Sensor signal noise suppression method and system based on tunable filter

Abstract

The application provides a sensor signal noise suppression method and system based on an adjustable filter, and relates to the technical field of signal processing. The method comprises the following steps: predicting signal entropy increase of a sensor, making impedance control decisions under entropy flow control according to the signal entropy increase value, determining impedance control instructions and applying the impedance control instructions to an accessed impedance matching network, cooperating with operation of the impedance matching network, performing source end signal collection of the sensor, determining a first sensor signal, performing frequency resonance detection on the first sensor signal by deploying a lightweight resonance detection array, performing noise suppression control according to a resonance detection result, and determining a filtering result. The technical problem that the sensor signal noise components are complex and dynamically change in the prior art, resulting in low sensor signal noise suppression efficiency and suppression quality is solved. Efficient and adaptive sensor signal noise suppression is achieved, and the technical effect of improving the quality and efficiency of signal processing is achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of signal processing, in particular to a sensor signal noise suppression method and system based on an adjustable filter. BACKGROUND

[0002] In the process of sensor signal acquisition and processing, the signal quality is often affected by multiple factors such as sensor body characteristics, external environmental interference, etc., so that a large amount of random noise and specific frequency interference are mixed in the original signal. In order to improve the quality of the sensor signal, the existing technology usually adopts the steps of parameter adjustment and filter separation to achieve the noise reduction target. However, in the process of parameter adjustment, the steps of parameter adjustment and filter separation need to be iterated multiple times, so the whole processing process is inefficient and it is difficult to meet the demand of real-time signal processing. Moreover, the filter cannot flexibly adjust the suppression strength, and if the parameter setting is improper, the effective signal component is filtered out, causing signal distortion or serious noise residue, thereby reducing the overall signal processing quality.

[0003] In summary, in the prior art, the noise component of the sensor signal is complex and dynamically changes, and the conventional filtering method cannot achieve adaptive regulation and control of different noise states, resulting in the technical problem of low sensor signal noise suppression efficiency and suppression quality. SUMMARY

[0004] The purpose of the present application is to provide a sensor signal noise suppression method and system based on an adjustable filter, which solves the technical problem in the prior art that the noise component of the sensor signal is complex and dynamically changes, and the conventional filtering method cannot achieve adaptive regulation and control of different noise states, resulting in low sensor signal noise suppression efficiency and suppression quality.

[0005] In view of the above problems, the present application provides a sensor signal noise suppression method and system based on an adjustable filter.

[0006] The first aspect of the present application provides a sensor signal noise suppression method based on an adjustable filter, which comprises: predicting the signal entropy increase of a sensor, making an impedance control decision under entropy flow control according to the signal entropy increase value, determining an impedance control instruction and acting on an accessed impedance matching network; in cooperation with the operation of the impedance matching network, performing source-end signal acquisition of the sensor to determine a first sensor signal; performing frequency resonance detection on the first sensor signal by deploying a lightweight resonance verification array, and performing noise suppression control according to the resonance detection result to determine a filtering result; wherein the noise suppression control mode comprises: if the resonance detection result is a non-resonance state, terminating the noise suppression process; if the resonance detection result is a resonance state, activating the filter unit to generate control.

[0007] Optionally, entropy increase rules of the read sensor signal sequence are read to determine a first entropy increase trend value; a second entropy increase value based on environmental interference is analyzed according to an environmental variable; impedance control analysis is performed on the first entropy increase trend value and the second entropy increase value according to an entropy flow controller to output an impedance control instruction; and the impedance matching network accessed is regulated according to the impedance control instruction, wherein the regulation parameters of the impedance matching network at least include resistance, capacitance, and inductance parameters.

[0008] Optionally, the impedance control instruction is the negative entropy flow strength and direction required to offset the signal entropy increase based on the first entropy increase trend value and the second entropy increase value, and the impedance state of the sensor signal source and the environmental internal resistance reach a conjugate matching relationship for the purpose of regulation.

[0009] Optionally, a lightweight second-order IIR filter array is deployed as a resonance verification array; filter records of the sensor are obtained to mine a noise frequency set; and the resonance verification array is tuned according to the noise frequency set, wherein each item in the resonance verification array corresponds to a noise frequency, and the resonance verification array is used to perceive the frequency component change of the corresponding noise.

[0010] Optionally, the first sensor signal is subjected to resonance verification according to the resonance verification array to determine a resonance detection result, wherein a preset tolerance interval is used as a constraint, an amplitude satisfying a zero value is taken as a non-resonance condition, and an abnormally high amplitude frequency is taken as a resonance condition; and the resonance detection result is determined, and if all the resonance detection result is a non-resonance condition, a direct storage instruction is generated to store the first sensor signal.

[0011] Optionally, if there is a resonance condition in the resonance detection result, a matching verification unit in the resonance verification array is located; a frequency component vector of the matching verification unit is integrated to determine to-be-suppressed source data in combination with the first sensor signal, and a noise suppression instruction is generated.

[0012] Optionally, a filter architecture is determined, wherein the filter architecture is composed of an adjustable filter-mechanism discriminator, the adjustable filter receives noise-containing data to generate recovered data, and the mechanism discriminator performs mechanism fingerprint quality inspection based on a pure signal; for the filter architecture, supervised training convergence based on an adversarial network is performed to generate a filter unit and deploy the filter unit in the resonance verification array.

[0013] Optionally, the filter unit is activated with the sending of the noise suppression instruction; the to-be-suppressed source data is input into the adjustable filter of the filter unit to generate a filter result; and the filter result is subjected to mechanism fingerprint determination according to the mechanism discriminator to output a filter result satisfying a pure signal standard.

[0014] Optionally, the resonance detection array comprises a first operation mode and a second operation mode, wherein the first operation mode is a full operation state of the array, and the second operation mode is a low-power monitoring state, and the second operation mode reserves the resonance detection operation of the core frequency; the first operation mode is driven in a high-entropy period and a key sensing period, and the second operation mode is determined in a low-entropy period and a non-key sensing period.

[0015] In a second aspect of the present application, a sensor signal noise suppression system based on an adjustable filter is provided, which comprises: an instruction determination module configured to predict signal entropy increase of a sensor, make impedance control decisions under entropy flow control according to the signal entropy increase value, determine impedance control instructions, and act on an accessed impedance matching network; a signal acquisition module configured to perform source-end signal acquisition of the sensor in cooperation with operation of the impedance matching network, and determine a first sensor signal; a noise suppression control module configured to perform frequency resonance detection on the first sensor signal by deploying a lightweight resonance detection array, perform noise suppression control according to the resonance detection result, and determine a filtering result; wherein the noise suppression control mode in the noise suppression control module comprises: if the resonance detection result is a non-resonance state, terminating the noise suppression process; and if the resonance detection result is a resonance state, activating a filtering unit to generate control.

[0016] The one or more technical solutions provided in the present application have at least the following technical effects or advantages:

[0017] The method provided in the embodiments of the present application predicts signal entropy increase of a sensor, makes impedance control decisions under entropy flow control according to the signal entropy increase value, determines impedance control instructions, and acts on an accessed impedance matching network; cooperates with operation of the impedance matching network to perform source-end signal acquisition of the sensor, and determines a first sensor signal; performs frequency resonance detection on the first sensor signal by deploying a lightweight resonance detection array, performs noise suppression control according to the resonance detection result, and determines a filtering result, thereby achieving efficient and adaptive sensor signal noise suppression and improving the quality and efficiency of signal processing.

[0018] The above description is only a summary of the technical solutions of the present application. In order to more clearly understand the technical means of the present application, the embodiments of the present application can be implemented in accordance with the content of the description, and in order to make the above and other purposes, characteristics and advantages of the present application more obvious and easy to understand, the following specific embodiments of the present application are described. It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0019] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0020] Figure 1 A flowchart illustrating the sensor signal noise suppression method based on an adjustable filter provided in this application.

[0021] Figure 2 This is a schematic diagram of the sensor signal noise suppression system based on an adjustable filter provided in this application.

[0022] Explanation of reference numerals in the attached diagram: Command determination module 11, signal acquisition module 12, noise suppression and control module 13. Detailed Implementation

[0023] This application provides a sensor signal noise suppression method and system based on an adjustable filter. It addresses the technical problem in existing technologies where sensor signal noise components are complex and dynamically changing, making it difficult for conventional filtering methods to adaptively adjust to different noise states, resulting in low noise suppression efficiency and quality. The method achieves efficient and adaptive sensor signal noise suppression, improving the quality and efficiency of signal processing.

[0024] The technical solutions of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be understood that the present invention is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention. It should also be noted that, for ease of description, only the parts related to the present invention are shown in the accompanying drawings, not all of them.

[0025] Example 1, as Figure 1 As shown, this application provides a sensor signal noise suppression method based on an adjustable filter, the sensor signal noise suppression method based on an adjustable filter includes:

[0026] The signal entropy increase of the sensor is predicted, and impedance control decision is made under entropy flow control based on the signal entropy increase. The impedance control command is determined and applied to the connected impedance matching network.

[0027] Further, according to the signal entropy increase value, the impedance control decision under the entropy flow control is determined, the impedance control instruction is determined, and the impedance matching network connected is affected, which comprises: reading the entropy increase law of the upper sensor signal sequence, determining the first entropy increase trend value; according to the environmental variable, analyzing the second entropy increase value based on environmental interference; according to the entropy flow controller, the first entropy increase trend value and the second entropy increase value are analyzed for impedance control, and the impedance control instruction is output; according to the impedance control instruction, the impedance matching network connected is regulated, wherein the regulation parameters of the impedance matching network at least include resistance, capacitance and inductance parameters.

[0028] Specifically, the signal entropy increase of the sensor is predicted to obtain the signal entropy value, which includes a first entropy increase trend value and a second entropy value. The process involves: acquiring a continuous sequence of upper-level sensor signals output by the sensor (the upper-level sensor signal sequence refers to all output data of the sensor within a certain period); normalizing the signal data; then segmenting the signal sequence according to a fixed-length sliding time window to construct a time series sample set; quantizing the signal amplitude within each window to form several discrete amplitude intervals; calculating the probability of occurrence of each interval to obtain the probability distribution of the time window; and based on the probability distribution, using an entropy calculation formula, such as the Shannon entropy formula, calculating the entropy value corresponding to each time window to obtain the signal entropy value. This signal entropy value characterizes the uncertainty and disorder of the sensor signal during that time period. The entropy increments of continuous time windows are arranged chronologically, and the difference sequence of adjacent entropy increments is calculated to obtain the growth rate of entropy increment over time, forming an entropy increment sequence. Then, using methods such as polynomial fitting or Kalman filtering, the trend of the entropy increment sequence is estimated, outputting a first entropy increment trend value. This first entropy increment trend value represents the entropy increase trend of the sensor signal itself under the absence of significant external interference. Environmental sensing units deployed in the same area as the sensor, such as temperature sensors, humidity sensors, and electromagnetic field strength meters, collect environmental information about the current sensor area, obtaining environmental variables such as temperature, humidity, and electromagnetic interference intensity. The collected environmental variables are normalized, and the characteristic probability distributions of the environmental variables, such as fluctuation frequency, amplitude variance, and sudden interference intensity, are statistically analyzed. Based on the Shannon entropy formula, the entropy value of the characteristic probability distribution is calculated. The calculated base entropy value is compared with the baseline entropy value of the sensor signal itself, and the difference is extracted as the entropy increment caused by environmental noise, obtaining a second entropy increment based on environmental interference. This second entropy increment reflects the additional disorder brought to the sensor signal by environmental interference. The first entropy increase trend value and the second entropy increase value are input into the entropy flow controller to perform impedance control analysis. The entropy flow controller determines the specific parameter values ​​of the resistor, capacitor, and inductor that need to be adjusted based on these values ​​to achieve conjugate matching between the signal source impedance and the environmental internal resistance, outputting a precise impedance control command. The entropy flow controller is a control device based on the principle of information entropy flow, used to analyze the impedance adjustment required to achieve a stable and low-noise state for the sensor signal under the current signal condition. According to the impedance control command, the connected impedance matching network is adjusted. This network is used to achieve impedance adjustment and can employ various structures, as long as it allows for fine control of the resistance, capacitance, and inductance parameters. For example, it can use a digital potentiometer to achieve continuous resistance adjustment or a MOSFET array to construct a reactive network; no specific limitations are imposed here.The regulation parameters of the impedance matching network at least include resistance, capacitance and inductance parameters, and the regulation parameters are accurately adjusted by the impedance control instruction, which is used to change the characteristics of the impedance matching network, so that the impedance matching network and the impedance state of the sensor signal source and the environmental internal resistance reach a conjugate matching relationship, thereby effectively reducing the reflection and loss of the sensing signal in the transmission process and improving the transmission efficiency and quality of the sensing signal.

[0029] Further, the impedance control instruction is the negative entropy flow intensity and direction required to offset the signal entropy increase based on the first entropy increase trend value and the second entropy increase value, and the impedance state of the sensor signal source and the environmental internal resistance reach a conjugate matching relationship for the purpose of regulation.

[0030] Specifically, the signal entropy increase quantifies the degree of disorder or uncertainty of the signal, and the entropy increase represents the intensification of noise interference and the degradation of signal quality. The impedance control instruction output by the entropy flow controller is the negative entropy flow intensity and direction required to offset the entropy increase caused by the first entropy increase trend value and the second entropy increase value. The negative entropy flow refers to a reverse adjustment amount applied to offset the entropy increase of the sensor signal caused by changes in its own characteristics and environmental interference. The negative entropy flow intensity represents the size of the reverse adjustment amount required to reduce signal entropy increase and improve signal purity. The intensity size depends on the degree of current signal entropy increase, and the negative entropy flow direction is the direction of reducing signal disorder and making the signal stable and pure. Based on the maximum power transmission principle, when the signal source internal resistance and the load impedance satisfy the conjugate matching relationship, the energy transmission efficiency is the highest, and the external interference introduced is the least. The conjugate matching refers to the condition that the signal source impedance and the load impedance reach the maximum power transmission when the real part is equal and the imaginary part is the opposite number. By regulating the resistance, capacitance and inductance parameters through the impedance control instruction, the required negative entropy flow intensity and direction are generated to realize the conjugate matching of the impedance state of the sensor signal source and the environmental internal resistance. By regulating based on the impedance control instruction, not only the signal transmission efficiency is optimized, but also the introduction of external interference is reduced from the physical layer, so that a relatively pure sensing signal is obtained, and the transmission efficiency and quality of the sensing signal are improved.

[0031] In cooperation with the operation of the impedance matching network, the source end signal of the sensor is collected to determine the first sensor signal.

[0032] A light resonance verification array is deployed to perform frequency resonance detection on the first sensor signal, noise suppression control is performed according to the resonance detection result, and a filtering result is determined. The noise suppression control mode includes: if the resonance detection result is a non-resonance state, the noise suppression process is terminated; and if the resonance detection result is a resonance state, a filtering unit is activated to generate control.

[0033] Specifically, when the impedance control instruction acts on the accessed impedance matching network, the impedance matching network dynamically adjusts the resistance, capacitance and inductance parameters, so that the equivalent output impedance of the sensor signal source and the environmental resistance form a conjugate matching relationship, the source end signal output by the sensor is collected to form a first sensor signal. And input the first sensor signal into the pre-deployed lightweight resonance detection array, the frequency resonance of the first sensor signal is detected, wherein the resonance detection array includes a plurality of filters for detecting specific frequency components in the signal and identifying the resonance frequency in the signal, so as to determine whether there is noise interference: when the signal amplitude is close to zero, it is determined that it is in a non-resonance state; When the signal amplitude is significantly increased and exceeds the preset threshold, it is determined that there is a resonance state, that is, the frequency component may be affected by noise interference or external resonance source. Further, according to the resonance detection result, noise suppression control is performed: if the resonance detection result is a non-resonance state, that is, the signal amplitude is close to zero or within the normal fluctuation range, it indicates that the current modulation can collect interference-free sensing signals, and further processing is not required. At this time, the noise suppression process is terminated. If the resonance detection result is a resonance state, the filter unit is activated, and the signal component corresponding to the frequency is input into the filter unit as the to-be-suppressed source data for dynamic adjustment, noise suppression control is performed, and the filter result is determined. Through frequency-level resonance detection, accurate identification and dynamic suppression of sensor signal noise are realized, and the quality and stability of the signal in the transmission and processing process are improved.

[0034] Further, a lightweight resonance detection array is deployed, including: deploying a lightweight second-order IIR filter array as a resonance detection array; obtaining a filter record of the sensor to mine a noise frequency set; and tuning the resonance detection array according to the noise frequency set, wherein each item in the resonance detection array corresponds to a noise frequency, and the resonance detection array is used to perceive the frequency component change of the corresponding noise.

[0035] Specifically, a resonance detection array is formed by deploying a plurality of lightweight second-order IIR filters, which are used to perceive changes in specific frequency components. The lightweight second-order IIR filter only retains the signal frequency perception function and does not have the full filtering processing capability of a traditional filter. It does not need to adjust the filtering parameters for filtering processing, but responds quickly to the amplitude changes of specific frequencies to achieve real-time detection of noise frequency bands, ensuring high efficiency and low delay of online interference detection. Collecting filtering records of sensors under different environmental conditions, including the interference frequency, time distribution and intensity information marked during the past noise suppression process, clustering analysis algorithm DBSCAN or K-means is used to cluster analyze the data in the filtering records, and the signal energy of different frequency bands is counted to form a noise frequency set. According to the noise frequency set, each second-order IIR filter in the resonance detection array is assigned a target frequency, that is, each item in the resonance detection array corresponds to a noise frequency. By adjusting the normalized center frequency parameter and bandwidth parameter, the resonance response peak value is aligned with the target noise frequency, so as to ensure that the filter is sensitive to the amplitude change of the target frequency component and responds quickly. Specifically, the resonance peak width is adjusted according to the noise intensity, narrow bandwidth is used for strong interference to improve selectivity, and wide bandwidth is used for weak interference to avoid missing detection. The tuning algorithm uses a greedy strategy to preferentially assign high-frequency noise to high-frequency band filters with better anti-aliasing performance, while avoiding assigning adjacent frequencies to the same filter channel through a conflict detection mechanism. For example, a resonance detection array is formed by three second-order IIR filters, the noise frequency set includes low frequency 0-1kHz, medium frequency 1-5kHz and high frequency 5-10kHz, and according to the noise frequency, the parameters of each second-order IIR filter are adjusted to make each filter detect the corresponding noise frequency range, so as to ensure that the filter can accurately detect the change of the corresponding noise frequency component.

[0036] By deploying a lightweight second-order IIR filter array, changes in specific frequency components can be quickly perceived, thereby achieving rapid detection of noise. Tuning the resonance detection array according to the noise frequency set can ensure that each filter accurately detects the corresponding noise frequency, improving the detection accuracy and response speed of the filter and ensuring the high quality and stability of the signal during transmission and processing.

[0037] Further, the first sensor signal is subjected to frequency resonance detection, and noise suppression control is performed according to the resonance detection result, including: performing resonance detection on the first sensor signal according to the resonance detection array to determine the resonance detection result, wherein a predetermined tolerance interval is used as a constraint, an amplitude satisfying zero value is regarded as a non-resonance condition, and an abnormally high amplitude frequency is regarded as a resonance condition; determining the resonance detection result, if all the resonance detection results are non-resonance conditions, a direct storage instruction is generated, and the first sensor signal is stored.

[0038] Specifically, the first sensor signal is input into the resonance detection array, the resonance detection array detects the first sensor signal by frequency band through a plurality of filters, and identifies the resonance frequency component in the first sensor signal, and the output signal amplitude of each filter reflects the resonance in the frequency band. By statistically analyzing the historical signal data collected by the sensor under normal operation, the typical fluctuation range of the sensor signal is determined, and a preset tolerance interval is set, for example, by the standard deviation method, the mean ± 3 times the standard deviation of the sensor signal is taken as the tolerance interval. The preset tolerance interval is a preset allowable error range for determining whether the amplitude of the sensor signal meets the resonance or non-resonance condition, which can be dynamically adjusted according to the real-time characteristics of the sensor signal. By setting the preset tolerance interval, small fluctuations or errors in actual situations can be adapted to ensure that small fluctuations or errors are not misjudged as resonance or non-resonance conditions. By analyzing the output signal amplitude of each filter, it is determined whether there is a resonance state in the signal. When the output signal amplitude meets the zero value, it indicates a non-resonance condition, and when the amplitude frequency abnormally increases, it indicates a resonance condition, which prompts that the signal may be disturbed by noise. The resonance detection result is obtained by resonance detection, the amplitude in the resonance detection result is determined, if all frequencies in the resonance detection result meet the zero value, it indicates that the first sensor signal is in a non-resonance state, and there is no noise interference in the first sensor signal, a direct storage instruction is generated, the first sensor signal is stored, and no further processing is required, thereby saving computing resources and power consumption.

[0039] Further, if there is a resonance condition in the resonance detection result, a matching detection unit in the resonance detection array is located; the frequency component vector of the matching detection unit is integrated, the first sensor signal is combined to determine the to-be-suppressed source data, and a noise suppression instruction is generated.

[0040] Specifically, if the resonance detection result indicates that there is a resonance state in the first sensing signal, that is, the output signal amplitude of one or more filters is abnormally high, the output signal amplitude of each filter is analyzed to determine which filters have abnormally high output signal amplitudes. According to the amplitude analysis result, the matching detection unit in the resonance detection array is located, for example, if the output signal amplitude of the low-frequency filter is abnormally high, the low-frequency filter is located as the matching detection unit. The frequency, amplitude and other component information are extracted from the matching detection unit, and the extracted component information is integrated to form a frequency component vector, which refers to the frequency and amplitude characteristics of the first sensing signal affected by noise. The frequency component vector and the first sensor signal are jointly analyzed, and the vector superposition or weighted fusion method is used to quantify the influence strength of different frequency interference. At the same time, the first sensor signal is mapped in the frequency domain by using fast Fourier transform or discrete wavelet transform, and the signal segment or signal component affected by the specific frequency interference is identified and determined as the to-be-suppressed source data. According to the noise frequency range and strength of the to-be-suppressed source data, a noise suppression instruction is generated to optimize the noise suppression strategy. The noise suppression instruction is used to guide the noise suppression operation of the filter unit, including the cutoff frequency, gain and other parameters of the filter.

[0041] By locating the matching unit in the resonance detection array and integrating the frequency characteristics, the noise source can be accurately identified, the noise suppression can be activated according to the frequency and demand, the blind filtering of the overall signal can be avoided, the high-quality and target-oriented noise suppression instruction can be provided for the adjustable filter, the noise suppression strategy can be optimized, the signal noise can be effectively suppressed, and the signal purity can be improved.

[0042] Further, before activating the filter unit to generate the control, the filter unit is constructed, including: determining a filter architecture, wherein the filter architecture is composed of an adjustable filter and a mechanism discriminator. The adjustable filter receives the noisy data and generates the recovered data, and the mechanism discriminator performs mechanism fingerprint inspection based on the pure signal; for the filter architecture, the supervised training convergence based on the adversarial network is performed to generate the filter unit and deploy it in the resonance detection array.

[0043] Specifically, the construction of the filtering unit first determines the filtering architecture, which is specifically composed of two parts of an adjustable filter and a mechanism discriminator. The adjustable filter is used to receive noisy data and generate recovered data, and the filtering parameters of the adjustable filter are dynamically adjusted according to the noise characteristics of the input signal to achieve accurate suppression of specific frequency noise. For example, the adjustable filter can adopt a band-stop filter or a notch filter to effectively suppress noise in a specific frequency range. The mechanism discriminator is a verification component of the filtering architecture, which performs quality inspection on the filtering output based on the mechanism fingerprint of the pure signal. The mechanism fingerprint is a verification feature formed based on the physical characteristics and generation mechanism of the first sensing signal, including specific frequency components, amplitude distribution, phase relationship, etc., which is used to judge whether the recovered data meets the standard of the pure signal. The training data set is called, which includes noisy data and corresponding pure signal data. Based on the training data set, the filtering architecture is supervised trained by using an adversarial network such as a generative adversarial network (GAN). The adversarial network is composed of a generator adjustable filter and a mechanism discriminator. The generator is used to generate recovered data, and the mechanism discriminator is used to judge whether the recovered data meets the standard of the pure signal. During the training process, the generator continuously adjusts its parameters to generate recovered data closer to the pure signal, and the discriminator continuously adjusts its parameters to more accurately identify the purity of the recovered data. Through continuous iterative training, the generator and the discriminator compete with each other, and finally reach a convergence state to obtain the filtering unit. The trained filtering unit is deployed at the back of the resonance verification array, so that it can be activated and perform noise suppression operation immediately when resonance is detected, improving the response speed and processing efficiency of signal noise suppression, effectively removing interference, and improving the purity of the signal.

[0044] Further, after generating the noise suppression instruction, the method further includes: activating the filtering unit with the sending of the noise suppression instruction; inputting the to-be-suppressed source data into the adjustable filter of the filtering unit to generate a filtering result; and performing mechanism fingerprint determination on the filtering result according to the mechanism discriminator to output a filtering result meeting the pure signal standard.

[0045] Specifically, after generating the noise suppression instruction, the post-deployed filter unit of the resonance verification array is activated, the source data to be suppressed is input into the filter unit as input data, the adjustable filter in the filter unit dynamically adjusts the center frequency, bandwidth and gain parameters according to the previous training results and noise characteristics, performs directional filtering on the disturbed frequency band in the signal, and generates a filtering result. The mechanism discriminator in the filter unit performs mechanism fingerprinting on the filtering result input by the adjustable filter, compares the consistency of the signal data in the filtering result with the frequency components, amplitude distribution, phase relationship, etc. of the pure sensor signal, and judges whether the filtering result meets the pure signal standard. When it is judged that the filtering result meets the pure signal standard, the filtering result meeting the pure signal standard is output. If it does not meet the standard, the filter parameters are adjusted again or further processed. Through the cooperative operation of the filter unit and the mechanism discriminator, on-demand noise suppression and signal fidelity double guarantee are realized, not only the interference components of the resonance detection positioning are effectively removed, but also the output signal meets the physical or statistical mechanism standard, improving the accuracy and stability of the sensor signal in the transmission and processing process.

[0046] Further, the resonance verification array includes a first operating mode and a second operating mode, wherein the first operating mode is an array full operating state, and the second operating mode is a low-power monitoring state, and the second operating mode reserves the resonance verification operation of the core frequency; the first operating mode is driven in a high-entropy period and a key sensing period, and the second operating mode is determined in a low-entropy period and a non-key sensing period.

[0047] Specifically, the operation mode of the resonance detection array includes two types of first operation mode and second operation mode, wherein the first operation mode is a full operation state of the array, that is, all second-order IIR filter units in the resonance detection array are in an active state, which can cover all frequency bands in the noise frequency set and ensure comprehensive monitoring of signals during periods of high interference or signal sensitivity. The second operation mode is a low-power monitoring state, which only retains the resonance detection operation of the core frequency, reducing the calculation and energy consumption overhead. The operation mode of the resonance detection array is determined by monitoring the complexity and interference level of the sensor signal through the entropy value. By continuously monitoring the entropy increment value of the signal, the uncertainty of the signal is evaluated in real time. When the entropy increment value is greater than or equal to the preset entropy increment threshold, the sensor signal is disordered, noise is frequent, and the signal stability is poor, which is determined as a high-entropy period. When the entropy increment value is less than the preset entropy increment threshold, it indicates that the signal is stable and the noise interference is low, which is determined as a low-entropy period. When the first sensor signal is detected as a high-entropy period and a key sensing period, the first operation mode is automatically driven to ensure maximum frequency detection coverage. The key sensing period refers to the sensing data acquisition period of safety monitoring and key control nodes. When the first sensor signal is detected as a low-entropy period and a non-key sensing period, the second operation mode is automatically switched to, and only the resonance detection operation of the core frequency band is retained, thereby effectively reducing power consumption and unnecessary data processing. The low-entropy period refers to a period of stable signal and low noise interference.

[0048] By introducing the dual-mode operation mechanism, the dynamic balance between sensor signal detection accuracy and energy efficiency is achieved, which can ensure full-band noise monitoring during high-risk periods and reduce energy consumption during low-risk periods, thereby improving the flexibility of the resonance detection array under different working conditions and achieving the best performance and energy efficiency balance.

[0049] In summary, the sensor signal noise suppression method based on the adjustable filter provided in the present application has the following beneficial effects:

[0050] By using an impedance matching network, the energy transfer efficiency of the sensor signal is maximized under a conjugate matching relationship, effectively reducing signal reflection and suppressing the introduction of external interference, ensuring that the sensor signal is in a relatively pure state. Based on this, a resonance detection mechanism is introduced to quickly determine whether interference exists in the acquired signal. When the detection result shows no resonance interference, it indicates that the front-end modulation has brought the signal to a low-interference or interference-relaxed state, and it can be directly stored and used. When the detection result shows resonance interference, combined with the located interference frequency and the acquired source signal, a pre-trained filter unit directly outputs a pure signal based on the training results, eliminating the need for parameter tuning and separation steps under conventional filtering. This avoids the high complexity and delay problems of conventional filtering methods, achieving fast, accurate, and adaptive noise suppression. This technical effect improves the efficiency and quality of sensor signal processing while ensuring signal authenticity.

[0051] Example 2 is based on the same inventive concept as the sensor signal noise suppression method based on tunable filters in the previous examples, such as... Figure 2 As shown, this application provides a sensor signal noise suppression system based on an adjustable filter, wherein the sensor signal noise suppression system based on an adjustable filter includes:

[0052] The instruction determination module 11 is used to predict the signal entropy increase of the sensor, make impedance control decisions under entropy flow control based on the signal entropy increase, determine the impedance control instruction, and apply it to the connected impedance matching network; the signal acquisition module 12 is used to coordinate the operation of the impedance matching network, perform source-end signal acquisition of the sensor, and determine the first sensor signal; the noise suppression and control module 13 is used to perform frequency resonance detection on the first sensor signal by deploying a lightweight resonance detection array, perform noise suppression and control based on the resonance detection result, and determine the filtering result; wherein, the noise suppression and control method in the noise suppression and control module 13 includes: if the resonance detection result is a non-resonance state, terminating the noise suppression process; if the resonance detection result is a resonance state, activating the filtering unit for generation control.

[0053] Furthermore, the instruction determination module 11 in the sensor signal noise suppression system based on the adjustable filter is also used to: read the entropy increase law of the upper-level sensor signal sequence and determine the first entropy increase trend value; analyze the second entropy increase value based on environmental interference according to environmental variables; perform impedance control analysis on the first entropy increase trend value and the second entropy increase value according to the entropy flow controller, and output an impedance control instruction; and regulate the connected impedance matching network according to the impedance control instruction, wherein the regulation parameters of the impedance matching network include at least resistance, capacitance, and inductance parameters.

[0054] Further, the instruction determination module 11 in the adjustable filter-based sensor signal noise suppression system is further configured to determine the impedance control instruction, which is the negative entropy flow intensity and direction required to offset the signal entropy increase based on the first entropy increase trend value and the second entropy increase value, so as to achieve a conjugate matching relationship between the impedance state of the sensor signal source and the environmental resistance for the purpose of impedance regulation.

[0055] Further, the noise suppression control module 13 in the adjustable filter-based sensor signal noise suppression system is further configured to deploy a lightweight second-order IIR filter array as a resonance detection array, obtain a filter record of the sensor, mine a noise frequency set, and tune the resonance detection array according to the noise frequency set, wherein each item in the resonance detection array corresponds to a noise frequency, and the resonance detection array is used to perceive the frequency component change of the corresponding noise.

[0056] Further, the noise suppression control module 13 in the adjustable filter-based sensor signal noise suppression system is further configured to perform resonance detection on the first sensor signal according to the resonance detection array to determine a resonance detection result, wherein a preset tolerance interval is used as a constraint, an amplitude satisfying a zero value is taken as a non-resonance condition, and an abnormally high amplitude frequency is taken as a resonance condition, and the resonance detection result is determined.

[0057] Further, the noise suppression control module 13 in the adjustable filter-based sensor signal noise suppression system is further configured to, if there is a resonance condition in the resonance detection result, locate a matching detection unit in the resonance detection array, integrate a frequency component vector of the matching detection unit, combine the first sensor signal, determine to-be-suppressed source data, and generate a noise suppression instruction.

[0058] Further, the noise suppression control module 13 in the adjustable filter-based sensor signal noise suppression system is further configured to determine a filter architecture, wherein the filter architecture is composed of an adjustable filter and a mechanism discriminator, the adjustable filter receives noisy data and generates recovered data, and the mechanism discriminator performs mechanism fingerprint quality inspection based on a pure signal, and perform supervised training convergence based on an adversarial network for the filter architecture, generate a filter unit, and deploy the filter unit in the resonance detection array.

[0059] Further, the noise suppression control module 13 in the adjustable filter-based sensor signal noise suppression system is further configured to activate the filter unit with the sending of the noise suppression instruction, input the to-be-suppressed source data into the adjustable filter of the filter unit to generate a filter result, and perform mechanism fingerprint determination on the filter result according to the mechanism discriminator to output a filter result satisfying a pure signal standard.

[0060] Further, the noise suppression management module 13 in the adjustable filter-based sensor signal noise suppression system is further configured to: the resonance detection array includes a first operating mode and a second operating mode, wherein the first operating mode is a full operating state of the array, and the second operating mode is a low-power monitoring state, and the second operating mode reserves resonance detection operation of the core frequency; the first operating mode is driven in a high-entropy period and a key sensing period, and the second operating mode is determined in a low-entropy period and a non-key sensing period.

[0061] The various embodiments in the specification are described in a progressive manner, and each embodiment focuses on the difference from other embodiments. The foregoing Figure 1 The adjustable filter-based sensor signal noise suppression method and specific examples in Embodiment One are also applicable to the adjustable filter-based sensor signal noise suppression system of the present embodiment. Through the foregoing detailed description of the adjustable filter-based sensor signal noise suppression method, those skilled in the art can clearly understand the adjustable filter-based sensor signal noise suppression system in the present embodiment. Therefore, in the interest of brevity, the adjustable filter-based sensor signal noise suppression system in the present embodiment will not be described in detail.

[0062] The above description of the disclosed embodiments enables a person skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

[0063] Obviously, for those skilled in the art, without departing from the principles of the present application, the present application can be improved and modified in several ways, and these improvements and modifications also fall within the protection scope of the present application.

Claims

1. A sensor signal noise suppression method based on an adjustable filter, characterized in that, The method includes: The signal entropy increase of the sensor is predicted, and impedance control decision is made under entropy flow control based on the signal entropy increase. The impedance control command is determined and applied to the connected impedance matching network. In coordination with the operation of the impedance matching network, the source-end signal of the sensor is acquired to determine the first sensor signal; By deploying a lightweight resonance detection array, frequency resonance detection is performed on the first sensor signal, and noise suppression and control are performed based on the resonance detection results to determine the filtering result; The noise suppression and control methods include: If the resonance detection result is a non-resonance state, the noise suppression process is terminated. If the resonance detection result indicates the presence of a resonance state, the filtering unit is activated for generation control. Deploying a lightweight resonance detection array includes: Deploy a lightweight second-order IIR filter array as a resonance detection array; Obtain the filtered records from the sensors and mine the noise frequency set; The resonance detection array is tuned according to the noise frequency set, wherein each item in the resonance detection array corresponds to a noise frequency, and the resonance detection array is used to sense the frequency component changes of the corresponding noise. Frequency resonance detection is performed on the first sensor signal, and noise suppression and control are performed based on the resonance detection results, including: According to the resonance detection array, the first sensor signal is resonantly detected to determine the resonance detection result. The preset tolerance range is used as a constraint. The amplitude is zero as a non-resonance condition and the amplitude frequency is abnormally increased as a resonance condition. The resonance detection results are determined. If all the resonance detection results are non-resonance cases, a direct storage instruction is generated to store the first sensor signal. If resonance is detected in the resonance detection results, locate the matching verification unit within the resonance verification array; By integrating the frequency component vectors of the matching verification unit and combining them with the first sensor signal, the source data to be suppressed is determined, and a noise suppression command is generated. Before activating the filtering unit for generation control, the construction of the filtering unit includes: A filtering architecture is determined, wherein the filtering architecture consists of an adjustable filter and a mechanism discriminator. The adjustable filter receives noisy data and generates recovered data, and the mechanism discriminator performs mechanism fingerprint quality inspection based on the clean signal. For the aforementioned filtering architecture, supervised training convergence based on adversarial networks is performed to generate filtering units, which are then deployed in the resonance calibrator array. After generating the noise suppression command, it includes: Upon receiving the noise suppression command, the filtering unit is activated; The source data to be suppressed is input into the adjustable filter of the filtering unit to generate a filtering result; Based on the mechanism discriminator, the filtering result is subjected to mechanism fingerprinting, and a filtering result that meets the clean signal standard is output.

2. The sensor signal noise suppression method based on an adjustable filter as described in claim 1, characterized in that, Impedance control decisions are made based on the increase in signal entropy, under entropy flow control. Impedance control commands are then determined and applied to the connected impedance matching network, including: Read the entropy increase pattern of the upper-level sensor signal sequence and determine the first entropy increase trend value; Based on environmental variables, analyze the second entropy increment based on environmental disturbances; Based on the entropy flow controller, impedance control analysis is performed on the first entropy increase trend value and the second entropy increase value, and an impedance control command is output. According to the impedance control command, the connected impedance matching network is adjusted, wherein the adjustment parameters of the impedance matching network include at least resistance, capacitance and inductance parameters.

3. The sensor signal noise suppression method based on an adjustable filter as described in claim 2, characterized in that, The impedance control command is to counteract the negative entropy flow intensity and direction required to offset the signal entropy increase based on the first entropy increase trend value and the second entropy increase value, with the aim of achieving a conjugate matching relationship between the impedance state of the sensor signal source and the internal resistance of the environment.

4. The sensor signal noise suppression method based on an adjustable filter as described in claim 1, characterized in that, The resonance detection array includes a first operating mode and a second operating mode. The first operating mode is the full operating state of the array, and the second operating mode is the low-power monitoring state. The second operating mode retains the resonance detection operation of the core frequency. The first operating mode is driven by high-entropy periods and critical sensing periods, and the second operating mode is determined by low-entropy periods and non-critical sensing periods.

5. A sensor signal noise suppression system based on an adjustable filter, characterized in that, The steps for implementing the sensor signal noise suppression method based on an adjustable filter according to any one of claims 1 to 4 include: The instruction determination module is used to predict the signal entropy increase of the sensor, make impedance control decisions under entropy flow control based on the signal entropy increase, determine the impedance control instruction and apply it to the connected impedance matching network. The signal acquisition module is used to coordinate the operation of the impedance matching network, perform source-end signal acquisition of the sensor, and determine the first sensor signal; The noise suppression and control module is used to perform frequency resonance detection on the first sensor signal by deploying a lightweight resonance detection array, perform noise suppression and control based on the resonance detection results, and determine the filtering result. The noise suppression and control methods in the noise suppression and control module include: If the resonance detection result is a non-resonance state, the noise suppression process is terminated. If the resonance detection result indicates the presence of a resonance state, the filtering unit is activated for generation control.

Images