An intelligent system for dynamically tracking noise sources of a tower and directionally reducing noise

Abstract

The application discloses an intelligent system for dynamically tracking a noise source of a tower body and directional noise reduction, and belongs to the technical field of cooling tower noise reduction. The intelligent system comprises a microphone array circumferentially distributed at the bottom of the tower, an audio denoising module, an audio tracking and positioning module, a directional noise reduction module and an audio synthesis module. The microphone array collects multiple audio signals, and the audio signals are filtered, windowed, Fourier transformed, spectrum subtracted and inverse Fourier transformed to obtain denoised audio signals. The audio tracking and positioning module extracts an envelope from the denoised audio signals and performs cross-correlation calculation to obtain a time delay difference of each microphone pair. Under near-field conditions, a nonlinear equation set is constructed to solve the three-dimensional coordinates of the noise source, and dynamic tracking is realized. The directional noise reduction module positions a sector according to the coordinates of the noise source, calculates the root mean square value of the denoised audio signals in the sector as the noise intensity, and controls the opening degree of the corresponding air inlet louvers to be reduced when the noise intensity exceeds a threshold value, so that directional noise reduction is realized. The application can adapt to the dynamic drift of noise hotspots, and can balance noise reduction effect and cooling efficiency.

Description

Technical Field

[0001] This invention relates to the field of cooling tower noise reduction technology, and more specifically, to an intelligent system that dynamically tracks the noise source of the cooling tower and reduces noise in a targeted manner. Background Technology

[0002] A cooling tower is a heat dissipation device that facilitates heat exchange through direct contact between air and water. It is widely used in industrial and civil applications such as thermal power plants, chemical plants, and central air conditioning systems. Its working principle involves spraying circulating hot water onto the surface of the packing material inside the tower. Ventilation allows for full contact between the air and the hot water, lowering the water temperature through evaporation and heat transfer. Cooling towers are typically hyperbolic or square in shape, with an annular air inlet at the bottom. The bottom area of ​​the tower contains packing material, water distribution pipes, and a water collection tray. Cooling towers generate significant noise during operation, causing disturbance to the surrounding environment and equipment operators. This noise not only affects the quality of life for residents but may also lead to excessive noise levels at factory boundaries, forcing companies to implement noise reduction measures. Because cooling towers often operate continuously, their noise problem is persistent and complex.

[0003] During cooling tower operation, multiple noise sources exist in the bottom area: aerodynamic sound is generated by the interaction of airflow entering through the inlet with the tower structure and packing; impact noise is generated by water droplets falling from the water spray system hitting the packing layer and the bottom water collection tray; in addition, airflow pulsation and water flow impact may also excite the packing, support components, and other thin-walled structures to generate vibration radiation noise. To reduce cooling tower noise, existing technologies add sound-absorbing materials to the louvers of all air inlets. These solutions generally adopt a globally uniform configuration, which cannot provide independent and differentiated control based on the noise differences in each area, and is highly dependent on the stability of the operating conditions. However, natural wind is not a steady-state artificial wind; changes in wind conditions will change the airflow distribution characteristics at the bottom of the tower; the temperature of the hot water to be cooled is not fixed and will change arbitrarily. Water temperature fluctuations will cause changes in the stiffness of the packing and the damping of its internal support springs, and the vibration response characteristics of the tower structure will also change. Therefore, the high noise sources inside the tower are not fixed, but dynamically drift with changes in environmental wind conditions and spray side operating conditions.

[0004] While directly installing sound-absorbing material on the louvers can provide some passive noise reduction, it significantly increases air intake resistance, reduces cooling efficiency, and increases fan energy consumption. Sound-absorbing materials are prone to dust and moisture accumulation, which can lead to narrowing of air ducts or reduction of sound-absorbing performance, resulting in high maintenance costs. Once sound-absorbing materials are fixed to louvers, their noise reduction capability is static and cannot be adjusted according to real-time wind direction or the movement of noise sources. In low-noise areas, noise reduction configurations are often excessive, resulting in unnecessary increases in wind resistance and energy consumption; while in high-noise areas, noise reduction configurations may be insufficient, leading to substandard overall noise reduction effects.

[0005] Therefore, there is an urgent need for an intelligent system that can target high-noise locations for noise reduction, in order to adapt to the complex and ever-changing operating environment of cooling towers and achieve synergistic optimization of precise noise reduction and energy-saving operation. Summary of the Invention

[0006] The purpose of this invention is to provide an intelligent system that dynamically tracks tower noise sources and performs targeted noise reduction, in order to solve the aforementioned technical problems.

[0007] To achieve the above objectives, the present invention provides the following technical solution: An intelligent system for dynamically tracking and directionally reducing noise sources in a tower, the system's actuators include a microphone array, and the system includes an audio noise reduction module, an audio tracking and positioning module, an audio synthesis module, and a directional noise reduction module; Several microphones of the microphone array are distributed circumferentially around the cooling tower axis at the bottom of the tower and are positioned opposite to several air inlets distributed circumferentially around the cooling tower. The water spray area at the bottom of the tower is divided into several uniform sectors. The system establishes a three-dimensional coordinate system with the center of the microphone array as the origin. The microphone array collects audio data during normal operation of the cooling tower as additive noise to be removed during spectral subtraction. During the operation of the cooling tower, the microphone array collects audio data and transmits it to the audio noise reduction module. The audio denoising module converts the multi-channel audio data collected by the microphone array into multi-channel digital audio data through analog-to-digital conversion. The denoised audio data is then obtained by sequentially filtering, windowing, Fourier transform, spectral subtraction, and inverse Fourier transform. The denoised audio data is then input into the audio tracking and positioning module, the directional noise reduction module, and the audio synthesis module. The audio synthesis module converts the denoised audio data from digital to analog into analog audio data and stores it. The audio tracking and localization module extracts the envelopes of multiple denoised audio data streams and performs cross-correlation calculations to obtain the time delay difference between each microphone pair. One microphone in the array is selected as the reference first microphone, and the time delay difference τ between the remaining microphones and the first microphone is obtained. i1 Under near-field conditions of the cooling tower, the known coordinates of the first microphone (x1, y1, z1), the speed of sound c, and the coordinates of the remaining microphones (x1, y1, z1) are given. i ,y i ,z i ), delay difference τ i1 With unknown noise source coordinates (x) s ,y s ,z s It satisfies the following system of nonlinear equations: ; The directional noise reduction module solves the above equations based on the coordinates of multiple other microphones and their corresponding time delay differences to obtain the noise source coordinates (x). s ,y s,z s Based on this, noise sources can be dynamically tracked; The directional noise reduction module obtains the sector where the noise source is located based on the coordinates of the noise source, extracts the noise reduction frequency data of the microphone corresponding to the sector, calculates the sound signal intensity feature value before the current moment, and generates the louver opening control signal of the air inlet of the sector based on the sound signal intensity feature value, and adjusts the opening of the corresponding louver.

[0008] Optionally, the audio tracking and positioning module performs Hilbert transform on the multiple denoised audio data to construct their analytical signals, calculates the magnitude of the analytical signals, obtains the amplitude curves of each signal as the extracted envelopes, and performs cross-correlation calculations based on the envelopes to obtain the time delay difference between each microphone pair.

[0009] Optionally, the directional noise reduction module sets a preset window duration, and continuously takes N preset duration segments backward from the current moment to obtain N preset windows. It extracts the denoised frequency data of the microphone corresponding to the sector where the noise source is located in the obtained preset windows, and calculates the root mean square of the amplitude of all sampling points in the preset window as the sound signal intensity characteristic value of the preset window. If the root mean square calculated by N consecutive preset windows exceeds the threshold, the opening of the louver at the air inlet of the sector where the noise source is located is reduced, where N is a positive integer.

[0010] Optionally, the preset window duration is 5-20 seconds, and N is 2-4.

[0011] Optionally, the audio noise reduction module performs windowing operations including the following steps: Digital audio data is processed by dividing it into frames according to a preset frame length. The frames are separated by a preset frame shift length. The window function used for windowing satisfies the constant overlap and addition condition: when the window function is shifted according to the preset frame shift length, the sum of all shifted window functions at any time point is always equal to 1.

[0012] Optionally, the preset frame shift length is half of the preset frame length, and the window function is one of the following: triangular window, Hanning window, and sine window.

[0013] Optionally, the system's actuator also includes a water distribution mechanism. The water distribution mechanism is equipped with a main water distribution pipe, which is connected upstream to the hot water supply device to be cooled and downstream to several branch water distribution pipes, spraying hot water toward the packing inside the tower. The branch water distribution pipes of each sector are independent of each other, and each branch water distribution pipe is equipped with a water distribution valve. The directional noise reduction module generates a control signal based on the sound signal intensity characteristic value, and adjusts the opening degree of the louvers at the air inlet of the sector where the noise source is located and the opening degree of the water distribution valve corresponding to the coordinates of the noise source.

[0014] Optionally, each sector is provided with several water distribution branch pipes. The directional noise reduction module generates a control signal based on the sound signal intensity characteristic value to adjust the opening degree of the corresponding louver and the opening degree of the water distribution valve corresponding to the coordinate of the noise source.

[0015] Optionally, the audio denoising module sequentially performs filtering, windowing, Fourier transform, spectral subtraction, cepstral transformation, and inverse Fourier transform on the digital audio data to obtain denoised audio data.

[0016] Optionally, the logarithm of the frequency domain amplitude spectrum obtained after spectral subtraction is taken, and the cepstrum is obtained by inverse Fourier transform. In the cepstrum domain, components higher than the preset inverse frequency threshold are attenuated or set to zero to suppress reverberation. The amplitude spectrum is then recovered by Fourier transform and exponential operation, and inverse Fourier transform is performed in combination with the original phase to output the time-domain denoised audio data as denoised audio data.

[0017] This invention effectively removes background noise during normal cooling tower operation while retaining abnormal noise components by distributing a microphone array circumferentially at the bottom of the tower and establishing a three-dimensional coordinate system with the array center as the origin. Combined with an audio denoising module, it performs filtering, windowing, Fourier transform, spectral subtraction, and inverse Fourier transform processing on multiple audio data streams. The audio tracking and positioning module uses envelope cross-correlation to obtain the time delay difference between each microphone pair and constructs a nonlinear TDOA equation system under near-field conditions to accurately solve for the three-dimensional coordinates of the noise source, achieving dynamic tracking of noise hotspots at the bottom of the tower. The directional noise reduction module automatically locates the sector where the noise source is located based on its coordinates, extracts the characteristic intensity value of the denoised frequency signal for that sector, and generates a control signal to adjust the opening of the corresponding air inlet louvers. This achieves closed-loop control for directional reduction of noise at high-noise locations; the larger the characteristic intensity value, the smaller the opening of the louvers at the air inlet of that sector after adjustment. Compared with existing fixed sound-absorbing louvers with a uniform global configuration, this invention can adapt to the drift of noise hotspots caused by changes in wind direction, water temperature, and water distribution conditions. It only implements local noise reduction in sectors with excessive noise, avoiding the increase in wind resistance and energy consumption caused by excessive noise reduction in low-noise areas. At the same time, it ensures that high-noise areas receive sufficient noise reduction, significantly improving the synergy between noise reduction effect and cooling efficiency. Detailed Implementation

[0018] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0019] This invention provides an intelligent system for dynamically tracking and directionally reducing noise sources on a tower. The system's actuator includes a microphone array, and the system itself includes an audio noise reduction module, an audio tracking and positioning module, an audio synthesis module, and a directional noise reduction module.

[0020] Several microphones of the microphone array are distributed circumferentially around the cooling tower axis at the bottom of the tower, and are positioned opposite to several air inlets distributed circumferentially around the cooling tower. The water spray area at the bottom of the tower is divided into several uniform sectors. The system establishes a three-dimensional coordinate system with the center of the microphone array as the origin. The microphone array collects audio data during normal operation of the cooling tower as additive noise to be removed during spectral subtraction. During the operation of the cooling tower, the microphone array collects audio data and transmits it to the audio noise reduction module.

[0021] The audio denoising module converts the multi-channel audio data collected by the microphone array into multi-channel digital audio data. The denoised audio data is then obtained by sequentially filtering, windowing, Fourier transform, spectral subtraction, and inverse Fourier transform. The denoised audio data is then input into the audio tracking and positioning module, the directional noise reduction module, and the audio synthesis module. The audio synthesis module converts the denoised audio data from digital to analog and stores it.

[0022] The audio tracking and localization module extracts the envelopes of multiple denoised audio data streams and performs cross-correlation calculations to obtain the time delay difference between each microphone pair. One microphone in the array is selected as the reference first microphone, and the time delay difference τ between the remaining microphones and the first microphone is obtained. i1 Under near-field conditions of the cooling tower, the known coordinates of the first microphone (x1, y1, z1), the speed of sound c, and the coordinates of the remaining microphones (x1, y1, z1) are given. i ,y i ,z i ), delay difference τ i1 With unknown noise source coordinates (x) s ,y s ,z s It satisfies the following system of nonlinear equations: ; The directional noise reduction module solves the above equations based on the coordinates of multiple other microphones and their corresponding time delay differences to obtain the noise source coordinates (x). s ,y s ,z s Based on this, the noise source can be dynamically tracked. Optionally, the above equations can be solved using the least squares method or Chan's algorithm to obtain the noise source coordinates (x, y). s ,y s ,z s Based on this, the noise source is dynamically tracked.

[0023] The directional noise reduction module obtains the sector where the noise source is located based on the coordinates of the noise source, extracts the noise reduction frequency data of the microphone corresponding to the sector, calculates the sound signal intensity feature value before the current moment, and generates the louver opening control signal of the air inlet of the sector based on the sound signal intensity feature value, and adjusts the opening of the corresponding louver.

[0024] The present invention has the following beneficial effects: 1. A closed-loop control system for dynamic tracking and directional noise reduction of noise hotspots was achieved. This invention distributes a microphone array circumferentially at the bottom of the cooling tower and establishes a three-dimensional coordinate system with the array center as the origin. Combined with an audio denoising module, multiple audio data are filtered, windowed, subjected to Fourier transform, spectral subtraction, and inverse Fourier transform processing. This effectively removes background noise during normal cooling tower operation while preserving abnormal noise components in the sound field. The audio tracking and positioning module uses envelope cross-correlation to obtain the time delay difference between each microphone pair and constructs a nonlinear TDOA equation system under near-field conditions to accurately solve for the three-dimensional coordinates of the noise source, achieving continuous, real-time dynamic tracking of the noise hotspot location at the bottom of the tower. The directional noise reduction module automatically locates the sector where the noise source is located based on its coordinates, extracts the characteristic intensity value of the denoised frequency signal in that sector, and generates a control signal to adjust the opening of the corresponding air inlet louvers. This forms a complete closed-loop control system of "acquisition-denoising-positioning-decision-execution," achieving precise noise reduction.

[0025] 2. Overcomes the drawback of balancing wind resistance and noise reduction. Existing technologies uniformly apply sound-absorbing material to the louvers and use the same noise reduction configuration for all air intake channels, resulting in excessive noise reduction and unnecessary increase in wind resistance in low-noise areas, while insufficient noise reduction and unsatisfactory results in high-noise areas. This invention adjusts the louver opening only locally in the sectors where noise hotspots are detected in real time. A larger opening is maintained in low-noise areas to reduce wind resistance and ensure cooling efficiency, while the opening is appropriately reduced in high-noise areas to enhance noise reduction, achieving synergistic optimization of noise reduction and ventilation efficiency.

[0026] 3. Adapting to the complex and ever-changing operating conditions of cooling towers. Random changes in natural wind conditions, fluctuations in circulating water temperature, and adjustments to water distribution conditions can all cause dynamic shifts in the spatial location of noise hotspots at the base of the tower. Existing fixed noise reduction solutions cannot respond to such changes, as their noise reduction capability is locked once installed. This invention tracks the movement trajectory of noise hotspots in real time through continuous frame positioning and adjusts the operating sector of the actuator accordingly. Regardless of how the hotspots drift, the system can automatically follow and implement directional noise reduction, exhibiting strong adaptability and robustness.

[0027] 4. This invention avoids performance degradation and maintenance difficulties caused by dust accumulation and moisture in sound-absorbing materials. Existing technologies involve directly applying sound-absorbing cotton, micro-perforated panels, and other sound-absorbing materials to the louvers. However, the cooling tower inlet environment is humid and dusty, making the sound-absorbing materials prone to dust accumulation and mold growth. Moisture absorption significantly reduces sound absorption performance, and the narrowed airflow further increases air resistance, leading to high maintenance costs. This invention employs an active noise reduction strategy, adjusting the louver opening to change the airflow state and reduce noise. It eliminates the need for sound-absorbing materials on the louvers, fundamentally reducing the risk of material aging and maintenance costs.

[0028] 5. Improved positioning accuracy. This invention explicitly adopts a near-field spherical wave model to construct the time delay difference equations, which, compared to the far-field plane wave assumption, better reflects the actual physical propagation laws, resulting in smaller positioning errors and accurate differentiation of adjacent sectors, providing a reliable spatial coordinate basis for directional noise reduction.

[0029] It should be noted that, for potential multi-source noise, the "location," "tracking," and "direction" in the intelligent system for dynamically tracking and directionally reducing noise sources in the tower provided by this invention refer to estimating and locating the spatial center of noise energy in the sound field at the base of the tower. This is achieved by using multiple audio signals collected by a microphone array to obtain an equivalent noise source coordinate. This coordinate represents the weighted center position of noise intensity within the annular area at the base of the tower at the current moment, rather than referring to a specific isolated physical sound source. Because different sound sources are close together and their spectra overlap, the system does not distinguish the specific number or location of each independent sound source, nor does it aim to resolve differences in sound sources with an angle smaller than the sector width. "Location" refers to assigning the representative center of the noise source to a sector to guide the directional noise reduction actuator to adjust that sector, thereby affecting the operating conditions of the actual sound sources distributed in that sector or at least in its vicinity. Since the actuator's adjustment targets are themselves divided by sector, this sector-level positioning accuracy is sufficient to meet the practical engineering needs of louver opening adjustment and water distribution valve control.

[0030] Optionally, the microphone array is a 16-channel microphone array.

[0031] In one possible implementation, the audio tracking and localization module performs Hilbert transform on multiple denoised audio data streams to construct their analytic signals, calculates the magnitude of the analytic signals, and obtains the amplitude curves of each signal as extracted envelopes. Cross-correlation calculations are then performed based on these envelopes to obtain the time delay difference between each microphone pair. This effectively suppresses the impact of high-frequency attenuation and waveform distortion on time delay estimation under the high humidity environment inside the cooling tower. The envelope curves retain the fluctuation characteristics of sound energy over time, eliminating phase distortion and amplitude fluctuations caused by differences in propagation paths, material reflection, and air absorption in the original waveform. This makes the peak value of the cross-correlation function sharper and more stable, significantly improving the estimation accuracy and anti-interference capability of the time delay difference. It provides reliable input parameters for subsequent solving of the near-field TDOA equations, ultimately improving the accuracy of noise source coordinate localization and the continuity of dynamic tracking.

[0032] In one possible implementation, the directional noise reduction module sets a preset window duration. Starting from the current moment, it continuously takes N preset duration segments to obtain N preset windows. It extracts the denoised frequency data of the microphone corresponding to the sector where the noise source is located within the obtained preset windows. It calculates the root mean square (RMS) of the amplitudes of all sampling points within the preset window as the sound signal intensity characteristic value of that preset window. If the RMS calculated for N consecutive preset windows exceeds a threshold, the opening of the louvers at the air inlet of the sector where the noise source is located is reduced, where N is a positive integer. Optionally, the preset window duration is 5-20 seconds, and N is 2-4. This invention, by setting multiple windows for sound signal intensity characteristic value comparison, triggers louver opening adjustment only when consecutive thresholds are exceeded, effectively avoiding erroneous actions caused by instantaneous noise fluctuations or occasional interference, and improving the stability and reliability of noise reduction decisions. Simultaneously, this invention uses the RMS as the characteristic value, which can objectively reflect the average energy level of the noise in the sector, overcoming the shortcomings of peak detection being easily affected by single-point impacts and the average value being insufficiently sensitive to small amplitudes, making the adjustment basis consistent with subjective auditory loudness and actual noise reduction requirements. In addition, by making continuous multi-window judgments, the system can adaptively filter out short-term random noise, ensuring that adjustment actions are only performed when the noise continuously exceeds the standard, thereby reducing the frequent operation of the actuator, extending the equipment life, and maintaining the stability of wind resistance.

[0033] In one possible implementation, the audio denoising module performs windowing operations including the following steps: Dividing the digital audio data into frames according to a preset frame length, with a preset frame shift length separating adjacent frames; the window function used for windowing satisfies the constant overlap and addition condition: when the window function is shifted according to the preset frame shift length, the sum of all shifted window functions at any time point is always equal to 1. Optionally, the preset frame shift length is half the preset frame length, and the window function is one of a triangular window, a Hanning window, and a sine window.

[0034] This invention employs a window function that satisfies a constant overlap and addition condition for frame-by-frame windowing processing, and sets a fixed frame shift length. This ensures that when the time-domain frames obtained after frequency domain processing (such as spectral subtraction and cepstrum) and inverse Fourier transform are directly overlapped and added with the same frame shift, the sum of all shifted window functions at any given time point is always equal to 1, thus eliminating the need for additional amplitude normalization. This characteristic avoids signal distortion or gain fluctuations caused by amplitude superposition in the inter-frame overlap region, ensuring that the amplitude of the synthesized continuous audio is consistent with the original signal, simplifying subsequent processing, reducing computational overhead, and improving real-time performance. Furthermore, this condition does not depend on a specific ratio between frame shift and frame length; as long as the window function design matches the frame shift, it provides flexibility in system parameter selection.

[0035] In one possible implementation, the system's actuator also includes a water distribution mechanism. The water distribution mechanism is equipped with a main water distribution pipe, which is connected upstream to the hot water supply device to be cooled and downstream to several branch water distribution pipes. Hot water is sprayed toward the packing material inside the tower. The branch water distribution pipes of each sector are independent of each other and are equipped with water distribution valves. The directional noise reduction module generates a control signal based on the sound signal intensity characteristic value to adjust the opening degree of the louvers at the air inlet of the sector where the noise source is located and the opening degree of the water distribution valve corresponding to the coordinates of the noise source.

[0036] Optionally, each sector is equipped with several water distribution branch pipes. The directional noise reduction module generates control signals based on the sound signal intensity characteristics to adjust the opening of the corresponding louvers and the opening of the water distribution valve corresponding to the noise source coordinates. The presence of several water distribution branch pipes in each sector allows for further subdivision and independent control of the spatial distribution of water spray within the same sector. The directional noise reduction module simultaneously adjusts the louver opening and the water distribution valve opening corresponding to the noise source coordinates based on the sound signal intensity characteristics, achieving coordinated noise reduction on both the wind side (air intake) and the water side (water spray). When the noise source is located in a localized area within a sector, the system not only reduces the air intake speed in that sector by closing the louvers to suppress airflow noise, but also reduces the local water spray density and decreases water droplet impact noise by closing the corresponding water distribution valve. This dual adjustment mechanism avoids the limitations of a single adjustment method, simultaneously addressing aerodynamic noise caused by wind conditions and impact noise caused by uneven water distribution, significantly improving the noise reduction effect. Meanwhile, since each sector corresponds to multiple water distribution branches, the adjustment range is more precise, avoiding large-scale reduction of water flow across the entire sector that would affect cooling efficiency, thus achieving a refined balance between noise reduction and heat exchange performance. In addition, coordinated adjustment of the air side and water side can reduce the magnitude of their respective adjustments, reduce wind resistance fluctuations and water flow impact changes, and improve the stability of system operation.

[0037] In one possible implementation, the audio denoising module sequentially performs filtering, windowing, Fourier transform, spectral subtraction, cepstral transformation, and inverse Fourier transform on the digital audio data to obtain denoised audio data. Optionally, the logarithm of the frequency domain amplitude spectrum obtained after spectral subtraction is taken, and an inverse Fourier transform is performed to obtain the cepstral spectrum. In the cepstral domain, components higher than a preset cepstral frequency threshold are attenuated or zeroed to suppress reverberation. The amplitude spectrum is then recovered through Fourier transform and exponential operation, and combined with the original phase, an inverse Fourier transform is performed to output the time-domain denoised audio data as the denoised audio data.

[0038] This invention effectively suppresses the interference of reverberation (i.e., the trailing effect caused by sound after multiple reflections through the tower wall, packing, and water surface) on audio signals by introducing cepstral processing. Specifically, after taking the logarithm of the frequency domain amplitude spectrum after spectral subtraction and transforming it to the cepstral domain, the direct sound energy is concentrated in the low cepstral frequency region, while the reverberation component exhibits periodic fluctuations in the high cepstral frequency region. By setting a cepstral frequency threshold, the high cepstral frequency components above the threshold are attenuated or zeroed out. Then, the amplitude spectrum is recovered through inverse transformation, allowing the reverberation component to be separated and removed from the original signal while preserving the spectral details of the direct sound. This processing significantly improves the clarity of the audio signal, making the time delay difference peaks in subsequent envelope extraction and cross-correlation calculations sharper and more accurate, avoiding false peaks or peak broadening caused by reverberation, thereby improving the accuracy of noise source localization and the reliability of dynamic tracking. Furthermore, cepstral processing does not rely on a priori models of the reverberation environment and can adaptively suppress time-varying reverberation, making it suitable for the complex and variable acoustic environment inside cooling towers.

[0039] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent system for dynamically tracking and directionally reducing noise sources in a tower structure, characterized in that, The system's actuators include a microphone array, and the system includes an audio noise reduction module, an audio tracking and positioning module, an audio synthesis module, and a directional noise reduction module. The microphone array has several microphones distributed circumferentially around the cooling tower axis at the bottom of the tower, and is positioned opposite to several air inlets distributed circumferentially around the cooling tower. The water spray area at the bottom of the tower is divided into several uniform sectors. The system establishes a three-dimensional coordinate system with the center of the microphone array as the origin. The microphone array collects audio data during normal operation of the cooling tower as additive noise to be removed during spectral subtraction. During the operation of the cooling tower, the microphone array collects audio data and transmits it to the audio noise reduction module. The audio denoising module converts the multi-channel audio data collected by the microphone array into multi-channel digital audio data through analog-to-digital conversion. The denoised audio data is then obtained by sequentially filtering, windowing, Fourier transform, spectral subtraction, and inverse Fourier transform. The denoised audio data is then input into the audio tracking and positioning module, the directional noise reduction module, and the audio synthesis module. The audio synthesis module converts the denoised audio data from digital to analog into analog audio data and stores it. The audio tracking and localization module extracts the envelopes of multiple denoised audio data streams and performs cross-correlation calculations to obtain the time delay difference between each microphone pair. It then selects one microphone in the array as the reference first microphone and obtains the time delay difference τ between the remaining microphones and the first microphone. i1 Under near-field conditions of the cooling tower, the known coordinates of the first microphone (x1, y1, z1), the speed of sound c, and the coordinates of the remaining microphones (x1, y1, z1) are given. i ,y i ,z i ), delay difference τ i1 With unknown noise source coordinates (x) s ,y s ,z s It satisfies the following system of nonlinear equations: ； The directional noise reduction module solves the above equations based on the coordinates of multiple other microphones and their corresponding time delay differences to obtain the noise source coordinates (x... s ,y s ,z s Based on this, noise sources can be dynamically tracked; The directional noise reduction module obtains the sector where the noise source is located based on the noise source coordinates, extracts the noise reduction frequency data of the microphone corresponding to the sector, calculates the sound signal intensity feature value before the current moment, generates the louver opening control signal of the air inlet of the sector based on the sound signal intensity feature value, and adjusts the opening of the corresponding louver.

2. The intelligent system for dynamically tracking tower noise sources and directional noise reduction according to claim 1, characterized in that, The audio tracking and positioning module performs Hilbert transform on the multiple denoised audio data to construct their analytical signals, calculates the magnitude of the analytical signals, and obtains the amplitude curves of each signal as the extracted envelopes. Based on the envelopes, cross-correlation calculations are performed to obtain the time delay difference between each microphone pair.

3. The intelligent system for dynamically tracking tower noise sources and directional noise reduction according to claim 1, characterized in that, The directional noise reduction module sets a preset window duration, and continuously takes N preset duration segments forward from the current moment to obtain N preset windows. It extracts the denoised frequency data of the microphone corresponding to the sector where the noise source is located in the obtained preset windows, and calculates the root mean square of the amplitude of all sampling points in the preset window as the sound signal intensity feature value of the preset window. If the root mean square calculated by N consecutive preset windows exceeds the threshold, the opening of the louver at the air inlet of the sector where the noise source is located is reduced, where N is a positive integer.

4. The intelligent system for dynamically tracking tower noise sources and directional noise reduction according to claim 3, characterized in that, The preset window duration is 5-20 seconds, and N is 2-4.

5. The intelligent system for dynamically tracking tower noise sources and directional noise reduction according to claim 1, characterized in that, The audio noise reduction module performs the windowing operation including the following steps: Digital audio data is processed by dividing it into frames according to a preset frame length. The frames are separated by a preset frame shift length. The window function used for windowing satisfies the constant overlap and addition condition: when the window function is shifted according to the preset frame shift length, the sum of all shifted window functions at any time point is always equal to 1.

6. The intelligent system for dynamically tracking tower noise sources and directional noise reduction according to claim 5, characterized in that, The preset frame shift length is half of the preset frame length, and the window function is one of the following: triangular window, Hanning window, and sine window.

7. The intelligent system for dynamically tracking tower noise sources and directional noise reduction according to claim 1, characterized in that, The system's actuator also includes a water distribution mechanism, which is equipped with a main water distribution pipe connected upstream to a hot water supply device to be cooled and downstream to several branch water distribution pipes that spray hot water toward the packing material inside the tower. The branch water distribution pipes of each sector are independent of each other, and each branch water distribution pipe is equipped with a water distribution valve. The directional noise reduction module generates a control signal based on the sound signal intensity characteristic value to adjust the opening degree of the louvers at the air inlet of the sector where the noise source is located and the opening degree of the water distribution valve at the coordinates of the noise source.

8. The intelligent system for dynamically tracking tower noise sources and directional noise reduction according to claim 7, characterized in that, Each sector is equipped with several water distribution branch pipes. The directional noise reduction module generates a control signal based on the sound signal intensity characteristic value to adjust the opening degree of the corresponding louver and the opening degree of the water distribution valve corresponding to the noise source coordinate.

9. The intelligent system for dynamically tracking tower noise sources and directional noise reduction according to claim 1, characterized in that, The audio denoising module sequentially performs filtering, windowing, Fourier transform, spectral subtraction, cepstral transformation, and inverse Fourier transform on the digital audio data to obtain denoised audio data.

10. The intelligent system for dynamically tracking tower noise sources and directional noise reduction according to claim 9, characterized in that, The frequency domain amplitude spectrum obtained after spectral subtraction is logarithmically taken, and then inverse Fourier transform is performed to obtain the cepstrum. In the cepstrum domain, components above a preset inverse frequency threshold are attenuated or set to zero to suppress reverberation. The amplitude spectrum is then recovered through Fourier transform and exponential operation. Combined with the original phase, inverse Fourier transform is performed to output the time-domain denoised audio data as denoised audio data.

Images