedded DSP (supporting 2048-point fast Fourier transform (FFT) real-time analysis), processes the data collected by the microphone array and vibration sensor. By analyzing the spectral characteristics of the sound waves and vibrations, the source and nature of the noise can be determined, thereby separating the air conditioner's operating noise from the ambient noise. In actual testing, the noise separation error was kept within 3dB.

[0074] Specifically, the microphone array collects sound signals in the air at a sampling rate of 48kHz (a common digital MEMS microphone standard) and generates a time-domain waveform. The vibration sensor collects mechanical vibration signals from the air conditioning equipment and outputs triaxial acceleration data at a sampling rate of 1kHz.

[0075] Based on the Nyquist sampling theorem, a 2048-point Fast Fourier Transform (FFT) is performed on the sound signal collected by the microphone, converting the time domain signal into the frequency domain, resulting in a spectrum with a frequency range of 0-24kHz. An FFT is also performed on the vibration signal to analyze the vibration components within the frequency range of 0-500Hz. Air conditioner operating noise typically has specific frequency characteristics, such as the 60Hz fundamental frequency of the compressor and its harmonics (120Hz, 180Hz, etc.), while ambient noise may include broadband random noise or high-frequency impulse noise.

[0076] Next, feature extraction is performed. This involves calculating the spectral centroid (spectral_centroid), which reflects the frequency distribution of noise energy. The spectral centroid of air conditioning noise is typically between 0 and 1 kHz, while ambient noise may tend to be higher frequencies. Impulse characteristics (impulse_ratio) are also extracted. By measuring the ratio of the signal's peak value to its mean, the presence of sudden noise (such as the sound of an object falling) is determined. Characteristic frequencies (tonal_components), such as the 60Hz harmonics of the compressor, are then identified to label air conditioning noise.

[0077] In specific embodiments, a pre-trained machine learning model (such as a support vector machine (SVM) or a simple threshold classifier) can be used to distinguish air conditioning noise from ambient noise based on extracted features. For example, if the proportion of 60Hz harmonics in the signal is greater than 10%, it is determined to be air conditioning noise. Vibration signal verification can also be combined. For example, if a vibration sensor detects periodic vibrations related to 60Hz and matches the microphone signal, it can be further confirmed as air conditioning noise. In addition, spatial positioning can be used, such as using the four-channel data of the microphone array through beamforming or time difference of arrival (TDOA) technology to calculate the spatial location of the noise source. If the direction of the noise source aligns with the location of the air conditioner, it is classified as air conditioning noise; otherwise, it is considered ambient noise.

[0078] Through the above operations, the noise information can be finally separated into two groups of signals, namely the operating noise of the air conditioner and the indoor environmental noise.

[0079] In practical applications, noise information can be processed by noise feature engineering, where the key parameters are as follows:

[0080] noise_features = {

[0081] 'A_weighting': True, # Human ear equivalent loudness

[0082] 'spectral_centroid': 0-4kHz, # spectral center

[0083] 'impulse_ratio': 0-100%, # burst noise ratio

[0084] 'tonal_components': [], # Characteristic frequency components (such as compressor 60Hz harmonics)

[0085] }

[0086] The purpose of noise feature engineering is to extract key quantitative features from the original noise signal so that the system can accurately analyze, separate, and control the noise. Specific functions include:

[0087] Quantifying noise characteristics: Converting complex sound and vibration signals into calculable parameters, such as spectral center of gravity and burst noise ratio;

[0088] Support noise separation: Provides input to the voiceprint analysis module to help distinguish air conditioning noise from ambient noise;

[0089] Guiding noise reduction strategies: Characteristic data is used to determine noise levels and types, thereby triggering appropriate control measures.

[0090] Noise feature engineering relies on the raw signals collected by microphone arrays and vibration sensors. The collected sound and vibration data are the inputs for feature extraction. For example, the microphone signal is used to calculate the center of gravity of the spectrum, and the vibration signal is used to verify the characteristic frequency. The extracted features will directly affect the selection of subsequent noise reduction modes. For example, if the proportion of burst noise (impulse_ratio) is greater than 50%, the "emergency mute" mode may be triggered; if the center of gravity of the spectrum is low and contains 60Hz harmonics, the compressor frequency may be adjusted. For example:

[0091] A_weighting: simulates human ear perception, and the weighted noise value is used to evaluate the actual interference level;

[0092] spectral_centroid: reflects energy distribution and guides fan or compressor adjustment;

[0093] impulse_ratio: identifies sudden noise and triggers transient noise cancellation;

[0094] tonal_components: locks the characteristic frequency of the air conditioner and optimizes the variable frequency control.

[0095] Specifically, A_weighting simulates human ear perception and adjusts noise measurements to reflect the human ear's sensitivity to different frequencies. Low and high frequencies are attenuated, while mid-range frequencies (1-4kHz) are enhanced. This is achieved by applying an A-weighting filter to the spectrum collected by the microphone. The calculation formula is:

[0096] ;

[0097] Where W(f) is the A-weighting function and X(f) is the spectrum amplitude.

[0098] spectral_centroid: represents the frequency "center of gravity" of the noise signal, which reflects the energy distribution and is used to distinguish low-frequency air conditioning noise from high-frequency ambient noise. It can be calculated using the following formula:

[0099] spectral_centroid= ;

[0100] Among them, f k is the kth frequency point, X(k) is the corresponding amplitude, ranging from 0 to 4kHz.

[0101] impulse_rati is used to quantify the proportion of burst noise, suitable for triggering emergency silent mode. It detects the peak-to-mean ratio of the signal and compares it with a preset threshold. If it exceeds the preset threshold (for example, the peak value is greater than 3 times the mean), it is determined to be a burst noise. Within a time window (for example, 1 second, 48,000 sample points), the proportion of sample points exceeding the threshold is calculated. For example:

[0102] impulse_ratio=N impulse / N total ×100%;

[0103] Among them, N impulse represents burst noise, N total Indicates the total amount of noise.

[0104] Tonal_components can identify specific frequency components (such as the 60Hz harmonics of a compressor) to label air conditioning noise. A peak detection algorithm is applied to the spectrum to find frequencies with amplitudes twice the average, such as 60Hz and 120Hz.

[0105] In one embodiment, the air conditioner control method further includes:

[0106] According to the following formula, the noise information is compensated using a dynamic compensation algorithm:

[0107] ;

[0108] in, Indicates the noise compensation value, K v Represents the vibration compensation coefficient, K a represents the correction value of the acoustic transfer function, x represents the vibration displacement, and t represents the time.

[0109] In this embodiment, the dynamic compensation algorithm is used to further improve the accuracy of noise separation and the robustness of the system. In real-world environments, noise transmission and reception may be affected by a variety of factors, such as the sound absorption properties of walls, room layout, and furniture placement. These factors may cause the sound signal received by the microphone array to differ from the actual air conditioning noise. By introducing a dynamic compensation algorithm, the sound signal can be corrected in real time based on the mechanical vibration information measured by the vibration sensor, thereby more accurately reflecting the actual air conditioning noise.

[0110] Specifically, the vibration compensation coefficient K v It reflects the correlation between the vibration signal and the sound signal, which can be obtained based on historical data and experimental measurements. a This algorithm takes into account the influence of the indoor acoustic environment, including the propagation path, reflection, and attenuation of sound waves, which can be obtained through pre-measurement and modeling. In actual applications, the system will dynamically calculate the noise compensation value based on the real-time collected vibration displacement x and time t, and apply it to the sound signal, thereby achieving accurate compensation for noise information. By introducing a dynamic compensation algorithm, this embodiment can more effectively separate the air conditioner operating noise from the ambient noise, providing more accurate input for subsequent noise reduction control. This not only improves the noise reduction effect of the system, but also enhances the user experience, making the air conditioner quieter and more comfortable during operation.

[0111] In one embodiment, if Figure 4 As shown, the sleep state includes the awake state and the asleep state, and the step S103 includes: steps S401 to S403.

[0112] Step S401: When the sleeping state is changed to the awake state and the awake state lasts for a preset time threshold, a basic noise reduction control instruction is generated;

[0113] Here, the basic noise reduction control instruction may specifically refer to adjusting the current angle of the air guide plate of the air conditioner and adjusting the current frequency of the compressor;

[0114] Adjusting the angle of the deflectors within the air duct can smooth the airflow and reduce turbulent noise. For example, if wind noise is detected concentrated at 2kHz (caused by turbulence), adjusting the deflector angle from 30° to 45° can increase the airflow's wall adhesion and reduce noise by approximately 3-5dB.

[0115] Fine-tuning the frequency can also change the vibration characteristics of the compressor, allowing you to find the lowest-noise operating point without affecting cooling. For example, if the compressor's current frequency is 50Hz, you can adjust it by +5Hz (55Hz) or -5Hz (45Hz) based on the noise test results. For example, if the current frequency is 60Hz and the harmonics are too strong, you can adjust it to -5Hz to avoid resonance.

[0116] Step S402: When the sleep state is a falling asleep state and the confidence level of the falling asleep state reaches a preset confidence threshold, a deep noise reduction control instruction is generated;

[0117] Here, the deep noise reduction control instruction may specifically refer to downshifting the current speed of the air conditioner fan and performing shutdown compensation processing on the compressor;

[0118] In Deep Noise Reduction mode, briefly stopping the compressor can reduce noise while also compensating for temperature fluctuations. For example, after stopping the compressor, cold storage materials (such as phase change materials) release cooling energy to maintain indoor temperature fluctuations below 1°C. The downtime can be set to meet actual needs, for example, less than 5 minutes.

[0119] Step S403: When the noise information contains sudden noise and the sudden noise reaches a preset noise threshold, an emergency mute control instruction is generated;

[0120] Here, the emergency mute control instruction may specifically refer to controlling the transient mute of the air conditioner and controlling the soft start and stop of the night mode of the air conditioner.

[0121] Transient noise cancellation uses active noise control (ANC) to generate anti-phase sound waves in real time to cancel out sudden noise, with a response time of less than 50ms. After canceling sudden noise, the air conditioner gradually restarts after a 0.5-second wait to avoid sudden noise. Night mode also features soft start / stop, where the fan speed gradually increases from 0 to 800 rpm (taking 2 seconds) and then decreases when it stops, preventing sudden noise changes.

[0122] In this embodiment, by combining multiple control strategies, not only can the precise separation and control of the air-conditioning operating noise be achieved, but also intelligent noise reduction adjustment can be performed according to the user's sleep state, thereby providing the user with a quieter and more comfortable resting environment. Combined with Table 3, specifically, in the basic noise reduction (continuous wakefulness > 30 minutes) mode, considering that the user is less sensitive to noise when awake, mild measures (such as air duct optimization and ±5Hz frequency conversion) are taken, which not only reduces noise but also maintains the cooling effect. In the deep noise reduction (confidence in falling asleep > 85%) mode, since the user is extremely sensitive to noise when falling asleep, it is necessary to significantly reduce the noise (such as fan speed ≤ 800rpm, compressor shutdown) to give priority to sleep. In the emergency silent mode (sudden noise exceeding the standard), since sudden noise (such as turning over) may wake the user up, it is necessary to immediately silence the noise and make a smooth transition, so the transient silence (0.5 second delay start) + night mode soft start and stop method is adopted for control;

[0123] Table 3

[0124]

[0125] In one embodiment, step S103 further includes:

[0126] The air conditioner is operated and controlled according to a preset interrupt priority control strategy; wherein the interrupt priorities include, from high to low, noise exceeding standard events, sleep state update events, and user configuration update events.

[0127] In this embodiment, by setting an interrupt priority control strategy, the air conditioner can ensure a quick and appropriate response to different types of events during operation. Specifically, referring to Table 4, this embodiment includes three interrupt sources: noise exceeding standard events, sleep state updates, and user configuration updates. The noise exceeding standard event has the highest priority, with a response period (the time it takes for the system to start processing after event detection) of less than 50 μs, achieving immediate noise reduction. The processing involves issuing an emergency speed reduction command, such as the MCU sending a PWM or CAN_FD command to the fan driver or compressor controller to reduce the fan speed, such as from 1000 rpm to 800 rpm. The sleep state update has a medium priority, with a response period of 200 ms. The processing primarily involves adjusting the mode, switching modes based on the sleep state and noise level, such as switching from basic noise reduction to deep noise reduction when the confidence level is greater than 85%. The user configuration update has the lowest priority, with a response period of 1 second. The processing primarily involves loading new rules, parsing new user rules (such as "weekend nap mode"), and updating the control logic in memory for application in the next cycle.

[0128] Table 4

[0129]

[0130] In actual application scenarios, a user-defined rule engine can be built to support personalized noise control logic, allowing users to set preferences through a visual interface to meet the needs of different scenarios and enhance user satisfaction. The rule configuration system can be as follows:

[0131] {

[0132] "rule_name": "Weekend nap mode",

[0133] "condition": {

[0134] "time_range": "12:00-14:00",

[0135] "sleep_state": "Light sleep phase",

[0136] "noise_limit": "≤20dB(A)"

[0137] },

[0138] "action": {

[0139] "compressor": "eco_mode",

[0140] "fan_speed": "level2",

[0141] "airflow_direction": "ceiling"

[0142] }

[0143] }

[0144] The above rule configuration defines an air conditioning control rule called "Weekend Nap Mode." The rule is triggered when the user is in light sleep between 12:00 PM and 2:00 PM, and the ambient noise level does not exceed 20 decibels. Once these conditions are met, the air conditioner sets the compressor to energy-saving mode, adjusts the fan speed to level 2, and sets the air flow direction to the ceiling. These settings are designed to provide a quiet and comfortable environment for weekend naps.

[0145] When building a visual programming interface, you can use a drag-and-drop logic builder that supports IF-THEN-ELSE conditional chains (with a maximum nesting depth of 5 levels). You can also set up a parameter fine-tuning panel, such as sliders to adjust the noise threshold (1dB steps) and response delay (0-60 seconds).

[0146] Furthermore, to implement embedded control, the following hardware deployment plan is made:

[0147] The main control chip uses STM32H743VIT6 (dual-core Cortex-M7 / M4)

[0148] Real-time control bus:

[0149] graph LR

[0150] Radar-->|SPI@20MHz|MCU

[0151] Microphone-->|I2S|MCU

[0152] MCU-->|CAN_FD|Compressor

[0153] MCU-->|PWM|Fan_Driver

[0154] Here, the real-time control bus is the data communication network between the modules of the system;

[0155] Radar→MCU means transmitting radar data via the SPI bus at a rate of 20MHz;

[0156] Microphone → MCU, refers to the transmission of 48kHz audio through the I2S interface;

[0157] MCU→Compressor means sending frequency control instructions through the CAN_FD bus;

[0158] MCU → Fan_Driver: This controls the fan speed via PWM signals. This ensures real-time performance, for example, with a response time of <50μs for noise exceeding the specified limit.

[0159] In another practical application scenario, the air conditioner control method provided in this embodiment is verified and optimized in combination with Table 5;

[0160] Table 5

[0161]

[0162] Based on Table 5 above, it can be seen that the air conditioner control method provided by this embodiment has shown good performance in practical applications. Specifically, in the comparative test conducted in a semi-anechoic chamber, the noise level of the air conditioner measured at a distance of 1 meter in the deep noise reduction mode was less than 18 decibels, which meets the requirements of the national standard GB / T 4214. In the test of simulating the injection of user body motion signals, the time from the system detecting the signal to completing the mode switching was less than 1.2 seconds, ensuring the timeliness of the response. In addition, in the test of 1,000 preset scenarios, the false trigger rate of rule execution was less than 0.5%, verifying the accuracy and reliability of the rule configuration system. These test results not only prove the effectiveness of the air conditioner control method provided by this embodiment, but also provide strong support for its promotion in practical applications.

[0163] Overall, this embodiment significantly improves user experience and sleep quality through high-precision sleep monitoring, multimodal noise sensing and dynamic control, a user-defined rules engine, and optimized real-time performance and response speed. Specifically, in terms of high-precision sleep monitoring, the system utilizes a 60GHz millimeter-wave radar and a neural network algorithm to achieve over 90% accuracy in detecting falling asleep. This allows precise differentiation between sleep states and provides a reliable basis for dynamic noise reduction. In terms of multimodal noise sensing and dynamic control, a microphone array, vibration sensor, and voiceprint analysis module are used to separate air conditioning noise from ambient noise. Combined with a graded noise reduction strategy, this approach accurately reduces noise, improving noise reduction by 40% in practical applications. The user-defined rules engine allows users to customize noise control logic through a visual interface to meet the needs of different scenarios and enhance user satisfaction. In terms of real-time performance and response speed, a graded interrupt response mechanism ensures rapid response to noise-exceeding events (<50μs) and sleep status updates (<1.2 seconds), ensuring system real-time performance and improving overall performance.

[0164] Figure 8 This is a schematic block diagram of an air conditioner control device 800 provided in an embodiment of the present invention. The device 800 includes:

[0165] The first collecting unit 801 is used to collect the user's sleep information through radar detection technology, and predict and output the sleep information using a neural network model to obtain the user's sleep state;

[0166] The second collecting unit 802 is configured to collect indoor noise information through a multimodal noise sensing system, and perform voiceprint analysis on the noise information to obtain a voiceprint analysis result;

[0167] The operation control unit 803 is used to generate a control instruction based on the sleep state and the voiceprint analysis result, and control the operation of the air conditioner based on the control instruction.

[0168] In one embodiment, if Figure 9 As shown, the first acquisition unit 801 includes:

[0169] An information input unit 901 is used to input the sleep information into an input layer of a multi-layer perceptron model;

[0170] A feature processing unit 902 is configured to perform feature processing on the sleep information using the hidden layer of the multi-layer perceptron model to obtain sleep features corresponding to the sleep information;

[0171] The probability distribution unit 903 is configured to perform probability distribution prediction on the sleep feature using the output layer of the multi-layer perceptron model, and output the sleep state based on the result of the probability distribution prediction.

[0172] In one embodiment, if Figure 10 As shown, the second acquisition unit 802 includes:

[0173] The sound collection unit 1001 is used to collect indoor sound signals through a microphone array and convert the sound signals into time domain waveforms;

[0174] a vibration collection unit 1002 for collecting a mechanical vibration signal of the air conditioner through a vibration sensor and outputting triaxial acceleration data based on the mechanical vibration signal;

[0175] a spectrum conversion unit 1003 configured to perform a fast Fourier transform on the time domain waveform to obtain corresponding first spectrum information, and to perform a fast Fourier transform on the triaxial acceleration data to obtain corresponding second spectrum information;

[0176] The spectrum processing unit 1004 is configured to calculate the spectrum centroid of the first spectrum information and the second spectrum information respectively, and extract pulse characteristics;

[0177] The noise separation unit 1005 is used to perform noise separation on the noise information based on the calculation result of the spectrum center of gravity and the extraction result of the pulse characteristics to obtain the operating noise of the air conditioner and the indoor environmental noise; wherein the environmental noise includes broadband random noise and high-frequency pulse noise.

[0178] In one embodiment, the air conditioner control device 800 further includes:

[0179] The noise compensation unit is configured to compensate the noise information using a dynamic compensation algorithm according to the following formula:

[0180] ;

[0181] in, Indicates the noise compensation value, K vRepresents the vibration compensation coefficient, K a represents the correction value of the acoustic transfer function, x represents the vibration displacement, and t represents the time.

[0182] In one embodiment, if Figure 11 As shown, the operation control unit 803 includes:

[0183] The first instruction generating unit 1101 is configured to generate a basic noise reduction control instruction when the sleeping state is changed to the awake state and the awake state lasts for a preset time threshold;

[0184] The second instruction generating unit 1102 is configured to generate a deep noise reduction control instruction when the sleep state is a falling asleep state and the confidence level of the falling asleep state reaches a preset confidence threshold;

[0185] The third instruction generating unit 1103 is configured to generate an emergency mute control instruction when the noise information contains burst noise and the burst noise reaches a preset noise threshold.

[0186] In one embodiment, the basic noise reduction control instruction includes: adjusting the current angle of the air guide plate of the air conditioner and adjusting the current frequency of the compressor;

[0187] The deep noise reduction control instruction includes: downshifting the current speed of the air conditioner fan and performing shutdown compensation processing on the compressor;

[0188] The emergency mute control instruction includes: controlling the transient mute of the air conditioner and controlling the soft start and stop of the night mode of the air conditioner.

[0189] In one embodiment, the operation control unit 803 further includes:

[0190] The interrupt control unit is used to control the operation of the air conditioner according to a preset interrupt priority control strategy; wherein the interrupt priority, from high to low, includes noise exceeding standard events, sleep state update events and user configuration update events.

[0191] Since the embodiments of the apparatus part correspond to the embodiments of the method part, please refer to the description of the embodiments of the method part for the embodiments of the apparatus part, and they will not be repeated here.

[0192] See also Figure 12 , Figure 12 1 is a schematic block diagram of an air conditioner 1200 provided by an embodiment of the present invention. The air conditioner 1200 is a device capable of wireless communication and wired communication.

[0193] See Figure 12The air conditioner 1200 includes a processor 1202 , a memory, and a network interface 1205 connected via a system bus 1201 , wherein the memory may include a non-volatile storage medium 1203 and an internal memory 1204 .

[0194] The non-volatile storage medium 1203 can store an operating system 12031 and a computer program 12032. When the computer program 12032 is executed, the processor 1202 can execute an air conditioner control method.

[0195] The processor 1202 is used to provide computing and control capabilities to support the operation of the entire air conditioner 1200.

[0196] The internal memory 1204 provides an environment for the operation of the computer program 12032 in the non-volatile storage medium 1203. When the computer program 12032 is executed by the processor 1202, the processor 1202 can execute an air conditioner control method.

[0197] The network interface 1205 is used to communicate with other devices through the network. Figure 12 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention, and does not constitute a limitation on the air conditioner 1200 to which the solution of the present invention is applied. The specific air conditioner 1200 may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0198] The processor 1202 is configured to run a computer program 12032 stored in a memory to implement any embodiment of the above-mentioned air conditioner control method.

[0199] It should be understood that in the embodiment of the present invention, the processor 1202 may be a central processing unit (CPU), and the processor 1202 may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0200] Those skilled in the art will appreciate that all or part of the steps in the method of the above-described embodiment can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. The computer program is executed by at least one processor in the computer system to implement the steps in the method of the above-described embodiment.

[0201] An embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon. When executed, the computer program can implement the steps provided in the above embodiment. The storage medium may include a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, an optical disk, or other medium capable of storing program code.

[0202] The various embodiments in the specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same and similar parts between the various embodiments can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of this application, several improvements and modifications can be made to this application, and these improvements and modifications also fall within the scope of protection of the claims of this application.

[0203] It should also be noted that, in this specification, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus comprising the element.

Claims

1. A method for controlling an air conditioner, characterized in that: include: Collecting the user's sleep information through radar detection technology, and using a neural network model to predict and output the sleep information to obtain the user's sleep status; The sleep state includes a wakeful state and a asleep state; Indoor noise information is collected through a multimodal noise sensing system, and voiceprint analysis is performed on the noise information to obtain a voiceprint analysis result; the voiceprint analysis result includes the operating noise of the air conditioner and the indoor ambient noise, and the ambient noise includes broadband random noise and high-frequency impulse noise; generating a control instruction based on the sleep state and the voiceprint analysis result, and controlling the operation of the air conditioner based on the control instruction; The method of collecting the user's sleep information by radar detection technology and predicting and outputting the sleep information by using a neural network model to obtain the user's sleep state includes: Inputting the sleep information into the input layer of the multi-layer perceptron model; the input layer includes three input features, namely, respiratory frequency, body movement frequency, and body movement amplitude; performing feature processing on the sleep information using a hidden layer of the multi-layer perceptron model to obtain sleep features corresponding to the sleep information; Performing probability distribution prediction on the sleep characteristics using the output layer of the multi-layer perceptron model, and outputting the sleep state based on the result of the probability distribution prediction; The step of generating a control instruction by combining the sleep state and the voiceprint analysis result, and controlling the operation of the air conditioner based on the control instruction, includes: When the sleep state changes to the awake state and the awake state lasts for a preset time threshold, a basic noise reduction control instruction is generated; the basic noise reduction control instruction includes: adjusting the current angle of the air guide plate of the air conditioner and adjusting the current frequency of the compressor; When the sleep state is a falling asleep state and the confidence level of the falling asleep state reaches a preset confidence threshold, a deep noise reduction control instruction is generated; the deep noise reduction control instruction includes: downshifting the current speed of the air conditioner fan and performing a shutdown compensation process on the compressor; When the noise information contains sudden noise and the sudden noise reaches a preset noise threshold, an emergency mute control instruction is generated; the emergency mute control instruction includes: controlling the air conditioner to perform transient mute and controlling the air conditioner's night mode soft start and stop.

2. The air conditioner control method according to claim 1, wherein: The multimodal noise sensing system is used to collect indoor noise information, and the noise information is subjected to voiceprint analysis to obtain a voiceprint analysis result, including: Collecting indoor sound signals through a microphone array and converting the sound signals into time domain waveforms; collecting a mechanical vibration signal of the air conditioner through a vibration sensor, and outputting triaxial acceleration data based on the mechanical vibration signal; Performing a fast Fourier transform on the time domain waveform to obtain corresponding first spectrum information, and performing a fast Fourier transform on the three-axis acceleration data to obtain corresponding second spectrum information; Calculating spectrum centroids for the first spectrum information and the second spectrum information respectively, and extracting pulse characteristics; The noise information is subjected to noise separation based on the calculation result of the spectrum center of gravity and the extraction result of the pulse characteristic to obtain the operating noise of the air conditioner and the indoor environmental noise; wherein the environmental noise includes broadband random noise and high-frequency pulse noise.

3. The air conditioner control method according to claim 1, wherein: Also includes: According to the following formula, the noise information is compensated using a dynamic compensation algorithm: ； in, Indicates the noise compensation value, K v Indicates the vibration compensation coefficient, K a represents the correction value of the acoustic transfer function, x represents the vibration displacement, and t represents the time.

4. The air conditioner control method according to claim 1, wherein: The step of controlling the operation of the air conditioner based on the control instruction includes: The air conditioner is operated and controlled according to a preset interrupt priority control strategy; wherein the interrupt priorities include, from high to low, noise exceeding standard events, sleep state update events, and user configuration update events.

5. An air conditioner control device, characterized in that: include: A first collecting unit is configured to collect the user's sleep information through radar detection technology, and predict and output the sleep information using a neural network model to obtain the user's sleep state; The sleep state includes a wakeful state and a asleep state; a second acquisition unit, configured to collect indoor noise information through a multimodal noise sensing system, and perform voiceprint analysis on the noise information to obtain a voiceprint analysis result; the voiceprint analysis result includes the operating noise of the air conditioner and the indoor ambient noise, wherein the ambient noise includes broadband random noise and high-frequency impulse noise; an operation control unit, configured to generate a control instruction based on the sleep state and the voiceprint analysis result, and to control the operation of the air conditioner based on the control instruction; The first acquisition unit includes: An information input unit, configured to input the sleep information into an input layer of a multi-layer perceptron model; the input layer includes three input features, namely, respiratory rate, body movement frequency, and body movement amplitude; a feature processing unit, configured to perform feature processing on the sleep information using a hidden layer of the multi-layer perceptron model to obtain sleep features corresponding to the sleep information; a probability distribution unit, configured to perform probability distribution prediction on the sleep feature using the output layer of the multi-layer perceptron model, and output the sleep state based on the result of the probability distribution prediction; The operation control unit includes: a first instruction generating unit, configured to generate a basic noise reduction control instruction when the sleep state changes to the awake state and the awake state lasts for a preset time threshold; the basic noise reduction control instruction includes: adjusting a current angle of a deflector of the air conditioner and adjusting a current frequency of a compressor; a second instruction generating unit, configured to generate a deep noise reduction control instruction when the sleep state is a falling asleep state and the confidence level of the falling asleep state reaches a preset confidence threshold; the deep noise reduction control instruction includes: downshifting a current fan speed of the air conditioner and performing a shutdown compensation process on the compressor; The third instruction generating unit is used to generate an emergency mute control instruction when there is burst noise in the noise information and the burst noise reaches a preset noise threshold; the emergency mute control instruction includes: controlling the transient mute of the air conditioner and controlling the soft start and stop of the air conditioner in night mode.

6. An air conditioner, characterized in that: The method comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the air conditioner control method according to any one of claims 1 to 4 when executing the computer program.

7. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the air conditioner control method according to any one of claims 1 to 4 is implemented.

Images