High-voltage circuit breaker mechanical fault audio signal cluster acquisition system and method based on adaptive filtering and sampling

Abstract

The invention relates to a high-voltage circuit breaker mechanical fault audio signal cluster acquisition system and method based on adaptive filtering and sampling. The system comprises a mechanical fault simulation module, a fault audio signal transmission module, an adaptive sampling module, a noise adding processing module, an adaptive filtering module and an audio fusion module. The acquisition method comprises the following steps: simulating a mechanical fault of the circuit breaker in an operation process, and transmitting and sampling an audio signal of the circuit breaker in a clustered sensor network; simulating an on-site noise environment, and carrying out noise adding processing on the audio signal obtained by sampling; transmitting the audio signal after noise addition processing to an adaptive filtering module for filtering and noise reduction processing, and calculating a signal-to-noise ratio of the processed or unprocessed audio signal; and optimizing an energy consumption formula of the voiceprint sensor according to the obtained audio signal and the signal-to-noise ratio, calculating a channel needing frequency fusion and the rate of the voiceprint sensor through a self-adaptive sampling heuristic algorithm, performing maximum ratio combination, and outputting a fault sound source.

Description

Technical Field

[0001] The present invention belongs to the field of online monitoring of electric power equipment, and relates to a system and method for collecting clustered audio signals of mechanical faults of high-voltage circuit breakers based on adaptive filtering and sampling, and in particular to a system and method for online monitoring of electric power equipment and improving signal quality using signal processing technology. Background Art

[0002] In the power system, 10kV circuit breakers bear the important task of opening and closing circuits. However, high-voltage circuit breakers may fail and be damaged during long-term use. According to statistical analysis, mechanical failure is still the most common failure.

[0003] Circuit breakers may cause mechanical failures during operation. Common mechanical failures include: idling of energy storage motors, demagnetization of tripping electromagnets, complete jamming of tripping coils, mechanical locks not unlocked, chain breakage, and loose nuts. Circuit breaker fault detection is critical to the safe operation of power systems. Traditional methods rely on regular inspections and manual testing, which are inefficient and prone to overlooking potential problems. Voiceprint detection provides new possibilities for circuit breaker fault detection. By monitoring the vibration and sound signals during the operation of the circuit breaker, combined with advanced signal processing technology and pattern recognition algorithms, real-time monitoring of the working status and fault diagnosis can be achieved. This method improves accuracy and efficiency while reducing maintenance costs, and plays an important supporting role in the safe and stable operation of power systems.

[0004] However, the quality of the fault audio signal is crucial for the subsequent voiceprint fault identification. If the quality of the fault audio signal is low, the subsequent voiceprint fault identification rate will be greatly reduced, and it cannot provide a strong guarantee for the safe and stable operation of the power grid. The voiceprint sensor needs to sample at full rate to ensure that the captured fault audio signal maintains its integrity and high quality, so it usually requires a large power consumption. This may limit the use time of the voiceprint sensor or increase power consumption loss, so it is necessary to design a new solution to meet actual needs. Summary of the invention

[0005] The purpose of the present invention is to provide a high-voltage circuit breaker mechanical fault audio signal cluster acquisition system and method based on adaptive filtering and sampling to address the above shortcomings, simulate the mechanical faults that may be caused during operation, collect and analyze the mechanical fault soundprints and improve the quality through the audio detection system based on the cluster structure, and reduce the system power consumption through adaptive sampling. It effectively solves the problems of low quality of traditional fault audio signal acquisition and serious system power loss.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is:

[0007] A high-voltage circuit breaker mechanical fault audio signal cluster acquisition system based on adaptive filtering and sampling, comprising:

[0008] Mechanical fault simulation module, used to simulate mechanical faults of circuit breaker during operation;

[0009] A fault audio signal transmission module, used for transmitting a fault audio signal;

[0010] An adaptive sampling module for sampling fault audio signals;

[0011] A noise processing module, used for performing noise processing on the sampled fault audio signal;

[0012] Adaptive filtering (LMS) module, used to filter and reduce noise on fault audio signals;

[0013] The audio fusion module is used to perform audio fusion on the fault audio signal and output the audio.

[0014] Furthermore, the mechanical fault simulation module simulates possible mechanical faults of the circuit breaker during operation, including idling of the energy storage motor, demagnetization of the tripping electromagnet, complete jamming of the tripping coil, failure to unlock the mechanical lock, broken chain and loose nuts.

[0015] Furthermore, the fault audio signal transmission module adopts a clustered sensor network to transmit the mechanical fault audio signal in different channels; an adaptive sampling module is provided in each channel, and sampling is performed through the adaptive sampling module; the adaptive sampling module has a switch controller, a signal analysis unit and a control algorithm unit;

[0016] A voiceprint sensor is also provided in each channel, and the voiceprint sensor is connected to the switch controller of the adaptive sampling module; the switch controller receives the signal or trigger signal generated by the voiceprint sensor, monitors and analyzes the audio signal collected by the voiceprint sensor in the clustered sensor network in real time, controls the state of the sampling switch according to preset conditions or algorithm logic, and dynamically determines whether to sample the audio signal, that is, to start or stop data sampling;

[0017] The adaptive sampling module controls the acquisition rate of the voiceprint sensor through the switch controller, and makes judgments on the signal-to-noise ratio and data arrival rate according to the actual situation on site, thereby controlling the acquisition rate of the voiceprint sensor.

[0018] Furthermore, the fault audio signal transmission module is composed of a sensor interface, a signal amplifier, an analog-to-digital converter and a communication interface; when working, the mechanical fault audio signal is collected through the sensor interface, and after amplifying and converting it into a digital signal, the signal is transmitted to the adaptive sampling module through the communication interface; this module ensures data integrity and low latency during audio signal transmission, and avoids data loss due to signal attenuation or noise interference.

[0019] Furthermore, when the adaptive sampling module is working, according to the on-site channel conditions, the switch controller simulates the attenuation in the signal transmission process by multiplying the amplitude of the audio signal collected by the voiceprint sensor by a constant that meets the channel conditions, which is expressed as a sample value of the signal attenuation in the channel. The adaptive sampling module decides whether to perform sampling based on the signal-to-noise ratio and data arrival rate according to the actual on-site conditions.

[0020] Furthermore, the noise processing module creates an audio signal containing noise by adding simulated Gaussian white noise to the original audio signal, and transmits the noisy audio signal to the adaptive filtering module for filtering and noise reduction processing, thereby simulating a real noise environment, making the subsequent filtering process have practical significance.

[0021] Furthermore, the adaptive filtering module has a second-order filter, an error calculation unit and an adaptive filtering (LMS) algorithm control unit. By calculating the error between the current filtered output and the target output and adjusting the coefficient of the second-order filter according to the error, the error is gradually minimized, and the noise is effectively filtered out in a dynamically changing environment while retaining important fault feature information; the filtering algorithm of the adaptive filtering module adopts the Least Mean Square algorithm.

[0022] The second-order filter has two cascaded filtering stages, each stage includes a delay element and a weight coefficient; by selecting and adjusting the weight coefficient, the input signal can be effectively adjusted in the time domain and frequency domain to achieve filtering and denoising of the signal.

[0023] Furthermore, the second-order filter adopts an improved variable step-size algorithm to implement the filtering algorithm of the least mean square algorithm, specifically: adjusting the filter response in real time according to the characteristics of the input signal, and improving the convergence speed and stability by dynamically adjusting the step-size parameter.

[0024] Furthermore, before the audio signal is input into the adaptive filtering module, a front signal-to-noise ratio calculation module is provided for calculating the signal-to-noise ratio of the audio signal containing noise. The noise signal with a calculated signal-to-noise ratio higher than the average signal-to-noise ratio is not processed, and the noise signal with a signal-to-noise ratio lower than the average signal-to-noise ratio continues to be input into the adaptive filtering module for adaptive filtering processing, thereby reducing the overall power consumption of the system.

[0025] Furthermore, the audio fusion module has an MRC (maximal ratio combining) phase modulator, which performs phase modulation and maximum ratio combining on the received filtered signal for audio output; specifically, it adjusts the phase and amplitude of the received signal to maximize the quality and amplitude of the combined signal; and adopts complex phase adjustment and signal merging technology to ensure that the optimal mixed signal is extracted from multiple inputs, thereby optimizing the effect and clarity of audio fusion.

[0026] Furthermore, maximum ratio combining is performed in the audio fusion module, and the weight G is normalized. At the receiving end, multiple diversity branches are phase-adjusted and added in phase according to appropriate gain coefficients to output an audio signal after maximum ratio combining. According to actual use needs, the output audio signal can be sent to a detector for detection to determine the type of fault.

[0027] The present invention also provides a method for clustering audio signals of mechanical faults of high-voltage circuit breakers based on adaptive filtering and sampling, comprising the following steps:

[0028] S1, simulating a mechanical failure of a circuit breaker during operation, so that an audio signal of the mechanical failure is transmitted in a clustered sensor network and sampled;

[0029] S2, simulating the on-site noise environment, and performing noise processing on the audio signal sampled in the above steps;

[0030] S3, transmitting the above-mentioned audio signal after the noise addition processing to the adaptive filtering module for filtering and noise reduction processing, and calculating the signal-to-noise ratio of the processed or unprocessed audio signal;

[0031] S4. According to the audio signal and signal-to-noise ratio obtained in the above steps, the energy consumption formula of the voiceprint sensor is optimized, and the channels requiring audio fusion and the rate of the voiceprint sensor are calculated through the heuristic algorithm of adaptive sampling, and the maximum ratio merging is performed to output the fault sound source.

[0032] Furthermore, the mechanical failures include idling of the energy storage motor, demagnetization of the tripping electromagnet, complete jamming of the tripping coil, failure to unlock the mechanical lock, broken chain and loose nuts.

[0033] Furthermore, step S1 specifically includes: the mechanical fault audio is transmitted in several channels, and the attenuation in the signal transmission process is simulated by multiplying the signal amplitude by a constant, which represents the signal attenuation sample value of a certain channel of the system, and the switch controller of the adaptive sampling module decides whether to perform sampling.

[0034] Furthermore, step S2 specifically includes: adding simulated Gaussian white noise to the original sampled audio signal to create a signal containing noise.

[0035] The Gaussian white noise has a mean of 0 and a variance of 0.01 to 0.05, and is used to simulate the noise environment of the equipment around the high-voltage circuit breaker. The best value is that the mean of the Gaussian white noise is 0 and the variance is 0.025.

[0036] Furthermore, step S3 specifically includes:

[0037] Calculating the signal-to-noise ratio of the audio signal after the noise processing to obtain an average signal-to-noise ratio; the audio signal after the noise processing is also referred to as a noise signal in the present invention;

[0038] The noise signal with a signal-to-noise ratio lower than the average value is processed by LMS (Least Mean Squares algorithm) adaptive filtering. By adjusting the weight coefficient of the filter, the mean square error between the output signal of the filter and the expected signal is minimized, thereby improving the signal-to-noise ratio of the voiceprint signal.

[0039] Furthermore, the minimum mean difference algorithm adopts an improved variable step size filtering algorithm (adaptive variable step size filtering algorithm) to solve the second-order filter in the adaptive filtering module. The specific steps are as follows:

[0040] 1. Initialize the weight coefficient of the filter to zero;

[0041] 2. Adjust the weight coefficient according to the error between the actual output and the expected output;

[0042] y(n)=x(n)w T (n)

[0043] e(n)=d(n)-y(n)

[0044] w(n+1)=w(n)+2μe(n)x(n)

[0045] Among them, x(n) is the input signal, y(n) is the output signal, d(n) is the expected signal, e(n) is the error between the expected signal and the output signal, w(n) is the weight coefficient (vector) at the nth step, w(n+1) is the weight coefficient (vector) at the n+1th step, w T (n) is the transpose of the weight vector at the nth step, μe(n) is the product of the step factor and the error signal, μ(n) is the step factor, and the condition for the algorithm to converge is 0≤μ≤λ max , where λ max is the maximum eigenvalue of the autocorrelation matrix of x(n);

[0046] According to the variable step length guidance adjustment rule, the optimal variable step length formula is obtained as follows:

[0047]

[0048] Among them, the parameter α controls the response shape of the step factor μ(n) to the error e(n), affecting the adjustment speed and method under different errors; the parameter θ controls the value range of the step factor μ(n), affecting the maximum adjustment ability under different errors;

[0049] 3. Through the iterative process, when the error is small enough or reaches the preset convergence condition, the algorithm stops iterating.

[0050] Furthermore, the acquisition rate assigned to the voiceprint sensor in step S4 must meet the expected value η of the audio signal signal-to-noise ratio, that is, the signal-to-noise ratio at a certain moment after merging is greater than the expected value η; and the data loss rate is less than the critical value β of the data loss rate.

[0051] The heuristic algorithm for the adaptive sampling is:

[0052] Set the expected signal-to-noise ratio threshold so that branches with a signal-to-noise ratio higher than the expected signal-to-noise ratio threshold are sorted according to data arrival rate; branches with a signal-to-noise ratio lower than the expected signal-to-noise ratio threshold are sorted according to the signal-to-noise ratio. If the signal-to-noise ratios are the same, they are sorted according to data reachability.

[0053] The rate of each voiceprint sensor is calculated through theorem and corollary.

[0054] The specific steps of the algorithm include:

[0055] Q2-1, assuming that N branches are collected, the normalized collection rate of the i-th branch is r i , the data loss rate P of the i-th branch loss =1-r i p i ;

[0056] Assuming the total number of sampling points is N, the data loss rate of N branches is

[0057] For the rate r of the voiceprint sensor i , since the rate assigned to each channel sensor is related to the data loss rate of each channel, and the minimum data loss rate must be less than the critical value P loss <β, if the maximum ratios of the first N-1 branches cannot meet the signal-to-noise ratio requirement after being combined, the first N-1 branches are sampled at full rate to increase the signal sample value, and the rate of the voiceprint sensor of the Nth branch is Then the acquisition rate of the voiceprint sensors in the N branches satisfies {1,1,1,...,r N , 0, 0, ..., 0};

[0058] Q2-2: sort the branches with a signal-to-noise ratio higher than the expected signal-to-noise ratio threshold according to data arrival rate; sort the branches with a signal-to-noise ratio lower than the expected signal-to-noise ratio threshold according to the signal-to-noise ratio. If the signal-to-noise ratios are the same, sort them according to data reachability rate.

[0059] Q2-3, assuming that the branches after arrangement are arranged in order, for the first branch:

[0060] Q2-3-1, if the signal-to-noise ratio of the first branch meets the expected signal-to-noise ratio and the data loss rate is also less than the critical value, it means that the first branch can meet the system acquisition requirements. Then only the voiceprint sensor on the first branch is needed for acquisition, and the acquisition rate is 1. The acquisition rates of the voiceprint sensors on the other branches are 0;

[0061] Q2-3-2, if the signal-to-noise ratio of the first branch does not meet the expected signal-to-noise ratio or the data loss rate is greater than the critical value, it means that in order to make the first branch collect at full rate, the first branch needs to be combined with other branches at the maximum ratio to meet the system requirements. The fused signal-to-noise ratio is calculated for each maximum ratio combination with a branch.

[0062] If the fusion signal-to-noise ratio meets the expected signal-to-noise ratio requirement, then calculate whether the data loss rate meets the requirement. If so, then the rate of the Nth branch voiceprint sensor If not satisfied,

[0063]

[0064] If the fusion signal-to-noise ratio does not meet the expected signal-to-noise ratio requirement, go to step Q2-3-4;

[0065] Q2-3-4, and the next branch in the arrangement perform maximum ratio merging to calculate the fusion signal-to-noise ratio. If the fusion signal-to-noise ratio meets the expected signal-to-noise ratio requirement, calculate whether the data loss rate meets the requirement. If so, the rate of the voiceprint sensor of the Nth branch is If not satisfied, If the fused signal-to-noise ratio does not meet the expected signal-to-noise ratio requirement, proceed to step Q2-3-4 and perform maximum ratio merging again with the next branch in the arrangement;

[0066] If the fused signal-to-noise ratio of the last branch still cannot reach the expected signal-to-noise ratio, the last branch is sampled at full rate to meet the expected signal-to-noise ratio requirement as much as possible.

[0067] The beneficial effects of the present invention are:

[0068] (1) Aiming at the shortcomings of traditional audio acquisition systems such as low audio signal acquisition quality, an audio detection system based on cluster structure is proposed. This method has high robustness, detection accuracy and processing efficiency.

[0069] (2) By studying two special cases of high signal-to-noise ratio and low signal-to-noise ratio, a heuristic algorithm for adaptive sampling is synthesized. According to the results, it can be seen that the heuristic algorithm for adaptive sampling not only meets the system's requirements for audio signal quality, but also greatly reduces the overall power consumption of the system.

[0070] (3) Aiming at the shortcomings of the traditional LMS adaptive filtering algorithm, such as slow convergence speed and large steady-state error, a variable step-size LMS adaptive filtering algorithm is proposed. This method not only has fast convergence speed but also can improve steady-state performance. This method has better robustness and adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] The present invention will be further described below in conjunction with the accompanying drawings. The accompanying drawings are only used to illustrate preferred embodiments and are not to be considered as limiting the present application.

[0072] Figure 1 This is a schematic diagram of the acquisition system architecture of the present invention;

[0073] Figure 2 A simplified flow chart of the steps of the collection method of the present invention;

[0074] Figure 3 A simplified flow chart of the steps of the collection method of the present invention;

[0075] Figure 4 It is a logic diagram of the working principle of the acquisition system of the present invention;

[0076] Figure 5 A time domain diagram of a mechanical fault signal of a high-voltage circuit breaker simulated in the present invention;

[0077] Figure 6 It is a time domain diagram of the signal after the adaptive filtering process in the present invention;

[0078] Figure 7 is an algorithm flow chart of the adaptive algorithm in the present invention;

[0079] Figure 8 This is a power consumption comparison diagram of the heuristic adaptive sampling and oversampling in the present invention. DETAILED DESCRIPTION

[0080] The following embodiments of the technical solution of the present application will be described in detail in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present application, and are therefore only used as examples, and cannot be used to limit the scope of protection of the present application.

[0081] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by technicians in the technical field to which this application belongs; the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit this application; the term "including" and any variations thereof in the specification and claims of this application and the above-mentioned figure descriptions are intended to cover non-exclusive inclusions.

[0082] The present invention provides a high-voltage circuit breaker mechanical fault audio signal cluster acquisition system and method based on adaptive filtering and sampling, which simulates the mechanical faults that may be caused during operation, collects and analyzes the mechanical fault soundprints and improves the quality through the audio detection system based on the cluster structure, and reduces the system power consumption through adaptive sampling. It effectively solves the problems of low quality of traditional fault audio signal acquisition and serious system power consumption loss.

[0083] The present invention performs maximum ratio combining in the audio fusion area, and the weight G is normalized. At the receiving end, multiple diversity branches are phase-adjusted and added in phase according to appropriate gain coefficients, and then sent to the detector for detection.

[0084] The present invention also introduces an adaptive sampling heuristic algorithm. While ensuring high-quality collection and transmission of audio data, the acquisition rate of the voiceprint sensor on the channel is dynamically adjusted by analyzing the critical value of the data loss rate and the expected value of the audio signal-to-noise ratio. The algorithm ensures that the energy consumption formula of the voiceprint sensor is optimized to reach the minimum value under the conditions of meeting the preset data loss rate threshold and the expected value of the audio signal-to-noise ratio, and transforms the complex power consumption problem into the problem of seeking the optimal solution. The energy efficiency and sustainability of the system are improved. To this end, the present invention studies the adaptive sampling heuristic algorithm to limit the power consumption loss of the voiceprint sensor.

[0085] The data loss rate refers to the ratio of data packets or information lost during data transmission or processing to the total amount of data sent. It is usually used to measure the reliability and stability of data transmission and is one of the important indicators for evaluating the efficiency of data communication systems. The present invention uses a special network monitoring device network analyzer to monitor network traffic, device status and link quality in real time to evaluate data reachability.

[0086] The expected value of the audio signal-to-noise ratio refers to the ratio of the audio signal to the background noise that is expected to be maintained during the subsequent fault voiceprint recognition process. A higher signal-to-noise ratio means that the audio signal is clearer and more obvious than the noise, which helps to improve the accuracy and reliability of the voiceprint recognition system.

[0087] Embodiment 1:

[0088] See attached Figure 1, a high-voltage circuit breaker mechanical fault audio signal cluster acquisition system based on adaptive filtering and sampling, comprising:

[0089] Mechanical fault simulation module, used to simulate mechanical faults of circuit breaker during operation;

[0090] A fault audio signal transmission module, used for transmitting a fault audio signal;

[0091] An adaptive sampling module for sampling fault audio signals;

[0092] A noise processing module, used for performing noise processing on the sampled fault audio signal;

[0093] Adaptive filtering (LMS) module, used to filter and reduce noise on fault audio signals;

[0094] The audio fusion module is used to perform audio fusion on the fault audio signal and output the audio.

[0095] The fault audio signal transmission module adopts a clustered sensor network to transmit the mechanical fault audio signal in different channels; an adaptive sampling module is provided in each channel to perform sampling; the adaptive sampling module has a switch controller, a signal analysis unit and a control algorithm unit;

[0096] A voiceprint sensor is also provided in each channel, and the voiceprint sensor is connected to the switch controller of the adaptive sampling module; the switch controller receives the signal or trigger signal generated by the voiceprint sensor, monitors and analyzes the audio signal collected by the voiceprint sensor in the clustered sensor network in real time, controls the state of the sampling switch according to preset conditions or algorithm logic, and dynamically determines whether to sample the audio signal, that is, to start or stop data sampling;

[0097] The adaptive sampling module controls the acquisition rate of the voiceprint sensor through the switch controller, and makes judgments on the signal-to-noise ratio and data arrival rate according to the actual situation on site, thereby controlling the acquisition rate of the voiceprint sensor.

[0098] The fault audio signal transmission module consists of a sensor interface, a signal amplifier, an analog-to-digital converter and a communication interface. When working, the mechanical fault audio signal is collected through the sensor interface, amplified and converted into a digital signal, and then the signal is transmitted to the adaptive sampling module through the communication interface. This module ensures data integrity and low latency during audio signal transmission, avoiding data loss due to signal attenuation or noise interference.

[0099] When the adaptive sampling module is working, according to the on-site channel conditions, the switch controller simulates the attenuation of the signal transmission process by multiplying the amplitude of the audio signal collected by the voiceprint sensor by a constant that meets the channel conditions, which is expressed as a sample value of the signal attenuation in the channel. The adaptive sampling module decides whether to perform sampling based on the actual signal-to-noise ratio and data arrival rate on site.

[0100] The noise processing module creates an audio signal containing noise by adding simulated Gaussian white noise to the original audio signal, and transmits the noisy audio signal to the adaptive filtering module for filtering and noise reduction processing, thereby simulating a real noise environment, making the subsequent filtering process meaningful.

[0101] The adaptive filtering module has a second-order filter, an error calculation unit and an adaptive filtering (LMS) algorithm control unit. It calculates the error between the current filter output and the target output and adjusts the coefficient of the second-order filter according to the error to gradually minimize the error, effectively filter out noise in a dynamically changing environment, and retain important fault feature information. The filtering algorithm of the adaptive filtering module adopts the Least Mean Squares algorithm.

[0102] The second-order filter has two cascaded filtering stages, each stage includes a delay element and a weight coefficient; by selecting and adjusting the weight coefficient, the input signal can be effectively adjusted in the time domain and frequency domain to achieve filtering and denoising of the signal.

[0103] The second-order filter adopts an improved variable step-size filtering algorithm to implement the filtering algorithm of the least mean square algorithm, specifically: adjusting the filter response in real time according to the characteristics of the input signal, and improving the convergence speed and stability by dynamically adjusting the step-size parameter.

[0104] Before the audio signal is input into the adaptive filtering module, a front signal-to-noise ratio calculation module is also provided, which is used to calculate the signal-to-noise ratio of the audio signal containing noise. The noise signal with a calculated signal-to-noise ratio higher than the average value is not processed, and the noise signal with a signal-to-noise ratio lower than the average value is continuously input into the adaptive filtering module for adaptive filtering processing, thereby reducing the overall power consumption of the system.

[0105] The audio fusion module has an MRC (maximal ratio combining) phase modulator, which performs phase modulation and maximum ratio combining on the received filtered signal for audio output; specifically, it adjusts the phase and amplitude of the received signal to maximize the quality and amplitude of the combined signal; and adopts complex phase adjustment and signal merging technology to ensure that the optimal mixed signal is extracted from multiple inputs, thereby optimizing the effect and clarity of audio fusion.

[0106] The maximum ratio combining is performed in the audio fusion module, and the weight G is normalized. At the receiving end, multiple diversity branches are phase-adjusted and added in phase according to appropriate gain coefficients to output the audio signal after maximum ratio combining.

[0107] Embodiment 2:

[0108] See attached Figure 2 A method for collecting audio signals of mechanical faults of high-voltage circuit breakers in clusters based on adaptive filtering and sampling includes the following steps:

[0109] S1, simulating a mechanical failure of a circuit breaker during operation, so that an audio signal of the mechanical failure is transmitted in a clustered sensor network and sampled;

[0110] S2, simulating the on-site noise environment, and performing noise processing on the audio signal sampled in the above steps;

[0111] S3, transmitting the above-mentioned audio signal after the noise addition processing to the adaptive filtering module for filtering and noise reduction processing, and calculating the signal-to-noise ratio of the processed or unprocessed audio signal;

[0112] S4. According to the audio signal and signal-to-noise ratio obtained in the above steps, the energy consumption formula of the voiceprint sensor is optimized, and the channels requiring audio fusion and the rate of the voiceprint sensor are calculated through the heuristic algorithm of adaptive sampling, and the maximum ratio merging is performed to output the fault sound source.

[0113] Embodiment 3:

[0114] In order to explain the implementation process of the acquisition method of the present invention in detail, Figure 3 The present invention is described in detail by an embodiment, and the method for collecting audio signals of a 10kV circuit breaker mechanical fault based on adaptive filtering and sampling includes the following steps:

[0115] S1. Simulating a mechanical fault of a 10 kV circuit breaker during operation, so that an audio signal of the mechanical fault is transmitted in a clustered sensor network and sampled;

[0116] S2, simulate a Gaussian white noise with a mean of 0 and a variance of 0.025, and add it to the audio signal sampled in the above step for noise addition processing;

[0117] S3, transmitting the above-mentioned audio signal after the noise addition processing to the adaptive filtering module for filtering and noise reduction processing, and calculating the signal-to-noise ratio of the processed or unprocessed audio signal;

[0118] S4. According to the audio signal and signal-to-noise ratio obtained in the above steps, the energy consumption formula of the voiceprint sensor is optimized, and the channels requiring audio fusion and the rate of the voiceprint sensor are calculated through the heuristic algorithm of adaptive sampling, and the maximum ratio merging is performed to output the fault sound source.

[0119] Embodiment 4:

[0120] See attached Figure 4 In this embodiment, three non-contact microphones are used as audio acquisition devices. The non-contact microphones are attached to three sides of the switch cabinet circuit breaker, and the acquisition environment is kept quiet. The simulated circuit breaker may cause mechanical failures during operation, including idling of the energy storage motor, demagnetization of the tripping electromagnet, complete jamming of the tripping coil, mechanical lock failure, chain breakage, and loose nuts. The time domain diagram of the simulated 10kV circuit breaker mechanical failure signal is shown in the figure below. Figure 5 shown.

[0121] Figure 4 In the example, j is the sample value at the sampling moment. In order to facilitate calculation, j in this embodiment is set to 1, representing the sample value at the first moment; m j represents the fault sound source of the 10kV circuit breaker at the jth moment; L si represents the fading sample value of the signal in the i-th channel of the system; K ij Indicates whether the system performs sampling at the jth time in the i-th channel, K ij =1 means sampling, K ij =0 means no sampling.

[0122] Mechanical fault audio is transmitted in i channels, each of which includes an adaptive sampling area (adaptive sampling module), an LMS filtering area (adaptive filtering module) and an audio fusion area (audio fusion module); the switch controller of the adaptive sampling module in the sampling system monitors and analyzes the audio signals collected by each sensor in real time, and dynamically decides whether to perform sampling; when the controller detects a possible fault signal, it will turn on the sampling switch to sample and transmit the signal.

[0123] Among them, M ij represents the amplitude of the signal at the jth moment in the ith channel of the system, b ij Represents the Gaussian white noise of the signal of the system at the jth moment in the i-th channel; in this embodiment, a Gaussian white noise with a mean of 0 and a variance of 0.025 is simulated, and the Gaussian white noise is added to the original audio signal to create a signal containing noise, and the noisy audio signal is transmitted to an adaptive second-order filter for filtering and noise reduction processing (since a real noise environment is simulated, the subsequent filtering process has practical significance); thereafter, the preprocessed audio signal is transmitted to the adaptive filtering module for further filtering and noise reduction processing.

[0124] M' ij It represents the actual output signal of the system after the second-order filter at the jth time in the i-th channel, d(n ij ) represents the expected output signal of the system after passing through the second-order filter at the jth time in the ith channel, e(n ij ) represents the error between the expected output signal and the actual output signal of the system at the jth time after the second-order filter at the i-th channel, w(n ij +1) represents the weight of the second-order filter of the system at the jth time in the i-th channel, p i It represents the data reachability rate of the system in the i-th channel.

[0125] The actual input signal M of the i-th channel at the j-th moment before filtering ij =m j L si K ij +b ij ;

[0126] Signal power of the i-th channel at the j-th moment before filtering

[0127] Noise power of the ith channel at the jth moment before filtering

[0128] σ ij It represents the standard deviation of the noise at the jth moment in the ith channel before filtering;

[0129] The signal-to-noise ratio of the i-th channel at the j-th time before filtering

[0130] The actual output signal of the i-th channel at the j-th moment after filtering

[0131] M' ij =m j L si K ij +w T (n i )b ij

[0132] Among them, w(n ij ) is the coefficient vector of the second-order filter of the i-th channel at the j-th moment; the updated weight is

[0133] w(n i +1)=w(n i )+2μe(n i )M ij

[0134] Among them, w(n i ) is the weight coefficient (vector) of the ith channel at the nth step, w(n i+1) is the weight coefficient (vector) of the ith channel at the n+1th step, w T (n i ) is the transpose of the weight vector of the ith channel at the nth step, μe(n i ) is the product of the i-th channel step factor and the error signal.

[0135] The optimal variable step length formula is

[0136] e(n) is the error; in the present invention, according to the time-frequency domain characteristics of the 10kV circuit breaker fault signal and the on-site requirements, the value range of θ is 0.375-1.5, the value range of α is 0.5625-2.25, and according to the experiment, the optimal values ​​are θ=0.75, α=1.125.

[0137] The expected output signal d(n) of the i-th channel at the j-th moment after filtering ij )=m j L si K ij ;

[0138] The error is e(n ij )=d(n ij )-M' ij =m j L si K ij -(m j L si K ij +w T (n i )b ij );

[0139] The signal power of the i-th channel at the j-th moment after filtering

[0140] The noise suppression ratio λ reflects the degree of noise reduction of the filter. The noise power after filtering is It can be expressed as Ideally, the noise suppression ratio λ should be much smaller than 1, indicating that the noise is significantly suppressed. Considering the performance of the filter, it can be approximately considered that the more optimized the filter coefficient vector is, the smaller the noise suppression ratio is. Therefore, there is an approximate relationship between the noise suppression ratio λ and the filter coefficient vector. If the filter coefficient vector w T (n i ) represents the response of the filter to the input signal, then the modulus of the coefficient vector can be used to represent the strength of the filter, assuming that there is an approximate relationship between the modulus of the coefficient vector and λ

[0141] λ=|w T (n i )

[0142] The noise power of the i-th channel at the j-th moment after filtering

[0143] Therefore, the signal-to-noise ratio of the i-th channel at the j-th moment after filtering is

[0144]

[0145] The time domain diagram of the signal after being processed by the adaptive variable step size filtering algorithm (variable step size filtering algorithm) is as follows Figure 6 As shown in the figure, it can be seen from the waveform that after the adaptive variable step size filtering process, the noise in the mechanical fault signal of the 10kV circuit breaker is significantly reduced and the signal-to-noise ratio is significantly improved.

[0146] like Figure 4 As shown, G ij represents the weight of the system at the jth moment of the ith channel after MRC (maximum ratio combining) phase modulation, m' j It represents the fault sound source obtained after the system adjustment of the 10kV circuit breaker at the jth moment.

[0147] The audio amplitude M at the jth moment after MRC merging j =(m j L si K ij +w T (n i )b ij )G ij ;

[0148] Signal power at time j after MRC merging

[0149] Noise power at time j after MRC merging

[0150] If the signal-to-noise ratio at the jth moment after the filtered MRC merging is

[0151]

[0152] If there is no filtering after MRC merging, the signal-to-noise ratio at the jth moment is

[0153]

[0154] In the heuristic algorithm of adaptive sampling in step S4, in order to make the fused signal-to-noise ratio satisfy q" j>η, the present invention first selects some nodes with high signal-to-noise ratio; when the first requirement is met, according to the theorem, nodes with high data arrival rate are selected to meet the second requirement. In general, the present invention first sets a threshold of expected signal-to-noise ratio, and nodes above the threshold of expected signal-to-noise ratio are sorted according to data arrival rate; nodes below the threshold of expected signal-to-noise ratio are sorted according to signal-to-noise ratio, and if the signal-to-noise ratio is the same, they are sorted according to data reachability rate; then the rate r of each voiceprint sensor is calculated by theorem and inference. i Perform calculations.

[0155] The specific implementation method of the heuristic algorithm for adaptive sampling is as follows:

[0156] According to the defined network model and the practical problem to be solved, in the system of the present invention, the fusion node only needs to allocate the acquisition rate to some voiceprint sensor nodes to meet the coordinator's expected value η of the audio signal signal-to-noise ratio, that is, the signal-to-noise ratio q at the jth moment after merging" j >η; the data loss rate is less than the critical value of the data loss rate β, that is, P loss <β; some voiceprint sensor nodes only need to report according to the assigned collection rate, so the energy consumption of the voiceprint sensor can be reflected as:

[0157] Through the above analysis, the problem can be transformed into an optimal solution problem, that is, The minimum value of E min , the constraints are:

[0158]

[0159] The system has two requirements q" j >η and P loss <β;

[0160] By discussing two special cases, the present invention proposes a heuristic algorithm for sampling rate allocation.

[0161] like Figure 7 As shown, the signal-to-noise ratios and data reachability rates in the three branches are sorted, and the channels with signal-to-noise ratios higher than the expected value are sorted from high to low according to the data reachability rates; the channels with signal-to-noise ratios lower than the expected value are sorted from high to low according to the signal-to-noise ratios.

[0162] Assume that the order of the channels after sorting is 1, 2, and 3, and set the expected minimum signal-to-noise ratio after fusion to η according to the subsequent voiceprint fault identification;

[0163] Case 1: If the signal-to-noise ratio of the audio signal in the first channel is q1>η and the data loss rate is also less than the threshold value, only the first channel is needed to complete the subsequent fault identification work. r2=0, r3=0;

[0164] Case 2: If the signal-to-noise ratio q1 of the audio signal in the first channel is less than η or the data loss rate is greater than the threshold value, then only accepting the sample value of one channel cannot meet the audio quality requirements of the subsequent fault identification work, and the signal-to-noise ratio q2 of the audio signal in the second channel needs to be calculated;

[0165] After the signal in the first channel and the signal in the second channel are combined by the maximum ratio,

[0166] Case 2-1, if the fusion signal-to-noise ratio q 12 >η and the data loss rate is also less than the threshold value, then the first channel and the second channel need to work together to complete the subsequent fault identification work. At this time, r1=1, r3=0; if r2=1, a third node is required to sample at a partial rate;

[0167] Case 2-2: If the fusion signal-to-noise ratio meets the system requirements but the data arrival rate does not meet the requirements, then r1 = 1. r3=0; if r2=1, a third node is also required to sample at a partial rate;

[0168] Case 2-3: If the fusion signal-to-noise ratio does not meet the system requirements, continue to merge with the third node for the maximum ratio. If the fusion signal-to-noise ratio q 123 >η and the data loss rate is also less than the threshold value, then r1=1, r2=1, If the fusion signal-to-noise ratio meets the system requirements but the data arrival rate does not meet the requirements, then r1=1, r2=1, For the extreme case, if the signal-to-noise ratio q after the three branches are fused 123 <η, then let the voiceprint sensors of all channels sample at full rate to maximize the audio quality and complete the subsequent audio recognition work.

[0169] It should be noted that 10kV circuit breakers, as important electrical equipment in the power system, have the dual functions of optimizing the grid structure and protecting the grid safety. Such extreme situations rarely occur.

[0170] The sampling power consumption of the heuristic algorithm of adaptive sampling in the present invention is compared with the power consumption of conventional oversampling in the art, such as Figure 8 As shown in the results, it can be seen that the sampling power consumption of the adaptive sampling heuristic algorithm is much lower than that of the oversampling system. It not only ensures that the captured fault audio signal maintains its integrity and high quality, but also increases the battery life of the voiceprint sensor. The design cost is greatly reduced, which actively responds to the requirements of the sustainable development strategy.

[0171] Although the preferred embodiments of the present invention have been described above, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0172] Obviously, those skilled in the art can make various changes and modifications to the embodiments of the present invention without departing from the spirit and scope of the embodiments of the present invention. Thus, if these modifications and variations of the embodiments of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A high-voltage circuit breaker mechanical fault audio signal cluster acquisition system based on adaptive filtering and sampling, characterized in that: include: Mechanical fault simulation module, used to simulate mechanical faults of circuit breaker during operation; A fault audio signal transmission module, used for transmitting a fault audio signal; An adaptive sampling module for sampling fault audio signals; A noise processing module, used for performing noise processing on the sampled fault audio signal; An adaptive filtering module, used to filter and reduce noise on fault audio signals; The audio fusion module is used to perform audio fusion on the fault audio signal and output the audio.

2. The high-voltage circuit breaker mechanical fault audio signal cluster acquisition system based on adaptive filtering and sampling according to claim 1 is characterized in that: The mechanical faults simulated by the mechanical fault simulation module include idling of the energy storage motor, demagnetization of the opening electromagnet, complete jamming of the opening coil, failure to unlock the mechanical lock, broken chain and loose nut.

3. The high-voltage circuit breaker mechanical fault audio signal cluster acquisition system based on adaptive filtering and sampling according to claim 1 is characterized in that: The fault audio signal transmission module adopts a clustered sensor network to transmit the mechanical fault audio signal in different channels; An adaptive sampling module is provided in each channel, and sampling is performed through the adaptive sampling module; the adaptive sampling module has a switch controller, a signal analysis unit and a control algorithm unit; A voiceprint sensor is also provided in each channel, and the voiceprint sensor is connected to the switch controller of the adaptive sampling module; the switch controller receives the signal or trigger signal generated by the voiceprint sensor, monitors and analyzes the audio signal collected by the voiceprint sensor in the clustered sensor network in real time, controls the state of the sampling switch according to preset conditions or algorithm logic, and dynamically determines whether to sample the audio signal, that is, to start or stop data sampling; The adaptive sampling module controls the acquisition rate of the voiceprint sensor through the switch controller, and makes judgments on the signal-to-noise ratio and data arrival rate according to the actual situation on site, thereby controlling the acquisition rate of the voiceprint sensor.

4. The high-voltage circuit breaker mechanical fault audio signal cluster acquisition system based on adaptive filtering and sampling according to claim 3 is characterized in that: When the adaptive sampling module is working, according to the on-site channel conditions, the switch controller simulates the attenuation of the signal transmission process by multiplying the amplitude of the audio signal collected by the voiceprint sensor by a constant that meets the channel conditions, which is expressed as a sample value of the signal attenuation in the channel. The adaptive sampling module decides whether to perform sampling based on the actual signal-to-noise ratio and data arrival rate on site.

5. The high-voltage circuit breaker mechanical fault audio signal cluster acquisition system based on adaptive filtering and sampling according to claim 1 or 3, characterized in that: The fault audio signal transmission module consists of a sensor interface, a signal amplifier, an analog-to-digital converter and a communication interface. When working, the mechanical fault audio signal is collected through the sensor interface, amplified and converted into a digital signal, and then the signal is transmitted to the adaptive sampling module through the communication interface. The fault audio signal transmission module ensures data integrity and low latency during audio signal transmission, avoiding data loss due to signal attenuation or noise interference.

6. The high-voltage circuit breaker mechanical fault audio signal cluster acquisition system based on adaptive filtering and sampling according to claim 5 is characterized in that: The noise processing module creates an audio signal containing noise by adding simulated Gaussian white noise to the original audio signal, and transmits the noisy audio signal to the adaptive filtering module for filtering and noise reduction processing, thereby simulating a real noise environment, making the subsequent filtering process meaningful.

7. The high-voltage circuit breaker mechanical fault audio signal cluster acquisition system based on adaptive filtering and sampling according to claim 6 is characterized in that: The adaptive filtering module has a second-order filter, an error calculation unit and an adaptive filtering algorithm control unit. By calculating the error between the current filter output and the target output and adjusting the coefficient of the second-order filter according to the error, the error is gradually minimized, effectively filtering out noise in a dynamically changing environment while retaining important fault feature information; the filtering algorithm of the adaptive filtering module adopts the least mean square algorithm; The second-order filter has two cascaded filtering stages, each stage includes a delay element and a weight coefficient; by selecting and adjusting the weight coefficient, the input signal is adjusted in the time domain and the frequency domain to achieve filtering and denoising of the signal.

8. The high-voltage circuit breaker mechanical fault audio signal cluster acquisition system based on adaptive filtering and sampling according to claim 7 is characterized in that: The second-order filter adopts an improved variable step-size algorithm to implement the filtering algorithm of the least mean square algorithm, specifically: the filter response is adjusted in real time according to the characteristics of the input signal, and the convergence speed and stability are improved by dynamically adjusting the step-size parameter.

9. The high-voltage circuit breaker mechanical fault audio signal cluster acquisition system based on adaptive filtering and sampling according to claim 6, characterized in that: Before the audio signal is input into the adaptive filtering module, a front signal-to-noise ratio calculation module is also provided, which is used to calculate the signal-to-noise ratio of the audio signal containing noise. The noise signal with a calculated signal-to-noise ratio higher than the average value is not processed, and the noise signal with a signal-to-noise ratio lower than the average value is continuously input into the adaptive filtering module for adaptive filtering processing, thereby reducing the overall power consumption of the system.

10. The high-voltage circuit breaker mechanical fault audio signal cluster acquisition system based on adaptive filtering and sampling according to claim 6, characterized in that: The audio fusion module has an MRC maximum ratio combining phase modulator, which performs phase modulation and maximum ratio combining on the received filtered signal for audio output; specifically, it adjusts the phase and amplitude of the received signal to maximize the quality and amplitude of the combined signal; and adopts phase adjustment and signal combining technology to ensure that the optimal mixed signal is extracted from multiple inputs, thereby optimizing the effect and clarity of audio fusion.

11. The high-voltage circuit breaker mechanical fault audio signal cluster acquisition system based on adaptive filtering and sampling according to claim 10, characterized in that: The maximum ratio combining is performed in the audio fusion module, and the weight G is normalized. At the receiving end, multiple diversity branches are phase-adjusted and added in phase according to appropriate gain coefficients to output the audio signal after maximum ratio combining.

12. The acquisition method of the high-voltage circuit breaker mechanical fault audio signal cluster acquisition system based on adaptive filtering and sampling as claimed in claim 1, characterized in that: The steps include: S1, simulating a mechanical failure of a circuit breaker during operation, so that an audio signal of the mechanical failure is transmitted in a clustered sensor network and sampled; S2, simulating the on-site noise environment, and performing noise processing on the audio signal sampled in the above steps; S3, transmitting the above-mentioned audio signal after the noise addition processing to the adaptive filtering module for filtering and noise reduction processing, and calculating the signal-to-noise ratio of the processed or unprocessed audio signal; S4. According to the audio signal and signal-to-noise ratio obtained in the above steps, the energy consumption formula of the voiceprint sensor is optimized, and the channels requiring audio fusion and the rate of the voiceprint sensor are calculated through the heuristic algorithm of adaptive sampling, and the maximum ratio merging is performed to output the fault sound source.

13. The method for clustering audio signals of mechanical faults of high-voltage circuit breakers based on adaptive filtering and sampling according to claim 12, characterized in that: Step S1 specifically includes: the mechanical fault audio is transmitted in several channels, and the attenuation in the signal transmission process is simulated by multiplying the signal amplitude by a constant, which represents the attenuation sample value of the system signal in a certain channel. The switch controller of the adaptive sampling module decides whether to perform sampling.

14. The method for clustering audio signals of mechanical faults of high-voltage circuit breakers based on adaptive filtering and sampling according to claim 13, characterized in that: Step S2 specifically includes: adding simulated Gaussian white noise to the original sampled audio signal to create a signal containing noise.

15. The method for clustering audio signal of mechanical fault of high-voltage circuit breaker based on adaptive filtering and sampling according to claim 14, characterized in that: Step S3 specifically includes: Calculating the signal-to-noise ratio of the audio signal after the noise processing to obtain an average signal-to-noise ratio; the audio signal after the noise processing is also referred to as a noise signal in the present invention; The noise signal with a signal-to-noise ratio lower than the average value is adaptively filtered using the LMS minimum mean difference algorithm. By adjusting the weight coefficient of the filter, the mean square error between the output signal of the filter and the expected signal is minimized, thereby improving the signal-to-noise ratio of the voiceprint signal.

16. The method for clustering audio signal of mechanical fault of high-voltage circuit breaker based on adaptive filtering and sampling according to claim 15, characterized in that: The minimum mean difference algorithm adopts an improved variable step size filtering algorithm to solve the second-order filter in the adaptive filtering module. The specific steps are as follows:

1. Initialize the weight coefficient of the filter to zero; 2. Adjust the weight coefficient according to the error between the actual output and the expected output; y(n)=x(n)w T (n) e(n)=d(n)-y(n) w(n+1)=w(n)+2μe(n)x(n) Among them, x(n) is the input signal, y(n) is the output signal, d(n) is the expected signal, e(n) is the error between the expected signal and the output signal, w(n) is the weight coefficient (vector) at the nth step, w(n+1) is the weight coefficient (vector) at the n+1th step, w T (n) is the transpose of the weight vector at the nth step, μe(n) is the product of the step factor and the error signal, μ(n) is the step factor, and the condition for the algorithm to converge is 0≤μ≤λ max , where λ max is the maximum eigenvalue of the autocorrelation matrix of x(n); According to the variable step length guidance adjustment rule, the optimal variable step length formula is obtained as follows: Among them, the parameter α controls the response shape of the step factor μ(n) to the error e(n), affecting the adjustment speed and method under different errors; the parameter θ controls the value range of the step factor μ(n), affecting the maximum adjustment ability under different errors; 3. Through the iterative process, when the error is small enough or reaches the preset convergence condition, the algorithm stops iterating.

17. The method for clustering audio signals of mechanical faults of high-voltage circuit breakers based on adaptive filtering and sampling according to claim 16, characterized in that: The acquisition rate assigned to the voiceprint sensor in step S4 must meet the expected value η of the audio signal signal-to-noise ratio, that is, the signal-to-noise ratio at a certain moment after merging is greater than the expected value η; the data loss rate is less than the critical value β of the data loss rate; The heuristic algorithm for the adaptive sampling is: Set the expected signal-to-noise ratio threshold so that branches with a signal-to-noise ratio higher than the expected signal-to-noise ratio threshold are sorted according to data arrival rate; branches with a signal-to-noise ratio lower than the expected signal-to-noise ratio threshold are sorted according to the signal-to-noise ratio. If the signal-to-noise ratios are the same, they are sorted according to data reachability. Calculate the rate of each voiceprint sensor through theorem and corollary; The specific steps of the adaptive sampling heuristic algorithm include: Q2-1, assuming that N branches are collected, the normalized collection rate of the i-th branch is r i , the data loss rate P of the i-th branch loss =1-r i p i ; Assuming the total number of sampling points is N, the data loss rate of N branches is For the rate r of the voiceprint sensor i , since the rate assigned to each channel sensor is related to the data loss rate of each channel, and the minimum data loss rate must be less than the critical value P loss <β, if the maximum ratios of the first N-1 branches cannot meet the signal-to-noise ratio requirement after being combined, the first N-1 branches are sampled at full rate to increase the signal sample value, and the rate of the voiceprint sensor of the Nth branch is Then the acquisition rate of the voiceprint sensors in the N branches satisfies {1,1,1,...,r N , 0, 0, ..., 0}; Q2-2: sort the branches with a signal-to-noise ratio higher than the expected signal-to-noise ratio threshold according to data arrival rate; sort the branches with a signal-to-noise ratio lower than the expected signal-to-noise ratio threshold according to the signal-to-noise ratio. If the signal-to-noise ratios are the same, sort them according to data reachability rate. Q2-3, assuming that the branches after arrangement are arranged in order, for the first branch: Q2-3-1, if the signal-to-noise ratio of the first branch meets the expected signal-to-noise ratio and the data loss rate is also less than the critical value, it means that the first branch can meet the system acquisition requirements. Then only the voiceprint sensor on the first branch is needed for acquisition, and the acquisition rate is 1. The acquisition rates of the voiceprint sensors on the other branches are 0; Q2-3-2, if the signal-to-noise ratio of the first branch does not meet the expected signal-to-noise ratio or the data loss rate is greater than the critical value, it means that in order to make the first branch collect at full rate, the first branch needs to be combined with other branches at the maximum ratio to meet the system requirements. The fused signal-to-noise ratio is calculated for each maximum ratio combination with a branch. If the fusion signal-to-noise ratio meets the expected signal-to-noise ratio requirement, then calculate whether the data loss rate meets the requirement. If so, then the rate of the Nth branch voiceprint sensor If not satisfied, If the fusion signal-to-noise ratio does not meet the expected signal-to-noise ratio requirement, go to step Q2-3-4; Q2-3-4, and the next branch in the arrangement perform maximum ratio merging to calculate the fusion signal-to-noise ratio. If the fusion signal-to-noise ratio meets the expected signal-to-noise ratio requirement, calculate whether the data loss rate meets the requirement. If so, the rate of the voiceprint sensor of the Nth branch is If not satisfied, If the fused signal-to-noise ratio does not meet the expected signal-to-noise ratio requirement, proceed to step Q2-3-4 and perform maximum ratio merging again with the next branch in the arrangement; If the fused signal-to-noise ratio of the last branch still cannot reach the expected signal-to-noise ratio, the last branch is sampled at full rate to meet the requirement of the expected signal-to-noise ratio as much as possible.

Images