Composite noise reduction method based on seawater desulfurization aeration tank fan

Abstract

The invention discloses a composite noise reduction method based on a seawater desulfurization aeration tank fan. The composite noise reduction method comprises the following steps: S1, deploying a composite noise reduction system; s2, initializing a system; s3, secondary channel modeling: controlling a secondary sound source to play white noise, and collecting response through an error sensor so as to establish or update a secondary channel estimation model; s4, noise reduction operation: starting a fan and operating a composite FXLMS algorithm; and S5, monitoring and optimizing: monitoring the energy value of the error signal in real time, and dynamically adjusting the self-adaptive step size sum according to the convergence state. By combining the feedforward and feedback composite FXLMS structure, the system can process deterministic noise related to reference signals and random noise which cannot be directly measured at the same time, realizes more comprehensive suppression of broadband and non-stationary noise, and is especially suitable for complex sound source characteristics with periodicity and randomness in fan noise.

Description

Technical Field

[0001] This application belongs to the field of noise control technology, specifically relating to a composite noise reduction method based on the blower of the seawater desulfurization aeration tank. Background Technology

[0002] In seawater desulfurization systems, the aeration tank blower is one of the core pieces of equipment. During operation, it generates strong broadband noise, especially with significant peaks in the low-to-mid frequency range (around 500Hz). This type of noise not only pollutes the plant area and surrounding environment, affecting the health of workers, but may also cause structural fatigue of the equipment due to long-term vibration.

[0003] Common noise reduction methods include passive sound insulation and sound-absorbing material wrapping, but their effectiveness against low-frequency noise is limited, and their durability is poor due to corrosion and humidity in seawater environments. In recent years, active noise control technology has been gradually applied in industrial fields. Based on the principle of sound wave interference, it achieves noise cancellation by generating anti-phase sound waves. However, traditional single-structure active control systems (such as feedforward or feedback) suffer from slow convergence speed, insufficient stability, and poor adaptability when dealing with complex sound fields, time-varying operating conditions, and changes in secondary channels. They are difficult to achieve continuous and efficient noise reduction in real-world scenarios with strong interference and multiple variables, such as seawater desulfurization aeration tanks.

[0004] Therefore, a composite noise reduction method is urgently needed to solve the above problems. Summary of the Invention

[0005] This application provides a composite noise reduction method based on the blower of the seawater desulfurization aeration tank, which aims to solve the problems of slow convergence speed, insufficient stability and poor adaptability of the existing technology.

[0006] A composite noise reduction method based on the blower of a seawater desulfurization aeration tank includes the following steps:

[0007] S1: Deployment of the composite noise reduction system, which includes a reference sensor, an error sensor, a secondary sound source, and a controller. The reference sensor acquires the raw noise signal from the noise source. Error sensor collects residual noise signal The secondary sound source emits anti-phase sound waves, and the controller is electrically connected to the reference sensor, the error sensor, and the secondary sound source.

[0008] The controller is configured to execute the composite FXLMS adaptive algorithm.

[0009] S2: System initialization, power-on initialization of the controller, setting algorithm parameters, including sampling frequency fs and feedforward filter length. Feedback filter length Initial step size and ;

[0010] S3: Secondary channel modeling, controlling the secondary sound source to play white noise, and collecting the response through an error sensor to establish or update the secondary channel estimation model. ;

[0011] S4: Noise Reduction Operation: Start the fan and run the composite FXLMS algorithm, including the following steps:

[0012] Synchronous acquisition of reference signals Sum of error signals ;

[0013] Generate internal reference signal ;

[0014] Calculate controller output = + , and output to a secondary sound source (3);

[0015] According to the formula:

[0016] Update the weight coefficients of the feedforward and feedback controllers;

[0017] S5: Monitoring and Optimization, real-time monitoring of error signals. The energy value is determined, and the adaptive step size is dynamically adjusted based on the convergence state. and .

[0018] Optionally, S2 includes the following steps:

[0019] The system power-on and self-test are performed sequentially by the sensor power supply module, the secondary sound source amplifier, and the controller host.

[0020] The controller executes a power-on self-test program to check whether the ADC / DAC channels, memory and communication interfaces are normal. The self-test information is transmitted to the host computer monitoring software through the communication network.

[0021] Enter the key parameters and download them to the controller.

[0022] Optionally, the key parameter includes: sampling frequency fs;

[0023] Filter length: Feedforward filter length Feedback filter length ;

[0024] Adaptive step size: Feedforward step size Feedback step size ;

[0025] Variable step size threshold: Set the error energy high threshold TH1 and low threshold TH2.

[0026] Optionally, S3 includes the following steps:

[0027] Modeling signal transmission and acquisition: The controller controls one of the secondary sound sources to play a pseudo-random white noise with a duration of 2 seconds and a bandwidth of 50-1500Hz. Its amplitude is calibrated and is within the linear working range. All error sensors corresponding to the sound source synchronously acquire response signals. In order to improve the model accuracy, the four acquired response signals are spatially averaged to obtain an average response signal.

[0028] Model Identification and Update: The controller uses the LMS algorithm to run a modeling program, taking the played white noise as input and the average response signal as the desired response, adaptively updating a 128-bit FIR filter. The weight coefficients are determined, and the modeling process continues until the filter weight coefficients converge. The coefficients are stored in the controller's memory and used to generate the filter reference signal in all subsequent FXLMS calculations.

[0029] Optionally, S3 needs to be performed under the condition that the fan is running but active noise reduction is not activated.

[0030] Optionally, S4 includes the following steps:

[0031] S4.1: Synchronous signal acquisition. When each sampling clock interrupt occurs, the controller synchronously samples the reference signal through its multi-channel ADC. And four error signals, which are averaged in real time into a single error signal. For use by the algorithm;

[0032] S4.2: Internal reference signal generation and calculation , here It uses storage The model's output of the controller at the previous time step Perform convolution operations to obtain the original main noise. Estimate ;

[0033] S4.3: Generation of Filtered Reference Signal:

[0034] calculate ;

[0035] calculate ;

[0036] S4.4: Calculate and output the control signal, and calculate the feedforward output: ;

[0037] in, It is a length of The reference signal vector;

[0038] Calculation feedback section output: ;

[0039] in, It is a length of ; internal reference signal vector;

[0040] Calculate the total output: ;

[0041] This The amplifier, whose output is sent to the secondary sound source via the DAC, drives the speaker to emit anti-phase sound waves.

[0042] S4.5: Adaptive update of weight coefficients. The filter weights are updated immediately using the following formula to prepare for the next sampling time:

[0043] ;

[0044] .

[0045] Optionally, the dynamic step size adjustment specifically involves: when the error energy is detected to suddenly increase and exceed the threshold TH1, the system automatically adjusts the step size. and The value was temporarily increased to 1.5 times the initial value to accelerate the convergence speed and quickly recapture noise changes;

[0046] When the system is running stably and the error energy is below the threshold TH2, the system will restore the step size to the initial value to reduce the steady-state error and obtain a more refined noise reduction effect.

[0047] Optionally, S5 also includes periodic maintenance and model updates: the system is set to automatically perform a secondary channel modeling step every 24 hours to compensate for the effects of slow changes in temperature, humidity or dirt on the acoustic channel;

[0048] If the controller's built-in temperature and humidity sensor detects a sudden change in the environment, it will immediately trigger an additional channel modeling.

[0049] Compared with the prior art, this application has at least the following beneficial effects:

[0050] By combining a feedforward and feedback composite FXLMS structure, this application enables the system to simultaneously process deterministic noise related to the reference signal and random noise that cannot be directly measured, achieving more comprehensive suppression of wideband, non-stationary noise. It is especially suitable for complex sound source characteristics in wind turbine noise that have both periodicity and randomness.

[0051] This application also performs secondary channel modeling periodically or triggeredly during wind turbine operation. By using pseudo-random white noise excitation and spatial averaging with multiple error sensors, the acoustic path response is accurately estimated, effectively compensating for channel changes caused by factors such as temperature, humidity, and dirt accumulation, and significantly improving the robustness and convergence stability of the algorithm in real-world environments.

[0052] This application also achieves a balance between rapid response and fine noise reduction by real-time monitoring of error energy and setting dual thresholds. The system can automatically increase the step size to accelerate convergence when noise changes abruptly, and restore the small step size to reduce steady-state error when running stably. This adapts to the changes in sound field caused by fluctuations in the operating conditions of the wind turbine. Attached Figure Description

[0053] Figure 1 A flowchart of a composite noise reduction method based on a blower in a seawater desulfurization aeration tank, provided for this application. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments.

[0055] This application provides a composite noise reduction method based on a blower in a seawater desulfurization aeration tank, comprising the following steps:

[0056] S1: Deployment of the composite noise reduction system, wherein the reference sensor, error sensor and secondary sound source are installed at predetermined locations in the blower outlet pipe and aeration tank;

[0057] The operating aeration tank blower system was initially tested using an acoustic camera or portable sound level meter to determine the main noise radiation area and the dominant noise frequency. In the example, a significant peak was found at around 500Hz. The reference sensor installation point was selected on the straight pipe section after the blower outlet and before the first bend, at a distance of ≥5 times the pipe diameter from the bend. The airflow is relatively stable at this location, which can capture the signal with the highest correlation to the noise source.

[0058] Error sensor installation point: Selected at about 0.5 meters below the top cover of the aeration tank, four sensors are arranged in a 2×2 rectangular array, with the center of the array aligned with the center of the tank. This layout can effectively sense the overall sound field of the tank space and provide comprehensive error information for the controller.

[0059] Secondary sound source installation points: Select the upper part of both sides of the aeration tank and symmetrically install two waterproof loudspeakers with their sound axes pointing towards the center of the tank to avoid forming an acoustic shadow area.

[0060] Weld or bolt the stainless steel mounting bases to all selected locations, ensuring a seal. Connect the sensors and speakers to the controller cabinet located in the explosion-proof control room via marine-grade armored cables. All cable connectors use IP68 waterproof aviation plugs. After wiring is complete, use a megohmmeter to check the insulation resistance of all lines to ensure there is no risk of short circuits in humid environments.

[0061] S2: Initialization, power-on initialization of the controller, setting algorithm parameters, including sampling frequency fs and feedforward filter length. Feedback filter length Initial step size and ;

[0062] Specifically, the system power-on and self-test are performed sequentially by the sensor power supply module, the secondary sound source amplifier, and the controller host.

[0063] The controller executes a power-on self-test (POST) procedure to check the ADC / DAC channels, memory, and communication interfaces for proper functioning. The POST information is transmitted to the host computer monitoring software via Ethernet.

[0064] Enter the following key parameters through the host computer software interface and download them to the controller:

[0065] Sampling frequency fs: set to 4000Hz. This frequency is based on the Nyquist sampling theorem (fs>2*). The system processing capacity is determined to be sufficient to cover the target noise reduction frequency band (0-800Hz).

[0066] Filter length: Feedforward filter length Setting it to 64, a longer order can better simulate the response of the feedforward channel;

[0067] Feedback filter length The value is set to 32. The feedback mainly deals with narrowband noise, and a shorter length is chosen to reduce the computational load.

[0068] Adaptive step size: Feedforward step size The initial value is set to 0.02.

[0069] Feedback Step The initial value is set to 0.005. Feedback loops are usually more sensitive, so a smaller step size is used to ensure stability.

[0070] Variable step size thresholds: Set a high error energy threshold TH1 (for increasing the step size) and a low error energy threshold TH2 (for decreasing the step size). In this embodiment, TH1 corresponds to an error signal RMS value of 0.5V, and TH2 corresponds to 0.1V;

[0071] S3: Secondary channel modeling, controlling the secondary sound source to play white noise, and collecting the response through an error sensor to establish or update the secondary channel estimation model. ;

[0072] The modeling steps specifically include:

[0073] Modeling signal transmission and acquisition: Click the "Channel Modeling" button in the host computer software, and the controller controls one of the secondary sound sources to play a pseudo-random white noise with a duration of 2 seconds and a bandwidth of 50-1500Hz. Its amplitude is calibrated and is within the linear working range. All error sensors corresponding to this sound source synchronously acquire response signals. In order to improve the model accuracy, the four acquired response signals are spatially averaged to obtain an average response signal.

[0074] Model Identification and Update: The controller employs the LMS algorithm, running a separate modeling program. This program takes the played white noise as input and the average response signal as the desired response, adaptively updating a 128-bit FIR filter. The weight coefficients are calculated, and the modeling process lasts approximately 2 seconds until the filter weight coefficients converge. The coefficients are stored in the controller's memory and used to generate the filter reference signal in all subsequent FXLMS calculations;

[0075] S3 needs to be performed under the condition that the wind turbine is running but active noise reduction is not activated to ensure the accuracy of the modeling;

[0076] S4: Noise Reduction Operation: Start the fan and run the composite FXLMS algorithm, including the following steps:

[0077] Synchronous acquisition of reference signals Sum of error signals ;

[0078] Generate internal reference signal ;

[0079] Calculate controller output = + , and output to a secondary sound source (3);

[0080] According to the formula:

[0081] Update the weight coefficients of the feedforward and feedback controllers;

[0082] Specifically, S4 includes the following steps:

[0083] S4.1: Synchronous signal acquisition. When each sampling clock interrupt occurs, the controller synchronously samples the reference signal through its multi-channel ADC. And four error signals, which are averaged in real time into a single error signal. For use by the algorithm;

[0084] S4.2: Internal reference signal generation and calculation Here It uses storage The model's output of the controller at the previous time step The purpose of performing convolution operations is to cancel out the influence of the secondary sound source from the error signal, thereby obtaining the original main noise. Estimate ;

[0085] S4.3: Generation of Filtered Reference Signal:

[0086] calculate

[0087] calculate

[0088] These two steps are the core of the FXLMS algorithm. Their purpose is to "pre-distort" the reference signal to compensate for the phase and amplitude distortion caused by the secondary channel S(z), thereby ensuring the stability of the adaptive process.

[0089] S4.4: Calculate and output the control signal, and calculate the feedforward output: ;

[0090] in, It is a length of The reference signal vector;

[0091] Calculation feedback section output: ;

[0092] in, It is a length of ; internal reference signal vector;

[0093] Calculate the total output: ;

[0094] This The amplifier, whose output is sent to the secondary sound source via the DAC, drives the speaker to emit anti-phase sound waves.

[0095] S4.5: Adaptive update of weight coefficients. The filter weights are updated immediately using the following formula to prepare for the next sampling time:

[0096] ;

[0097] ;

[0098] S5: Monitoring and Optimization, real-time monitoring of error signals. The energy value is determined, and the adaptive step size is dynamically adjusted based on the convergence state. and ;

[0099] S5 specifically includes:

[0100] Real-time performance monitoring: The host computer software displays error signals in real time. The time-domain waveform and spectrum diagram allow operators to visually observe the decrease in noise level, while the controller internally calculates the short-term average energy of the error signal.

[0101] Dynamic step size adjustment (variable step size strategy): When the error energy is detected to suddenly increase and exceed the threshold TH1 (e.g., due to changes in wind turbine operating conditions), the system automatically adjusts the step size. and The value was temporarily increased to 1.5 times the initial value to accelerate the convergence speed and quickly recapture noise changes;

[0102] When the system is running stably and the error energy is below the threshold TH2, the system will restore the step size to the initial value to reduce the steady-state error and obtain a more refined noise reduction effect.

[0103] Periodic maintenance and model updates: The system is set to automatically perform a secondary channel modeling step every 24 hours to compensate for the effects of slow changes in temperature, humidity or dirt on the acoustic channels;

[0104] If the controller's built-in temperature and humidity sensor detects a sudden change in the environment, it will immediately trigger an additional channel modeling.

[0105] The composite noise reduction system includes: a reference sensor, which employs a seawater-corrosion-resistant probe-type microphone and is installed at the outlet pipe of the blower or the air inlet of the aeration tank, for collecting the raw noise signal from the noise source. ;

[0106] An error sensor, employing a seawater-corrosion-resistant probe-type microphone, is installed in the gas space or exhaust port of the aeration tank to collect residual noise signals. ;

[0107] The secondary sound source, which is a waterproof loudspeaker, is installed in the aeration tank or pipe to emit anti-phase sound waves.

[0108] A controller, which is electrically connected to the reference sensor, the error sensor, and the secondary sound source;

[0109] The controller is configured to execute the composite FXLMS adaptive algorithm.

[0110] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

Claims

1. A composite noise reduction method based on seawater desulfurization aeration tank fan, characterized in that, The method comprises the following steps: S1: composite noise reduction system is deployed, the composite noise reduction system includes a reference sensor, an error sensor, a secondary sound source and a controller, the reference sensor collects the original noise signal from the noise source , the error sensor collects the residual noise signal , the secondary sound source emits an inverted sound wave, and the controller is electrically connected with the reference sensor, the error sensor and the secondary sound source; The controller is configured to execute a composite FXLMS adaptive algorithm; S2: system initialization, power-on initialization to the controller, setting algorithm parameters, including sampling frequency fs, feedforward filter length , feedback filter length , initial step size , and ; S3: Secondary path modeling, controlling the secondary sound source to play white noise, and collecting the response through the error sensor to establish or update the secondary path estimation model ; S4: noise reduction operation: start the fan and run the composite FXLMS algorithm, comprising the following steps: Synchronizing acquisition of reference signals and error signals ; Generating an internal reference signal ; Computing controller output = + and output to the secondary sound source (3); According to the formula: Update the weight coefficients of the feedforward and feedback controllers; S5: Monitoring and optimization, real-time monitoring of error signal energy value, and dynamically adjusting the adaptive step size according to the convergence state and .

2. The seawater desulfurization aeration tank fan-based composite noise reduction method according to claim 1, characterized in that, The S2 comprises the following steps: System power-on and self-test, in turn, power on the sensor power supply module, the secondary sound source power amplifier, and the controller host; The controller executes a startup self-test program to check whether the ADC / DAC channel, memory and communication interface are normal, and transmits the self-test information to the upper computer monitoring software through the communication network; Input the key parameters and download them to the controller.

3. The seawater desulfurization aeration tank fan-based composite noise reduction method according to claim 2, characterized in that, The key parameters include: sampling frequency fs; Filter length: feedforward filter length feedback filter length ; Adaptive step size: feedforward step size , feedback step size ; Variable step threshold: set the high threshold TH1 and the low threshold TH2 of the error energy.

4. The seawater desulfurization aeration tank fan-based composite noise reduction method according to claim 1, characterized in that, The S3 comprises the following steps: Modeling signal emission and collection: the controller controls one of the secondary sound sources to play a pseudo-random white noise with a duration of 2 seconds and a bandwidth of 50-1500 Hz, and the amplitude is calibrated to be within the linear working area. All error sensors corresponding to the sound source synchronously collect the response signal, and the collected 4-way response signals are spatially averaged to obtain an average response signal to improve the model accuracy. Model identification and update: the controller employs LMS algorithm, runs modeling program, takes the played white noise as input, takes the average response signal as expected response, and adaptively updates the weight coefficients of a FIR filter with length of 128 The modeling process continues until the weight coefficients of the filter converge, and the converged coefficients are stored in the controller memory for generating the filtered reference signal in all subsequent FXLMS calculations.

5. The seawater desulfurization aeration tank fan-based composite noise reduction method according to claim 4, characterized in that, The S3 needs to be performed under the condition that the fan is running but the active noise reduction is not started.

6. The seawater desulfurization aeration tank fan-based composite noise reduction method according to claim 1, characterized in that, The S4 comprises the following steps: S4.1 : Signal synchronization acquisition, at each sampling clock interrupt arrival, the controller synchronously samples the reference signal through its multi-channel ADC and 4 error signals, the 4 error signals are averaged in real time into a single error signal for algorithm use; S4.2: internal reference signal generation, computation Here is the convolution of the stored model with the controller output from the previous time step, resulting in an estimate of the original primary noise ; S4.3: filter reference signal generation: Computing ; Computing ; S4.4: Calculate and output control signal, calculate feedforward part output: ; wherein is a reference signal vector of length ​ Computing the feedback portion output: ; wherein is an internal reference signal vector of length ​ Total output is calculated: ; This The power amplifier outputs the inverse sound wave to the secondary sound source through the DAC. S4.5: weight coefficient adaptive update, immediately update the filter weight using the following formula to prepare for the next sampling time: ； 。 7. The seawater desulfurization aeration tank fan-based composite noise reduction method according to claim 1, characterized in that, The dynamic step adjustment specifically includes: when it is monitored that the error energy suddenly increases and exceeds a threshold TH1, the system automatically increases the step size to 1.5 times of the initial value, so as to accelerate the convergence speed and quickly recapture the noise change. and temporarily to 1.5 times of the initial value, so as to accelerate the convergence speed and quickly recapture the noise change. When the system is running stably and the error energy is lower than the threshold TH2, the system restores the step length to the initial value to reduce the steady-state error and obtain a more fine noise reduction effect.

8. The seawater desulfurization aeration tank fan-based composite noise reduction method according to claim 1, characterized in that, The S5 further comprises periodic maintenance and model update: the system is set to automatically execute the secondary channel modeling step once every 24 hours to compensate for the effects of temperature, humidity or slow changes such as dirt accumulation on the acoustic channel; If the built-in temperature and humidity sensor of the controller detects a dramatic change in the environment, an additional channel modeling will be triggered immediately.

Images