Range hood infrared signal anti-interference method

Abstract

The invention discloses a range hood infrared signal anti-interference method, which comprises the following steps: S100, signal preprocessing and spatial feature extraction: performing frequency band screening on an infrared receiving signal through a dynamic band-pass filter circuit so as to perform out-of-band noise suppression, and acquiring spatial difference features of the signal by using a dual-channel differential sampling circuit; s200, feature-enhanced blind source separation: decomposing the preprocessed mixed signal based on a blind source separation algorithm, embedding combined feature constraints of a time domain, a frequency domain and a space domain in the decomposition process, eliminating interference and extracting multi-dimensional feature parameters; s300, interference level decision and adaptive response: classifying the feature parameters through a lightweight deep learning model, identifying gesture signals and interference signals, and executing an adaptive strategy according to the interference level; the method can dynamically adapt to a complex interference environment, solves the problem that fixed frequency band filtering cannot dynamically adapt to interference, improves the signal stability, and guarantees the real-time performance of user operation.

Description

Technical Field

[0001] The present invention belongs to the technical field of smart home control, and in particular relates to a range hood infrared signal anti-interference method. Background Art

[0002] In smart home applications, infrared sensors are widely used in applications such as range hood gesture control and robot vacuum cleaner obstacle avoidance due to their low cost and fast response. However, infrared frequency band resources are limited, and infrared sensors on different devices may interfere with each other due to overlapping or similar operating frequencies, leading to false triggering or malfunctioning of devices.

[0003] Take a range hood and a robot vacuum as an example: the range hood's active infrared gesture sensing module emits infrared light through a transmitter tube. When a hand is waved at close range (5-12 cm), the directionally reflected light triggers a signal from a receiver tube. Meanwhile, the robot vacuum's infrared obstacle avoidance sensor detects obstacles using diffuse reflection, scattering the infrared light in all directions. When these two sensors operate simultaneously, two types of interference issues can arise: First, the range hood's infrared signal can be mistakenly recognized as a gesture, leading to inadvertent activation or function switching; second, the range hood's infrared signal can be misinterpreted as an obstacle by the robot vacuum, causing path planning confusion or even unexpected shutdowns.

[0004] Existing infrared signal interference prevention technologies often rely on fixed-band filtering or simple signal threshold determination. However, fixed-band filtering cannot dynamically adapt to complex interference environments, and fixed-frequency infrared transmissions are prone to persistent conflicts with other devices. Relying solely on software-based filtering algorithms can increase signal processing latency (e.g., IIR / FIR filters require point-by-point calculations), impacting user experience. With the proliferation of smart home devices, the lack of unified standards for infrared sensor frequency band planning across different brands and models has led to increasingly prominent interference issues that require urgent resolution. Summary of the Invention

[0005] The present invention provides a range hood infrared signal anti-interference method, aiming to improve the problems of existing infrared signal anti-interference technology, such as the inability to dynamically adapt to complex interference environments and the increased signal processing delay.

[0006] The present invention is a method for preventing infrared signal interference of a range hood, comprising the following steps:

[0007] S100, signal preprocessing and spatial feature extraction: The infrared received signal is frequency-band filtered using a dynamic bandpass filter circuit to suppress out-of-band noise, while a dual-channel differential sampling circuit is used to obtain the spatial difference characteristics of the signal;

[0008] S200, feature-enhanced blind source separation: Decompose the preprocessed mixed signal based on the blind source separation algorithm, embed joint feature constraints in the time domain, frequency domain, and spatial domain during the decomposition process, eliminate interference, and extract multi-dimensional feature parameters;

[0009] S300, Interference Level Decision and Adaptive Response: This step uses a lightweight deep learning model to classify feature parameters, identify gesture signals and interference signals, and implement the following strategies based on the interference level:

[0010] If there is no interference, the range hood will operate in normal mode;

[0011] If the interference is mild, enable the software filtering algorithm to suppress the noise;

[0012] If the interference is severe, the hardware-level frequency jump is triggered to the preset backup frequency.

[0013] Furthermore, the dynamic bandpass filter circuit has a digitally controlled LC filter and a switched capacitor filter. The dynamic bandpass filter circuit achieves dynamic adjustment of the center frequency of 36-40KHz through the digitally controlled LC filter and achieves precise control of the ±1KHz bandwidth through the switched capacitor filter.

[0014] Furthermore, the dual-channel differential sampling circuit is deployed with two infrared receiving tubes, and the distance between the two receiving tubes is 5-15 cm. The dual-channel differential sampling circuit adopts ADC dual-channel synchronous sampling mode and obtains spatial difference characteristics by calculating the mutual correlation coefficient of the two-channel signals.

[0015] Furthermore, in step S200, the blind source separation algorithm includes a preprocessing process, which includes signal centering and whitening processing, wherein the whitening processing is used to remove the correlation between signal dimensions and is accelerated based on the ARMCMSIS-DSP library; the iterative calculation of the blind source separation adopts fixed-point number Q15 format operation, and the number of iterations is limited to ≤10 times, so that the signal processing delay is <5ms.

[0016] Furthermore, the mixed signal model is expressed as X=AS+N, where:

[0017] X is the observation signal matrix (m×n), which is composed of the infrared signals output by the dual-channel differential sampling circuit;

[0018] A is the unknown mixing matrix (m×k), which represents the signal mixing mode;

[0019] S is the unknown source signal matrix (k×n), which contains at least gesture signal and interference signal components;

[0020] N is the noise matrix, which satisfies the requirement that the number of sensors m ≥ the number of source signals k, and only one Gaussian source signal is allowed to exist.

[0021] Furthermore, the joint feature parameters of the time domain, frequency domain and spatial domain include:

[0022] Time domain: signal zero-crossing rate, pulse duty cycle;

[0023] Frequency domain: 38kHz±2kHz sub-band energy extracted by STM32 hardware FFT;

[0024] Spatial domain: Cross-correlation coefficient of two-channel signals.

[0025] Furthermore, in step S200, interference is eliminated by the following mechanism:

[0026] S210, calculating the time-frequency-space feature parameters of each decomposition component, and performing a matching evaluation with a preset gesture signal feature template;

[0027] S220: If the characteristic parameter of a component deviates from the template threshold range, it is determined to be interference and removed;

[0028] S230: Reorganize the filtered components into a pure gesture signal and extract its multi-dimensional feature parameters.

[0029] Furthermore, the lightweight deep learning model is a 1D-CNN+BiLSTM architecture. The input includes a 100ms time domain waveform and 13-dimensional MFCC features. The model is quantized into 8-bit integers through TensorFlowLite, the Flash occupies less than 50KB, and the AI ​​coprocessor of ESP32-S3 is used to accelerate reasoning.

[0030] Furthermore, in step S300, if the feature matching degree is ≥90% and the 38kHz±2kHz sub-band energy proportion is ≥60%, it is determined to be no interference; if the feature matching degree is <70% or the sub-band energy proportion is <40%, it is determined to be severe interference, and the others are mild interference; wherein, the feature matching degree is the Softmax probability value output by the lightweight deep learning model, and the sub-band energy proportion is based on the calculation result of the STM32 hardware FFT; when it is determined to be severe interference, the hardware-level frequency jumps to the preset backup frequency while broadcasting the current working frequency through BLE, and negotiates the frequency band with other smart home devices with infrared functions. When the negotiation is abnormal, the abnormal recovery mechanism of the hardware watchdog (WWDG) and the software heartbeat packet double check is triggered; in an abnormal state, the current working frequency, interference level and other environmental parameters are automatically saved to the Flash memory, and the parameter recovery is completed within 100ms after reset.

[0031] Furthermore, the following steps are included:

[0032] S400: Generate the following output instructions according to the interference level:

[0033] Mild interference processing: fine-tune the carrier frequency ±200HZ, and the LED light flashes slowly in yellow;

[0034] Severe interference processing: shut down infrared reception for 500ms, trigger a strong vibration prompt, and report to the cloud alarm.

[0035] The range hood infrared signal anti-interference method provided by this invention dynamically adapts to complex interference environments. It avoids interfering signals through dynamic bandpass filtering and hardware-level frequency hopping, addressing the inability of fixed-band filtering to dynamically adapt to interference and improving signal stability. Furthermore, it employs multi-dimensional feature fusion and a lightweight model to accurately identify gesture signals and interference signals while avoiding the latency issues associated with traditional software filtering, ensuring real-time user operation. Other advantages of this invention are described in the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is a flow chart of a method for preventing infrared signal interference from a range hood provided in an embodiment of the present invention;

[0037] Figure 2 This is a flow chart of the mechanism for eliminating interference in step S200. DETAILED DESCRIPTION

[0038] In the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," "connect," "fixed," etc. should be understood broadly. For example, they may refer to fixed connection, detachable connection, or integration; mechanical connection or electrical connection; direct connection or indirect connection through an intermediate medium; internal communication between two components or interaction between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0039] The following is a further description with reference to the accompanying drawings and specific embodiments:

[0040] The infrared anti-interference system hardware of the range hood mainly includes:

[0041] The dual-channel differential sampling circuit is equipped with two TSOP38238 infrared receiving tubes, and the distance between the two receiving tubes is 10 cm.

[0042] Dynamic bandpass filter circuit: Integrates digitally controlled LC filter (model LTC1562) and switched capacitor filter (model MAX7401) to achieve 36-40kHz dynamic filtering;

[0043] Main control chip: ESP32-S3-WROOM-1 module, built-in 240MHz main frequency processor and AI coprocessor, supports TensorFlowLite model inference;

[0044] Storage unit: 4MB Flash memory is configured to store model parameters and system logs;

[0045] BLE communication module: Integrates the built-in BLE5.0 function of ESP32-S3 to achieve frequency band negotiation between devices.

[0046] As shown in Figure 1, a method for preventing interference of infrared signals of a range hood includes the following steps:

[0047] S100, Signal Preprocessing and Spatial Feature Extraction: A dynamic bandpass filter circuit filters the infrared received signal to suppress out-of-band noise. The dynamic bandpass filter circuit features a digitally controlled LC filter and a switched capacitor filter. The digitally controlled LC filter dynamically adjusts the center frequency between 36 and 40 kHz, while the switched capacitor filter precisely controls the ±1 kHz bandwidth. A dual-channel differential sampling circuit employs an ADC dual-channel synchronous sampling mode, simultaneously acquiring two signals at a 1 MHz sampling rate. The ADC converts these signals into digital signals, and the cross-correlation coefficient between the two channel signals is calculated to obtain spatial difference features.

[0048] S200, feature-enhanced blind source separation: The preprocessed mixed signal enters the blind source separation stage, and is decomposed based on the blind source separation algorithm. During the decomposition process, joint feature constraints in the time domain, frequency domain, and spatial domain are embedded to eliminate interference and extract multi-dimensional feature parameters.

[0049] The blind source separation algorithm includes a preprocessing step, which includes signal centering and whitening. First, centering is performed to reduce the signal mean to zero. Then, whitening is performed to remove inter-dimensional correlation. This algorithm is accelerated using the ARM CMSIS-DSP library. The iterative calculations for blind source separation utilize fixed-point Q15 format, with the number of iterations limited to ≤10, ensuring signal processing latency <5ms. The mixed signal model is represented as X = AS + N, where: X is the observed signal matrix (m × n), consisting of the infrared signals output by the dual-channel differential sampling circuit; A is the unknown mixing matrix (m × k), representing the signal mixing pattern; S is the unknown source signal matrix (k × n), containing at least the gesture signal and interference signal components; and N is the noise matrix, satisfying the requirement that the number of sensors m ≥ the number of source signals k, and only one Gaussian source signal is allowed.

[0050] The joint feature parameters of time domain, frequency domain and spatial domain include:

[0051] Time domain: signal zero-crossing rate, pulse duty cycle. The normal gesture signal zero-crossing rate is about 50-80 times / ms, and the typical value of the pulse duty cycle is 30%-70%.

[0052] Frequency domain: 38kHz±2kHz sub-band energy extracted by STM32 hardware FFT; the energy of normal gesture signals in this frequency band typically accounts for >60%.

[0053] Spatial domain: Calculate the cross-correlation coefficient of the two-channel signals. The correlation coefficient of the gesture signal is usually greater than 0.7.

[0054] like Figure 2 As shown, in step S200, interference is eliminated through the following mechanism:

[0055] S210: Calculate the time-frequency-space feature parameters of each decomposition component and evaluate the matching degree with the preset gesture signal feature template.

[0056] After the blind source separation algorithm is used to decompose the signal into multiple components, the time domain, frequency domain and spatial domain feature parameters are extracted for each component.

[0057] In the time domain, the zero-crossing rate of the signal component—the number of times the signal amplitude crosses the zero axis per unit time—is calculated by counting the positive and negative amplitude changes between adjacent sampling points. Simultaneously, the pulse duty cycle—the proportion of time the signal is in a high-level state within a complete cycle—is calculated by identifying the start and end times of the signal pulse.

[0058] In the frequency domain, the STM32 hardware FFT function is used to perform a Fast Fourier Transform on the signal components to obtain their spectral distribution. The energy of the 38kHz±2kHz subband is extracted, and the sum of the squares of the spectral amplitudes within this frequency band is used as a measure of the subband energy.

[0059] In the spatial domain, the cross-correlation coefficient of the two channel signal components is calculated based on a dual-channel differential sampling circuit. Using the discrete signal cross-correlation coefficient calculation method, the dual-channel signals are first de-meaned, and then the covariance and standard deviation are calculated to obtain the cross-correlation coefficient, which reflects the spatial consistency of the signals.

[0060] The preset gesture signal feature template is derived through statistical analysis of the characteristic parameters of a large number of normal gesture signal samples. Each feature parameter has a corresponding standard range, such as the zero-crossing rate standard range of 50-80 times / second, the pulse duty cycle standard range of 30%-70%, the 38kHz±2kHz subband energy standard range of at least 60% of the total energy, and the dual-channel cross-correlation coefficient standard range of at least 0.7. The extracted feature parameters of each decomposed component are compared with the preset template, and the deviation between the feature parameters and the template is calculated using Euclidean distance or cosine similarity to assess the degree of match.

[0061] S220: If the characteristic parameter of a component deviates from the template threshold range, it is determined to be interference and is removed.

[0062] S230, reorganize the filtered components into a pure gesture signal and extract its multidimensional feature parameters. Reorganize the signal components retained after screening according to the inverse process of the blind source separation algorithm to restore the pure gesture signal. Again extract the multidimensional feature parameters of the time domain, frequency domain and spatial domain from the reorganized pure gesture signal. These parameters will serve as input data for the subsequent lightweight deep learning model for further interference level decision and adaptive response. When extracting multidimensional feature parameters, the same calculation method and standard as in step S210 are used to ensure the consistency and accuracy of the feature parameters.

[0063] S300, Interference Level Decision and Adaptive Response: A lightweight deep learning model classifies feature parameters and identifies gesture signals and interference signals. This lightweight deep learning model uses a 1D-CNN + BiLSTM architecture, with input consisting of a 100ms time-domain waveform and 13-dimensional MFCC features. The model is quantized to 8-bit integers using TensorFlow Lite, occupies less than 50KB of Flash memory, and utilizes the ESP32-S3's AI coprocessor to accelerate inference.

[0064] If the feature match is ≥90% and the 38kHz±2kHz sub-band energy ratio is ≥60%, it is determined to be interference-free and the range hood operates normally. If the feature match is <70% or the sub-band energy ratio is <40%, it is determined to be severe interference, triggering a hardware-level frequency jump to the preset backup frequency. Other cases are mild interference, and a software filtering algorithm is used to suppress noise. The feature match is the Softmax probability value output by the lightweight deep learning model, and the sub-band energy ratio is based on the calculation results of the STM32 hardware FFT. For example, if the model output match is 92% and the sub-band energy ratio is 65%, it is determined to be interference-free and the range hood responds normally to gesture commands. If the match is 75% and the sub-band energy ratio is 55%, it is determined to be mild interference and the system uses the Kalman filter algorithm to suppress noise. When severe interference is detected, the hardware-level frequency jumps to a preset backup frequency while simultaneously broadcasting the current operating frequency via BLE. This allows for frequency band negotiation with other infrared-enabled smart home devices. If negotiation fails, such as three consecutive failures, the hardware watchdog (WWDG) and software heartbeat packet double-checking are triggered to ensure proper program operation. In these abnormal conditions, environmental parameters such as the current operating frequency and interference level are automatically saved to Flash memory. After a reset, this data can be read to quickly restore the system to its pre-announcement state within 100ms, completing parameter recovery.

[0065] S400, the range hood generates the following output instructions according to the interference level:

[0066] Mild Interference Handling: Fine-tune the carrier frequency by ±200 Hz to avoid the center frequency of the interfering signal, achieving frequency domain isolation. Simultaneously, the LED slowly flashes yellow, providing visual feedback that mild interference is present and the system has automatically adjusted the frequency. The slow flash frequency is set to 1 per second, which does not affect user operation while providing a continuous reminder of environmental anomalies.

[0067] Severe interference handling: turn off infrared reception for 500ms, cut off interference input, and reset the signal link. Trigger a strong vibration prompt, such as using the vibration motor inside the range hood (such as an eccentric motor) to generate a strong vibration with a frequency of 2Hz and an amplitude of 0.5mm, which lasts for 2-3 seconds. Compared with simple light prompts, vibration feedback can attract more user attention, especially in a noisy kitchen environment. In addition, cloud alarms are reported to achieve remote monitoring, and it is convenient for operation and maintenance personnel to conduct background analysis. If the severe interference reporting rate of a certain model of range hood in a specific area exceeds the threshold (such as 10% users / week), the system automatically triggers an OTA upgrade and pushes the optimized feature template or frequency switching strategy.

[0068] Among the four main steps of the range hood infrared signal anti-interference method provided in this application, step S100 preliminarily filters the frequency band and obtains spatial features through hardware means, step S200 performs deep separation and feature extraction at the algorithm level based on the result of step S100, step S300 makes intelligent decisions based on the features extracted in step S200 to achieve graded response, and step S400 is further execution and feedback based on the interference level decision result of step S300. Through hardware actions, user interaction and system linkage, the abstract level judgment is converted into specific anti-interference execution instructions, forming a complete "detection-analysis-decision-execution" closed loop.

[0069] In summary, the range hood infrared signal anti-interference method provided in this application can dynamically adapt to complex interference environments. By using dynamic bandpass filtering and hardware-level frequency hopping to avoid interfering signals, it solves the problem of fixed-band filtering's inability to dynamically adapt to interference and improves signal stability. Furthermore, by employing multi-dimensional feature fusion and a lightweight model, it accurately identifies gesture signals and interference signals while avoiding the latency issues associated with traditional software filtering, ensuring real-time user operation.

[0070] The above is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that the present invention is susceptible to various modifications and variations. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A range hood infrared signal anti-interference method, characterized in that: The steps include: S100, signal preprocessing and spatial feature extraction: The infrared received signal is frequency-band filtered using a dynamic bandpass filter circuit to suppress out-of-band noise, while a dual-channel differential sampling circuit is used to obtain the spatial difference characteristics of the signal; S200, feature-enhanced blind source separation: Decompose the preprocessed mixed signal based on the blind source separation algorithm, embed joint feature constraints in the time domain, frequency domain, and spatial domain during the decomposition process, eliminate interference, and extract multi-dimensional feature parameters; S300, Interference Level Decision and Adaptive Response: This step uses a lightweight deep learning model to classify feature parameters, identify gesture signals and interference signals, and implement the following strategies based on the interference level: If there is no interference, the range hood will operate in normal mode; If the interference is mild, enable the software filtering algorithm to suppress the noise; If the interference is severe, the hardware-level frequency jump is triggered to the preset backup frequency.

2. The method for preventing infrared signal interference of a range hood according to claim 1, characterized in that: The dynamic bandpass filter circuit has a digitally controlled LC filter and a switched capacitor filter. The dynamic bandpass filter circuit achieves dynamic adjustment of the center frequency of 36-40KHz through the digitally controlled LC filter and achieves precise control of the ±1KHz bandwidth through the switched capacitor filter.

3. The method for preventing infrared signal interference of a range hood according to claim 1, characterized in that: The dual-channel differential sampling circuit is deployed with two infrared receiving tubes, and the distance between the two receiving tubes is 5-15 cm. The dual-channel differential sampling circuit adopts ADC dual-channel synchronous sampling mode and obtains spatial difference characteristics by calculating the mutual correlation coefficient of the two-channel signals.

4. The method for preventing infrared signal interference of a range hood according to claim 1, characterized in that: In step S200, the blind source separation algorithm includes a preprocessing process, which includes signal centering and whitening processing. The whitening processing is used to remove the correlation between signal dimensions and is accelerated based on the ARMCMSIS-DSP library. The iterative calculation of the blind source separation adopts fixed-point number Q15 format operation, and the number of iterations is limited to ≤10 times, so that the signal processing delay is <5ms.

5. The method for preventing infrared signal interference of a range hood according to claim 4, characterized in that: The mixed signal model is expressed as X=AS+N, where: X is the observation signal matrix (m×n), which is composed of the infrared signals output by the dual-channel differential sampling circuit; A is the unknown mixing matrix (m×k), which represents the signal mixing mode; S is the unknown source signal matrix (k×n), which contains at least gesture signal and interference signal components; N is the noise matrix, which satisfies the requirement that the number of sensors m ≥ the number of source signals k, and only one Gaussian source signal is allowed to exist.

6. The method for preventing infrared signal interference of a range hood according to claim 5, characterized in that: The joint feature parameters of the time domain, frequency domain and spatial domain include: Time domain: signal zero-crossing rate, pulse duty cycle; Frequency domain: 38kHz±2kHz sub-band energy extracted by STM32 hardware FFT; Spatial domain: Cross-correlation coefficient of two-channel signals.

7. The method for preventing infrared signal interference of a range hood according to claim 6, characterized in that: In step S200, interference is eliminated through the following mechanism: S210, calculating the time-frequency-space feature parameters of each decomposition component, and performing a matching evaluation with a preset gesture signal feature template; S220: If the characteristic parameter of a component deviates from the template threshold range, it is determined to be interference and removed; S230: Reorganize the filtered components into a pure gesture signal and extract its multi-dimensional feature parameters.

8. The method for preventing infrared signal interference of a range hood according to claim 7, characterized in that: The lightweight deep learning model uses a 1D-CNN+BiLSTM architecture. Its input includes a 100ms time-domain waveform and 13-dimensional MFCC features. The model is quantized into 8-bit integers using TensorFlowLite, occupies less than 50KB of Flash memory, and uses the ESP32-S3's AI coprocessor to accelerate inference.

9. The method for preventing infrared signal interference of a range hood according to claim 8, characterized in that: In step S300, if the feature matching degree is ≥90% and the 38kHz±2kHz sub-band energy ratio is ≥60%, it is determined that there is no interference; If the feature matching degree is less than 70% or the sub-band energy proportion is less than 40%, it is judged as severe interference, and the others are mild interference; wherein, the feature matching degree is the Softmax probability value output by the lightweight deep learning model, and the sub-band energy proportion is based on the calculation result of the STM32 hardware FFT; when it is judged as severe interference, the hardware-level frequency jumps to the preset backup frequency while broadcasting the current working frequency through BLE, and negotiates the frequency band with other smart home devices with infrared functions. When the negotiation is abnormal, the abnormal recovery mechanism of the hardware watchdog (WWDG) and the software heartbeat packet double check is triggered; in the abnormal state, the current working frequency, interference level and other environmental parameters are automatically saved to the Flash memory, and the parameter recovery is completed within 100ms after reset.

10. A range hood infrared signal anti-interference method according to any one of claims 1 to 9, characterized in that: The following steps are also included: S400: Generate the following output instructions according to the interference level: Mild interference processing: fine-tune the carrier frequency ±200HZ, and the LED light flashes slowly in yellow; Severe interference processing: shut down infrared reception for 500ms, trigger a strong vibration prompt, and report to the cloud alarm.

Images