Intelligent bathroom control system supporting voice and gesture interaction

Abstract

The application belongs to the technical field of intelligent bathroom power supply and interactive control, and discloses an intelligent bathroom integrated control system supporting voice and gesture interaction, which comprises an electric energy distribution monitoring unit, a pipeline vibration sensing unit, a space state parameter acquisition unit, a central logic control unit and an electric energy output modulation unit. The pipeline vibration sensing unit detects mechanical vibration excited by fluid and provides a reference signal. The central logic control unit identifies gestures based on the correlation strength between the reference signal and an echo sequence, and then drives the electric energy output modulation unit to adjust load power parameters. The application utilizes the physical causal correlation between echoes and vibrations generated by fluid dynamics processes, eliminates recognition interference generated by water splashing through a multi-modal signal decoupling mechanism, and enhances the stability of system power distribution control on the basis of maintaining interactive response sensitivity.

Description

Technical Field

[0001] This invention relates to an integrated intelligent bathroom control system that supports voice and gesture interaction, belonging to the field of intelligent bathroom power supply and interactive control technology. Background Technology

[0002] Currently, for electrical equipment in bathroom environments, non-contact human-machine interaction using acoustic sensing and electromagnetic wave detection technologies is the mainstream approach to improve terminal management efficiency and ensure electrical safety. These systems collect user voice signals or gesture characteristics, converting them into control commands for adjusting the power of loads such as water valve actuators and heating components, thus achieving intelligent energy distribution for bathroom equipment. Existing technical solutions typically rely on microphone arrays for beamforming or millimeter-wave radar to extract point cloud features. However, the bathroom operating environment inherently presents a conflict between the co-location of signal and noise sources. When the water actuator is in operation, high-speed fluid jets generate broadband aerodynamic noise, which highly overlaps spatially with the user's voice signal, and the distance between them is usually less than 50cm. This causes traditional noise reduction algorithms based on spatial gain to physically fail. Simultaneously, the breaking and splashing process of the fluid under gravity and pressure generates Doppler micro-motion characteristics, which overlap with human hand micro-motion characteristics in the frequency domain. This aliasing of physical features makes it difficult for radar detection systems to distinguish between fluid movement and gesture commands using conventional feature extraction methods, leading to false triggering or rejection of control loops.

[0003] While the industry has attempted to introduce various optimization algorithms to assist in identification, simply relying on algorithm model parameter optimization is insufficient to handle the complex fluid dynamic interference in the bathroom environment. Chinese invention patent CN121121951A discloses an intelligent bathroom fall warning system based on an intelligent sensing system. Although it integrates millimeter-wave radar, distributed pressure-sensing mats, and other sensors, and utilizes the fruit fly algorithm to optimize the identification model and reduce the false alarm rate, its essence remains the same: backend statistical filtering and feature extraction of mixed signals. This addresses the conflict between the co-location of signal and noise sources. However, when high-pressure water jets generate Doppler clutter that highly overlaps with the frequency domain of hand gestures, relying solely on algorithm threshold adjustment or model training is insufficient to overcome the limitations of conventional methods. Frequency domain filtering removes fluid interference with similar physical characteristics, but it cannot accurately capture and respond to subtle gesture commands under strong shower noise conditions. To address these challenges, the industry has attempted to introduce deep learning models or increase the number of hardware sensors to assist in recognition. However, such approaches face resource constraints in practical engineering applications. Bathroom terminal devices typically use low-power embedded processors, and massive data processing leads to real-time latency and reduced energy efficiency, limiting the deployment of advanced algorithms. Since the water flow pattern in bathroom scenarios is affected by unstructured factors such as water pressure fluctuations, nozzle buildup, and faucet rotation angles, spatial masking methods based on fixed geometric models lack reliability when facing dynamically changing physical environments.

[0004] Therefore, the technical problem to be solved by this invention is how to achieve highly reliable decoupling from complex interference environments by utilizing the physical correlation between fluid dynamics processes and signal transmission paths without increasing the burden on system hardware, and how to construct a dynamic interactive control closed loop with self-calibration capabilities. Summary of the Invention

[0005] To address the problems mentioned in the background art, the technical solution of the present invention is as follows: A smart bathroom integrated control system supporting voice and gesture interaction, the system comprising:

[0006] The system includes a power distribution monitoring unit, a pipeline vibration sensing unit, a spatial state parameter acquisition unit, a central logic control unit, and a power output modulation unit.

[0007] The power distribution monitoring unit is used to manage the power distribution of bathroom terminal loads and monitor the current status of the output circuit in real time.

[0008] The pipeline vibration sensing unit is attached to the surface of the valve body of the controlled pipeline to detect the mechanical vibration generated when the fluid flows in the controlled pipeline and output a reference vibration signal.

[0009] The spatial state parameter acquisition unit is used to acquire millimeter-wave radar echo data in the bathroom space and extract the target echo intensity sequence at different detection distances;

[0010] The signal input terminals of the central logic control unit are connected to the pipeline vibration sensing unit and the spatial state parameter acquisition unit, respectively. They are used to calculate the Pearson correlation coefficient between the target echo intensity sequence and the reference vibration signal within a preset sliding sampling window to generate a correlation intensity value. If the correlation intensity value is lower than the preset interference identification threshold, the central logic control unit determines that there is a valid gesture in the detection area and generates a corresponding power adjustment command based on the valid gesture.

[0011] The control terminal of the power output modulation unit is connected to the central logic control unit, and its power input terminal is connected to the power distribution monitoring unit. It is used to adjust the output voltage or duty cycle according to the power adjustment command to change the power parameters supplied to the bathroom terminal load.

[0012] Preferably, when calculating the associated intensity value, the central logic control unit is used to determine the background interference benchmark based on the reference vibration signal and to identify the water flow signal characteristics in the target echo intensity sequence in combination with the fluid dynamics law of the controlled pipeline; wherein, the central logic control unit is also used to dynamically adjust the step size of the sliding sampling window according to the envelope width of the reference vibration signal.

[0013] Preferably, the central logic control unit is also used to call the stored gravitational acceleration constant as a constraint reference to establish the range-velocity image curve of the fluid particle in free fall; the central logic control unit transforms the millimeter-wave radar echo data to range-Doppler space and performs notch filtering based on the range-velocity image curve.

[0014] Preferably, the central logic control unit calculates the correlation strength value in real time according to the following formula. : ,in, For the target echo intensity sequence, For reference vibration signal, Let the covariance of the two be , The standard deviation of the target echo intensity sequence. The standard deviation of the reference vibration signal.

[0015] Preferably, the spatial state parameter acquisition unit also stores the detection mask parameters; the central logic control unit is also used to extract the transient flow head motion trajectory at the moment the fluid output is turned on, fit the jet characteristic angle under the current working condition, and update the detection mask parameters according to the jet characteristic angle to perform dynamic calibration of the detection area.

[0016] Preferably, the central logic control unit is also used to calculate the frequency domain spectral entropy of the reference vibration signal; when the frequency domain spectral entropy exceeds the preset turbulence judgment limit, the central logic control unit triggers the dimensionality reduction processing of the power regulation logic and switches the interaction mode to the preset finite command set matching mode.

[0017] Preferably, the pipeline vibration sensing unit includes a piezoelectric sensor, which is attached to the housing of a ball valve or ceramic core valve in the controlled pipeline; the spatial state parameter acquisition unit includes a millimeter-wave radar sensor, which is orthogonally arranged with the piezoelectric sensor in the spatial arrangement.

[0018] Preferably, the power distribution monitoring unit is used to detect the loop impedance of each bathroom terminal load in real time, and, in conjunction with the interaction priority determined by the central logic control unit, to perform dynamic power limiting or load switching on the bathroom terminal load.

[0019] Preferably, when updating the detection mask parameters, the central logic control unit is used to delineate a signal shielding area in the detection space according to the jet characteristic angle, and to perform zeroing processing on the echo energy in the signal shielding area.

[0020] Preferably, the power distribution monitoring unit is also used to cut off the auxiliary power supply circuit of the power output modulation unit and switch the central logic control unit to a low-power standby mode with a standby power of less than 5mW when no reference vibration signal is detected and the target echo intensity sequence is in a preset no-load range.

[0021] Compared with the prior art, the beneficial effects of the present invention are:

[0022] 1. In the integrated control of intelligent bathrooms, an adaptive cancellation processing mechanism is constructed by utilizing the strong correlation between structurally transmitted vibration signals and water flow aerodynamic noise at their physical origins, as well as the strong orthogonality between vibration signals and air-transmitted speech signals along their propagation paths. By calculating the transmission mapping relationship from fluid mechanical vibration to aerodynamic noise in real time, the non-stationary water flow noise component is accurately removed from the mixed audio signal without damaging the speech characteristics. Compared with traditional noise reduction methods that rely on spatial separation or spectral statistical characteristics, this solution solves the bottleneck of command extraction distortion in controlled equipment under strong self-noise and co-location interference conditions by introducing a bypass physical reference source, ensuring the effective recognition of control commands in extreme sound field environments.

[0023] 2. Based on the principle of acoustic impedance coupling modulation between fluid and pipeline structure, the transient spectral shift of pipeline vibration signal generated when the user intervenes in the fluid is monitored, and the physical change of fluid impedance load is translated into an interactive ready interruption signal. This path transforms the uncertain acoustic wake-up process into a deterministic physical tactile perception, enabling the system to bypass the wake word detection link and directly respond to voice commands in high-decibel background noise environments. This collaborative mechanism achieves the dual functions of noise reduction reference and interactive triggering using the same sensor without adding additional sensing hardware, eliminating the recognition blind spot and response delay in the interactive initiation stage.

[0024] 3. By introducing the constant of gravitational acceleration as a kinematic constraint benchmark, a distance-velocity correlation curve characterizing fluid particles in free fall is constructed. The system maps radar echo data to the distance-Doppler feature space and performs dynamic notch filtering based on physical laws to achieve an essential distinction between hand gestures and random water splashes. This judgment logic based on physical constants rather than absolute spatial coordinates makes the control system naturally adaptable to mechanical posture changes such as the rotation and pulling of the faucet, and can maintain the stability of operation in complex interactive spaces without the need to add angle sensors. Attached Figure Description

[0025] Figure 1 This is a diagram showing the interactive control logic and power regulation signal flow of the system of the present invention;

[0026] Figure 2 This is a topology diagram of the hardware modules and sensor spatial deployment of the system of the present invention. Detailed Implementation

[0027] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0028] A smart bathroom integrated control system supporting voice and gesture interaction, the system includes:

[0029] The system includes a power distribution monitoring unit, a pipeline vibration sensing unit, a spatial state parameter acquisition unit, a central logic control unit, and a power output modulation unit.

[0030] The power distribution monitoring unit is used to manage the power distribution of bathroom terminal loads and monitor the current status of the output circuit in real time.

[0031] The pipeline vibration sensing unit is attached to the surface of the valve body of the controlled pipeline to detect the mechanical vibration generated when the fluid flows in the controlled pipeline and output a reference vibration signal.

[0032] The spatial state parameter acquisition unit is used to acquire millimeter-wave radar echo data in the bathroom space and extract the target echo intensity sequence at different detection distances;

[0033] The signal input terminals of the central logic control unit are connected to the pipeline vibration sensing unit and the spatial state parameter acquisition unit, respectively. They are used to calculate the Pearson correlation coefficient between the target echo intensity sequence and the reference vibration signal within a preset sliding sampling window to generate a correlation intensity value. If the correlation intensity value is lower than the preset interference identification threshold, the central logic control unit determines that there is a valid gesture in the detection area and generates a corresponding power adjustment command based on the valid gesture.

[0034] The control terminal of the power output modulation unit is connected to the central logic control unit, and its power input terminal is connected to the power distribution monitoring unit. It is used to adjust the output voltage or duty cycle according to the power adjustment command to change the power parameters supplied to the bathroom terminal load.

[0035] Preferably, when calculating the associated intensity value, the central logic control unit is used to determine the background interference benchmark based on the reference vibration signal and to identify the water flow signal characteristics in the target echo intensity sequence in combination with the fluid dynamics law of the controlled pipeline; wherein, the central logic control unit is also used to dynamically adjust the step size of the sliding sampling window according to the envelope width of the reference vibration signal.

[0036] Preferably, the central logic control unit is also used to call the stored gravitational acceleration constant as a constraint reference to establish the range-velocity image curve of the fluid particle in free fall; the central logic control unit transforms the millimeter-wave radar echo data to range-Doppler space and performs notch filtering based on the range-velocity image curve.

[0037] Preferably, the central logic control unit calculates the correlation strength value in real time according to the following formula. : ,in, For the target echo intensity sequence, For reference vibration signal, Let the covariance of the two be , The standard deviation of the target echo intensity sequence. The standard deviation of the reference vibration signal.

[0038] Preferably, the spatial state parameter acquisition unit also stores the detection mask parameters; the central logic control unit is also used to extract the transient flow head motion trajectory at the moment the fluid output is turned on, fit the jet characteristic angle under the current working condition, and update the detection mask parameters according to the jet characteristic angle to perform dynamic calibration of the detection area.

[0039] Preferably, the central logic control unit is also used to calculate the frequency domain spectral entropy of the reference vibration signal; when the frequency domain spectral entropy exceeds the preset turbulence judgment limit, the central logic control unit triggers the dimensionality reduction processing of the power regulation logic and switches the interaction mode to the preset finite command set matching mode.

[0040] Preferably, the pipeline vibration sensing unit includes a piezoelectric sensor, which is attached to the housing of a ball valve or ceramic core valve in the controlled pipeline; the spatial state parameter acquisition unit includes a millimeter-wave radar sensor, which is orthogonally arranged with the piezoelectric sensor in the spatial arrangement.

[0041] Preferably, the power distribution monitoring unit is used to detect the loop impedance of each bathroom terminal load in real time, and, in conjunction with the interaction priority determined by the central logic control unit, to perform dynamic power limiting or load switching on the bathroom terminal load.

[0042] Preferably, when updating the detection mask parameters, the central logic control unit is used to delineate a signal shielding area in the detection space according to the jet characteristic angle, and to perform zeroing processing on the echo energy in the signal shielding area.

[0043] Preferably, the power distribution monitoring unit is also used to cut off the auxiliary power supply circuit of the power output modulation unit and switch the central logic control unit to a low-power standby mode with a standby power of less than 5mW when no reference vibration signal is detected and the target echo intensity sequence is in a preset no-load range.

[0044] Example 1: In a typical smart bathroom application scenario involving high-velocity fluid control and human-machine interaction, the system faces extreme conditions where signal and noise sources highly overlap and coexist dynamically in physical space. User gestures often need to be performed within a high-frequency splash zone very close to the water outlet, accompanied by high-intensity broadband aerodynamic noise generated by the high-speed fluid flow within the pipes. To address these conditions, the system initiates an initialization program. The power distribution monitoring unit monitors and maintains the power supply stability of each functional module in real time. The pipe vibration sensing unit, attached to the surface of the controlled pipe valve, responds immediately upon fluid initiation, continuously acquiring the mechanical vibration waveforms generated by the fluid dynamics process and outputting a reference vibration signal as an interference benchmark. Simultaneously, the spatial state parameter acquisition unit scans the bathroom space at a preset frame rate, acquiring millimeter-wave radar echo data containing fluid motion clutter and potential user gesture characteristics, and extracting the target echo intensity sequence within different detection distance units. At the physical level, the synchronous acquisition of heterogeneous signals was completed. For determining the interference identification threshold, an adaptive calibration logic based on on-site statistics was adopted: upon initial system power-on or environmental reset, the control valve was in a fully open state to maintain stable fluid injection. Synchronous signal data for N sliding windows were continuously collected, with N ranging from 50 to 100 to satisfy statistical significance. The central logic control unit calculated the Pearson correlation coefficient for each window under pure fluid conditions and constructed a coefficient distribution histogram. The value corresponding to the 95th percentile of the histogram was superimposed with a safety margin of 0.05 as the final fixed interference identification threshold. Here, the Pearson correlation coefficient is a dimensionless value between -1 and +1. If the statistically obtained 95th percentile value is 0.70, the system will ultimately set the interference identification threshold to 0.75. This linear increment of 0.05 is used to construct a statistically reliable confidence interval, preventing edge misjudgments caused by sensor thermal noise. Based on the measured data of the current pipeline material and water pressure environment, the nonlinear error introduced by the acoustic impedance difference in the installation environment is eliminated.

[0045] The central logic control unit performs the core cross-modal signal decoupling operation. Utilizing the physical causal relationship between fluid-induced mechanical vibrations and radar echo clutter generated by the fluid, it performs time-domain correlation analysis on the acquired multimodal data within a preset sliding sampling window. Prior to this, the system performs time-domain alignment preprocessing for heterogeneous signals: the original sampling frequency of the pipeline vibration sensing unit is set to 12800Hz, and the central logic control unit calculates the root mean square amplitude of the vibration signal within this time period every 20 milliseconds, thus outputting a vibration envelope sequence with a frequency of 50Hz; simultaneously, the radar frame rate of the spatial state parameter acquisition unit is synchronously set to 50Hz. The central logic control unit aligns the above two to the same time axis and performs correlation analysis, specifically calculating the Pearson correlation coefficient ρ(r) between the target echo intensity sequence and the reference vibration signal in real time according to the following formula: Where r represents the radar's detection range cell index, For the target echo intensity sequence, For reference vibration signal, Let the covariance of the two be , The standard deviation of the target echo intensity sequence. To reference the standard deviation of the vibration signal, and considering the numerical anomaly boundary of the formula for calculating the associated intensity value ρ(r), the central logic control unit is configured with arithmetic protection logic: before performing the division operation between covariance and standard deviation, the standard deviation of the target echo intensity sequence is detected in real time. Standard deviation of reference vibration signal If any value is lower than the preset system noise floor benchmark (i.e., 0.1% of the sensor range), it indicates that the detection area is in a silent state or the sensor output amplitude is insufficient to support effective statistics. The correlation strength value ρ(r) is set to 0, and subsequent division operations are not performed to avoid the risk of calculation overflow or logic interruption caused by the denominator approaching zero. This ensures the stability of the program under intermittent water flow or low water pressure conditions in the control loop. Based on the above calculation results, the central logic control unit executes dynamic gating decision logic. For distance units where the calculated correlation strength value ρ(r) is higher than the preset interference identification threshold, the system determines that the radar echo at that location is mainly caused by fluid splashing or flow, and then marks it as an interference area and suppresses it. For areas where the correlation strength value is lower than the threshold, the system determines that its echo originates from user hand gestures independent of pipeline vibration. In this way, the system accurately locks the trajectory characteristics of effective gestures in the complex fluid interference background and generates corresponding power adjustment commands to drive the power output modulation unit to adjust the output voltage or duty cycle provided to the bathroom terminal load.

[0046] Example 2: To verify the anti-interference performance and engineering effectiveness of the technical solution of the present invention under extreme conditions where the signal source and noise source overlap, a verification platform simulating a real shower environment was built. Located in a semi-anechoic chamber, it was equipped with a standard ceramic valve core shower pipe and integrated a piezoelectric vibration sensor with a sensitivity of 100mV / g and a frequency response range of 10Hz to 10kHz, attached to the valve body surface. A millimeter-wave radar module with an operating frequency of 60GHz was fixed 0.4 meters directly above the water outlet to cover the water flow point and the user interaction area. The background conditions set for the experiment were: The pipeline water pressure was adjusted to 0.35 MPa, causing the nozzle to generate continuous turbulent jets and form a high-density random water droplet splash on the radar detection cross section. Measurements taken near the outlet using a standard sound level meter showed an average sound pressure level of 72 dB for the hydrodynamic noise. Under these conditions, two experimental groups were set up for simultaneous comparative testing: the control group used a constant false alarm rate (CFAR) detection algorithm that relied solely on radar echo amplitude characteristics; the experimental group used the anti-interference control scheme based on multimodal correlation analysis of this invention, where the central logic control unit was configured to perform Pearson correlation coefficient calculations and a sliding sampling window. The initial setting is 50ms.

[0047] In the first phase of the baseline fluid noise test, the system was in a pure water jet state without human intervention. At this time, the raw echo data (input data) received by the radar showed that there was a continuous clutter signal with random amplitude fluctuations within a range of 0.3 meters to 0.5 meters from the target cell, with an average signal-to-noise ratio (SNR) of only -5dB. At the same time, the reference signal collected by the pipeline vibration sensor showed a quasi-periodic envelope characteristic consistent with the fluid turbulence pulsation, with its main frequency concentrated in the range of 150Hz to 300Hz. The detection logic of the control group showed an extremely high false trigger rate under this condition. Because the echo intensity generated by water droplet splashing exceeded the preset static detection threshold multiple times in a short period of time, the system erroneously generated 12 gesture trigger commands within a 1-minute continuous test period. In the second phase of the core processing logic verification, the experimental group initiated cross-modal correlation calculation, and the central logic control unit extracted the echo intensity sequence within the radar target range cell in real time. envelope of reference vibration signal Data monitoring shows that in the pure water flow area, although the radar echo intensity fluctuates significantly, its fluctuation trend exhibits a high degree of temporal synchronization with the pipeline vibration envelope. The calculated Pearson correlation coefficient ρ remains stable within the range of 0.85 to 0.92, far exceeding the preset interference identification threshold of 0.6. Based on this, the system correctly identifies the signal in this area as fluid interference and suppresses it. A simulated hand gesture is introduced (using a standard metal sphere with a radar cross-section of 0.01 square meters to simulate a hand, cutting laterally into the water flow area at a speed of 0.5 m / s). Key intermediate data generated at this time indicates that the radar echo intensity... A drastic change occurs due to the intervention of the target, and the pipeline vibration signal... It remains controlled only by the water flow and maintains its original rhythm. The temporal envelope correlation between the two quickly decouples, and the calculated correlation coefficient ρ drops sharply from 0.88 to 0.15 within 30ms.

[0048] In the final performance evaluation of the third stage, based on the differences in the correlation coefficients mentioned above, the dynamic gating logic of the experimental group was triggered. Despite the presence of strong fluid clutter in the background, the system was still able to lock onto the target at 45ms after the gesture input, based on the criterion of ρ < 0.6, and output an effective power adjustment command. The final statistical results showed that in 100 consecutive gesture tests under high-pressure water flow, the recognition accuracy of the control group was only 42%, accompanied by high-frequency false alarms; while the recognition accuracy of the experimental group reached 96%, and the false alarm rate under pure water flow interference was 0%, confirming the effectiveness of this scheme in decoupling co-frequency and co-position interference at the physical level, and providing a basis for determining the sliding sampling window. To determine the optimal range of values, parameter gradient experiments were conducted. When set to 10ms, due to the small number of sample points, the calculated correlation coefficient ρ is affected by random noise, resulting in a variance of 0.2, which leads to unstable flickering decisions in the system; when When the time was increased to 100ms, although the stability of the correlation coefficient improved (variance decreased to 0.02), the system response latency exceeded 150ms, causing the initial segment of rapid gestures to be averaged and thus missed. Experimental results show that when... When the time is selected within the range of 40ms to 60ms, the system achieves the best engineering balance between interference suppression ratio and interaction response speed. At this time, the variance of the correlation coefficient is less than 0.05, and the system response delay is controlled within 80ms.

[0049] Example 3: In the scenario of adaptive system optimization and boundary condition defense for variable water supply network pressure, the technical solution of this invention faces the challenge of non-stationarity of fluid dynamic characteristics caused by hydraulic fluctuations, especially at the secondary water supply terminal of high-rise buildings. The transient jump in network pressure will cause the drift of the fluid turbulence main frequency and the nonlinear distortion of the water droplet splash velocity distribution, resulting in phase mismatch or signal-to-noise ratio deterioration in the correlation coefficient calculation of traditional correlation analysis algorithms based on fixed parameter models. In response to the uncertainties in the above-mentioned engineering implementation, the central logic control unit initiates feedforward calibration logic based on physical constraints, executes the gravity ballistic filtering procedure for water flow splash clutter, that is, constructs the distance and velocity mapping relationship of fluid particles in free fall state based on the determined Newtonian laws of mechanics, calls the pre-stored gravitational acceleration constant g as the absolute physical reference, and combines the radar installation height information to construct a theoretical water droplet splash curve in the range-Doppler space. For any detection range unit r, the corresponding theoretical Doppler velocity of the water droplet is... Follow the physical formula The constraints, among which The central logic control unit generates a notch mask with a preset Doppler tolerance width, which is the vertical coordinate origin of the water outlet. The radar echo energy falling within the range of the mask is subjected to joint zeroing in the spatial and frequency domains, thereby eliminating non-rigid water droplet interference signals that conform to the law of free fall and retaining the user gesture signal characteristics that do not conform to the law of gravity motion.

[0050] Based on this, in order to solve the problem of vibration frequency drift caused by changes in flow velocity, an adaptive frequency locking procedure with a sliding window was implemented. The central logic control unit performed a real-time fast Fourier transform (FFT) on the reference vibration signal output by the pipeline vibration sensing unit to extract the vibration dominant frequency characteristic value under the current fluid conditions. Based on this, the system dynamically calculates the optimal integral period to match it, and forces the sliding sampling window required for the Pearson correlation coefficient calculation according to the control law. Set to an integer multiple of the fundamental frequency period of the vibration signal, that is, satisfy... The relationship is defined by k, where k is the smallest positive integer coefficient to ensure statistical significance. Through this frequency following mechanism, regardless of the fluctuation of the pipeline pressure, the system can always ensure that the reference signal and the radar echo signal contain the complete energy fluctuation cycle within the time window, eliminating the spectral leakage and correlation calculation errors caused by the window truncation effect. Finally, the signal filtered by the above physical model and adaptive time-domain locking enters the decision layer, enabling the system to maintain a suppression ratio of more than 25dB against interference signals during the drastic fluctuation of pipeline pressure from 0.1MPa to 0.6MPa. This ensures that the control commands received by the power output modulation unit are entirely derived from deterministic gesture logic, rather than random disturbances from environmental noise.

[0051] Example 4: To ensure the engineering reproducibility and stability of the multimodal decoupling system of this invention from the source, and to eliminate potential implementation uncertainties, this example supplements the disclosure of a standardized system initialization calibration procedure. This procedure is used to determine key operating parameters under different installation environments and equipment batches, execute an offline database construction process targeting pipeline vibration baseline and radar clutter characteristics, and enter self-learning mode when the system is first powered on or when a significant environmental change is detected. At this time, the system closes the outlet valve and prompts the user to leave the detection area. The system collects at least 60 seconds of static environmental vibration noise data through the pipeline vibration sensing unit and calculates its amplitude distribution histogram to determine the environmental vibration noise floor. The system automatically opens the outlet valve to its maximum opening and runs continuously for no less than 120 seconds, simultaneously collecting vibration signals and radar echo data excited by steady-state fluid. By performing power spectral density analysis on the vibration signals, the system extracts the dominant frequency range caused by fluid turbulence. and the average vibration energy This will serve as the basis for subsequent dynamic adjustments to the sliding window. The physical reference standard.

[0052] After completing the aforementioned basic data acquisition, the system executes the on-site adaptive calibration logic for the interference identification threshold. The central logic control unit calls the synchronous data acquired during the aforementioned steady-state fluid testing phase. Without any human intervention, it traverses the preset threshold search space, calculates the Pearson correlation coefficient distribution within each detection range unit in real time, and calculates the false alarm rate under pure fluid interference, i.e., the probability that the correlation coefficient is lower than a certain candidate threshold. Based on the Neyman-Pearson criterion, and under the constraint that the false alarm rate is lower than 0.1%, the corresponding critical value of the correlation coefficient is selected as the optimal interference identification threshold for this specific operating condition. Through this automated calibration process, the system can automatically adapt to the fluid-structure coupling characteristics of different pipe materials and under different water pressure conditions, ensuring the stable operation of the interference suppression algorithm in diverse engineering sites and eliminating the technical risks of relying on empirical values.

[0053] Example 5: Addressing the risk of signal mismatch in smart bathroom systems caused by relative deviations between the radar detection coordinate system and the physical pipeline spatial position in non-standard installation environments, as well as the potential system failure due to deterioration of sensor coupling during long-term operation, this example constructs a standardized engineering procedure that includes spatial image self-calibration and operational integrity monitoring. Upon completion of the physical installation and initial activation of the fluid control function, the central logic control unit executes an automated spatial reference anchoring procedure. This procedure does not rely on preset fixed distance parameters but actively instructs the power output modulation unit to open the valve to full open. During the steady-state fluid jet cycle, it performs a comprehensive scan of the full-range range-Doppler heatmap output by the spatial state parameter acquisition unit, calculating the radar echo sequence and pipeline reference vibration signal at each distance unit r. The cross-correlation function, based on the physical causality law, is used to calculate the spatial distribution vector of the correlation coefficient. The global maximum index in the data is identified as the physical anchor point of the fluid jet source. That is, satisfying Then, using this physical anchor point as a reference, and based on the relative geometric model pre-stored in non-volatile memory, a dynamic three-dimensional attention window covering the user's gesture interaction area is automatically generated through coordinate transformation, thereby eliminating spatial alignment deviations caused by installation height or angle errors.

[0054] Based on this, to address the spectral aliasing black box problem in the target echo intensity sequence extraction process, a signal demodulation algorithm based on Doppler energy integration was defined. The spatial state parameter acquisition unit, in the extraction... Instead of simply extracting the energy peak within a range cell, an adaptive spectral mask based on constant false alarm rate (CFAR) is applied. The stopband range of this mask is determined by the theoretical droplet velocity range calculated from the aforementioned gravity ballistic model. This mask is used to filter out the main clutter components that travel at the same velocity as the fluid. The specific fitting and updating logic is as follows: The system extracts the Doppler velocity peak points of the first 5 range cells within a range of 0.1 meters to 0.3 meters from the outlet. The least squares method is used to perform linear regression on these 5 discrete points. The slope of the resulting line is the jet characteristic angle. Centered on this slope, all cells within a range of ±15 degrees in the Doppler frequency domain are marked as signal shielding areas. The corresponding spectral mask coefficients are forcibly set to zero. The system performs incoherent accumulation of the remaining Doppler frequency energy within the mask passband and generates an intensity sequence according to the following logic: ,in To filter out the effective Doppler passband after removing the principal velocity components of the fluid, To focus on the time-frequency signal spectrum matrix within the window, this step ensures that the radar signal input to the correlation analysis module contains only the energy of outliers with non-gravity motion characteristics, thus improving the signal-to-noise ratio from the signal source.

[0055] Finally, the system incorporates a runtime fault self-healing mechanism based on logic interlocks to address boundary anomalies such as sensor detachment or radar obstruction. The central logic control unit performs a state consistency check at the end of each control cycle: if the power distribution monitoring unit reports normal valve drive current, but the average energy output of the pipeline vibration sensing unit is below the normal value... Continuously below the preset minimum fluid excitation threshold The system determines that the vibration sensor is decoupled or the water supply is interrupted, and automatically switches to pure radar Doppler detection mode to maintain basic interaction capabilities; if the vibration signal is normal, but the physical anchor point... Correlation coefficient at location Below the system limit within N consecutive sliding windows The system determines that the radar antenna is blocked by a foreign object or that the signal link is faulty, and then triggers a fault alarm command and locks the high-risk automatic water discharge function.

[0056] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0057] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A smart bathroom integrated control system supporting voice and gesture interaction, characterized in that, The system includes: The system includes a power distribution monitoring unit, a pipeline vibration sensing unit, a spatial state parameter acquisition unit, a central logic control unit, and a power output modulation unit. The power distribution monitoring unit is used to manage the power distribution of bathroom terminal loads and monitor the current status of the output circuit in real time. The pipeline vibration sensing unit is attached to the surface of the valve body of the controlled pipeline to detect the mechanical vibration generated when the fluid flows in the controlled pipeline and output a reference vibration signal. The spatial state parameter acquisition unit is used to acquire millimeter-wave radar echo data in the bathroom space and extract the target echo intensity sequence at different detection distances; The signal input terminals of the central logic control unit are connected to the pipeline vibration sensing unit and the spatial state parameter acquisition unit, respectively, to calculate the Pearson correlation coefficient between the target echo intensity sequence and the reference vibration signal within a preset sliding sampling window, so as to generate the correlation intensity value. If the correlation strength value is lower than the preset interference recognition threshold, the central logic control unit determines that there is a valid gesture in the detection area and generates a corresponding power adjustment command based on the valid gesture. The control terminal of the power output modulation unit is connected to the central logic control unit, and its power input terminal is connected to the power distribution monitoring unit. It is used to adjust the output voltage or duty cycle according to the power adjustment command to change the power parameters supplied to the bathroom terminal load.

2. The intelligent bathroom integrated control system supporting voice and gesture interaction according to claim 1, characterized in that, When calculating the correlation intensity value, the central logic control unit is used to determine the background interference benchmark based on the reference vibration signal and to identify the water flow signal characteristics in the target echo intensity sequence in combination with the fluid dynamics law of the controlled pipeline. The central logic control unit is also used to dynamically adjust the step size of the sliding sampling window according to the envelope width of the reference vibration signal.

3. The intelligent bathroom integrated control system supporting voice and gesture interaction according to claim 1, characterized in that, The central logic control unit is also used to call the stored gravitational acceleration constant as a constraint reference to establish the range-velocity image curve of the fluid particle in free fall; the central logic control unit transforms the millimeter-wave radar echo data to range-Doppler space and performs notch filtering based on the range-velocity image curve.

4. The intelligent bathroom integrated control system supporting voice and gesture interaction according to claim 1, characterized in that, The central logic control unit calculates the correlation strength value in real time according to the following formula. : ,in, For the target echo intensity sequence, For reference vibration signal, Let the covariance of the two be , The standard deviation of the target echo intensity sequence. The standard deviation of the reference vibration signal.

5. The intelligent bathroom integrated control system supporting voice and gesture interaction according to claim 1, characterized in that, The spatial state parameter acquisition unit also stores the detection mask parameters; the central logic control unit is also used to extract the transient flow head motion trajectory at the moment the fluid output is turned on, fit the jet characteristic angle under the current working condition, and update the detection mask parameters according to the jet characteristic angle to perform dynamic calibration of the detection area.

6. The intelligent bathroom integrated control system supporting voice and gesture interaction according to claim 1, characterized in that, The central logic control unit is also used to calculate the frequency domain spectral entropy of the reference vibration signal; When the frequency domain spectral entropy exceeds the preset turbulence judgment limit, the central logic control unit triggers the dimensionality reduction processing of the power regulation logic, switching the interaction mode to the preset finite command set matching mode.

7. The intelligent bathroom integrated control system supporting voice and gesture interaction according to claim 1, characterized in that, The pipeline vibration sensing unit includes a piezoelectric sensor, which is attached to the housing of a ball valve or ceramic core valve in the controlled pipeline; the spatial state parameter acquisition unit includes a millimeter-wave radar sensor, which is orthogonally arranged with the piezoelectric sensor in the spatial arrangement.

8. The intelligent bathroom integrated control system supporting voice and gesture interaction according to claim 1, characterized in that, The power distribution monitoring unit is used to detect the loop impedance of each bathroom terminal load in real time, and, in conjunction with the interaction priority determined by the central logic control unit, to perform dynamic power limiting or load switching on the bathroom terminal load.

9. A smart bathroom integrated control system supporting voice and gesture interaction according to claim 5, characterized in that, When updating the detection mask parameters, the central logic control unit is used to delineate the signal shielding area in the detection space according to the jet characteristic angle, and to perform zeroing processing on the echo energy in the signal shielding area.

Images