A data processing method and system for an accelerometer driving and control system based on optoelectronic signals

Abstract

The present invention discloses a data processing method and system for an accelerometer drive and control system based on optoelectronic signals, which relates to the technical field of accelerometers. It includes obtaining signals to be processed, identifying and removing data, enhancing the noise suppression ability, purifying useful signals, collecting iterative characteristics to stabilize noise, removing the noise with stable characteristics, algorithm selection, judging the calculation time of the time-domain convolution algorithm and the moving average filtering algorithm when processing the same data according to the size of the data volume, and selecting the algorithm with shorter time, simplifying the algorithm iteration, simplifying the stop condition of the algorithm iteration process, so that the algorithm ends the loop on the premise of meeting the accuracy requirements and the number of iterations. The present invention can automatically adjust the filtering parameters according to the current environmental conditions to achieve the best noise suppression effect, reduce the calculation burden, improve the real-time response ability. At the same time, according to the size of the data volume, when processing the same data, select the algorithm with shorter time, and reduce the time required for data processing.

Description

Technical Field

[0001] The present invention relates to the technical field of accelerometers, and particularly to a data processing method and system for an accelerometer drive and control system based on optoelectronic signals. Background Art

[0002] With the increasing demand for precise motion detection in fields such as modern industry, aerospace, automotive safety, and consumer electronics, traditional accelerometers (such as piezoelectric, capacitive, etc.) gradually show limitations in terms of sensitivity, stability, and durability. To solve these problems, an accelerometer drive and control system based on optoelectronic signals has emerged. This new type of accelerometer uses optical principles to measure the acceleration changes of objects, providing a new approach different from traditional mechanical or electronic sensors.

[0003] An accelerometer control method, an accelerometer control device, and an intelligent terminal with the application publication number CN107389980A. The accelerometer control method includes: in the operating state of the accelerometer, obtaining the current vibration situation and reporting frequency of the intelligent terminal, where the reporting frequency is the frequency at which the drive layer of the intelligent terminal reports the output data of the accelerometer to the application layer of the intelligent terminal, and the vibration situation indicates whether the intelligent terminal is in a vibrating state; based on the current vibration situation and reporting frequency, determining the target filtering control parameter of the accelerometer based on preset filtering parameter adjustment information, where the filtering parameter adjustment information includes the corresponding relationship information among the vibration situation, reporting frequency, and filtering control parameter; based on the target filtering control parameter, adjusting the current filtering control parameter of the accelerometer, realizing the dynamic adjustment of the filtering control parameter of the accelerometer.

[0004] The algorithm processing real-time response ability of the existing optoelectronic signal accelerometer drive and control system is poor. Complex signals will cause data processing delays, affecting the real-time response ability of the system. In application scenarios that require fast feedback (such as high-speed motion detection), such delays may lead to a decline in the performance of the control system. Moreover, the existing optoelectronic signal accelerometer drive and control system has poor anti-noise interference processing. Although optoelectronic technology has good resistance to electromagnetic interference, in actual applications, environmental noise (such as mechanical vibration, etc.) will still affect the measurement accuracy, and the existing control software lacks an effective mechanism to dynamically adjust to adapt to these changes. Summary of the Invention

[0005] The purpose of the present invention is to provide a data processing method and system for an accelerometer drive and control system based on optoelectronic signals to solve the problems raised in the above background art.

[0006] To achieve the above object, the present invention provides the following technical solution: A data processing method for an accelerometer drive and control system based on optoelectronic signals, the method comprising:

[0007] Obtain the signal to be processed, and obtain the original signal collected by the optoelectronic sensor;

[0008] Identify and remove data, perform identification processing on the input original signal, and remove the irrelevant data therein;

[0009] Enhance the noise suppression ability, estimate the state of the system through the data containing noise, and adjust the filtering parameters according to the current environmental conditions to achieve the best noise suppression effect;

[0010] Purify the useful signal, use the reference signal to estimate and remove the noise component in the main signal to obtain a pure useful signal;

[0011] Collect and iterate on characteristics to stabilize noise, continuously detect the relatively stable noise in characteristics, and after reaching a certain frequency, record it in a specific noise database;

[0012] Remove the noise with stable characteristics, detect the relatively stable noise in the original signal through an algorithm, compare it with the specific noise database, and remove the noise in the specific noise database;

[0013] Algorithm selection, based on the amount of data, judge the calculation time of the time-domain convolution algorithm and the moving average filtering algorithm when processing the same data, and select the algorithm with shorter time. The specific steps are as follows:

[0014] S1: Set a threshold. When the amount of data is lower than the threshold, select the moving average filtering algorithm;

[0015] S2: Set a threshold. When the amount of data is higher than the threshold, select the time-domain convolution algorithm and convert the time-domain signal to the frequency-domain signal;

[0016] S3: Process the frequency-domain signal using the time-domain convolution algorithm. The algorithm formula is as follows:

[0017] Wherein, Y(k) is the frequency-domain representation of the output signal, X(k) is the frequency-domain representation of the input signal, and H(k) is the frequency-domain representation of the filter;

[0018] S4: After the calculation is completed, convert the frequency-domain signal back to the time-domain signal;

[0019] Simplify the algorithm iteration, simplify the stop condition of the algorithm iteration process, so that the algorithm ends the loop on the premise of meeting the accuracy requirements and the number of iterations.

[0020] The identified removal data performs signal preprocessing on the input original signal, removes the irrelevant parts through detrending processing, trims the irrelevant parts of the signal, and reduces the computational amount.

[0021] The enhanced noise suppression ability sets the initial state prediction value and its covariance matrix, predicts the current state based on the state prediction value of the previous moment, corrects the prediction value through signal data, adjusts the filter parameters, and dynamically suppresses the noise.

[0022] The steps for obtaining a pure useful signal are as follows:

[0023] L1: Collect the main signal, which consists of a useful signal and a noise signal;

[0024] L2: Set the reference signal, and the calculation formula for the reference signal is as follows:

[0025] where r is the i-th reference signal, h(t) is the filter for adjusting the i-th reference signal, y(t) is the adjusted i-th reference signal, and Fori indexes the value range of the variable i;

[0026] L3: Perform weighted summation on all the adjusted reference signals to obtain an estimate of the noise signal, and the formula is as follows:

[0027] where w is the weighting coefficient of the i-th adjusted reference signal, and n(t) is the estimated value of the noise signal;

[0028] L4: Extract the useful signal. Subtract the estimated noise signal from the main signal to obtain an estimated value of the pure useful signal, and the calculation formula is as follows:

[0029] where S(t) is the estimated value of the pure useful signal, x(t) is the main signal, and n(t) is the estimated value of the noise signal.

[0030] The collected iterative characteristics stabilize the noise. Continuously detect the noise with relatively stable characteristics. When the same noise with relatively stable characteristics reaches the set number of times, record it in a specific noise database for spectral subtraction data model calculation to reduce the computational burden.

[0031] The algorithm for removing the noise with relatively stable characteristics from the main signal is as follows:

[0032] where S(f) represents the estimated value of the spectrum of the pure useful signal, X(f) represents the spectrum of the main signal, α represents the adjustment factor, and N(f) represents the estimated value of the noise spectrum.

[0033] After the data volume is lower than the threshold, the moving average filtering algorithm is selected, and the algorithm formula is as follows:

[0034] Where: y(n) is the output signal, that is, the signal after moving average filtering processing, M1 represents the average coefficient of the signal samples within the moving window, ∑ M-1 K=0 represents the summation operation, x represents the input signal, and (n-k) represents the time index or sample index of the input signal.

[0035] After the data volume is higher than the threshold, the time-domain convolution algorithm is selected, and the time-domain signal is converted into a frequency-domain signal. The conversion formula is as follows:

[0036] Where: X(k) is the frequency-domain representation of the input signal x(n), x(n) is the original input signal in the time domain, and e is the complex exponential function.

[0037] The frequency-domain signal is reversely converted into a time-domain signal. The conversion formula is as follows:

[0038] Where, x(n) is the finally obtained time-domain signal, X(k) is the processed frequency-domain signal, e is the conjugate form of the complex exponential function, and 1 / N is the normalization factor.

[0039] Compared with the prior art, the beneficial effects of the present invention are:

[0040] The data processing method of the accelerometer drive and control system based on optoelectronic signals enhances the noise suppression ability, automatically adjusts the filtering parameters according to the current environmental conditions to achieve the best noise suppression effect, continuously detects relatively stable noise, and after reaching a certain frequency, enters it into a specific noise database. The algorithm detects relatively stable noise in the original signal, compares it with the specific noise database, removes the noise in the specific noise database, reduces the calculation burden, improves the real-time response ability, and avoids data processing delay caused by complex signals;

[0041] Based on the size of the data volume, when processing the same data, the calculation time of the time-domain convolution algorithm and the moving average filtering algorithm is judged, and the algorithm with shorter time is selected. After the data volume is lower than the threshold, the moving average filtering algorithm is selected. After the data volume is higher than the threshold, the time-domain convolution algorithm is selected to convert the time-domain signal into a frequency-domain signal. The time-domain convolution algorithm is used to process the frequency-domain signal. After the calculation is completed, the frequency-domain signal is reversely converted into a time-domain signal to reduce the time required for data processing. At the same time, the stop condition of the algorithm iteration process is simplified, so that the algorithm ends the loop on the premise of meeting the accuracy requirements and the number of iterations, further reducing the time required for data processing. Brief Description of the Drawings

[0042] Figure 1 is the schematic diagram of the principle structure of the present invention;

[0043] Figure 2 is the schematic diagram of the principle structure of the data removal principle of the present invention;

[0044] Figure 3 is the schematic diagram of the principle structure of the purification signal principle of the present invention;

[0045] Figure 4 is the schematic diagram of the principle structure for removing relatively stable noise characteristics in the present invention;

[0046] Figure 5 is the schematic diagram of the principle structure of the noise suppression principle of the present invention;

[0047] Figure 6 is the schematic diagram of the principle structure of the noise processing principle of the present invention;

[0048] Figure 7 is the schematic diagram of the principle structure of the simplified algorithm principle of the present invention. Detailed implementation manners

[0049] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0050] In this application, for the convenience of understanding, the method steps used do not necessarily need to be executed in the order of the steps in this embodiment during actual operation. In some other embodiments, these steps can be performed synchronously or the order can be changed.

[0051] Embodiment 1:

[0052] As Figures 1-7 shown, the present invention provides a technical solution: a data processing method for an accelerometer drive and control system based on optoelectronic signals, and the method includes:

[0053] Obtain the signal to be processed, and obtain the original signal collected by the optoelectronic sensor;

[0054] Identify and remove data, perform identification processing on the input original signal, and remove the irrelevant data therein;

[0055] Enhance the noise suppression ability, estimate the state of the system through the data containing noise, and adjust the filtering parameters according to the current environmental conditions to achieve the best noise suppression effect;

[0056] A pure and useful signal, using a reference signal to estimate and remove the noise component in the main signal to obtain a pure and useful signal;

[0057] Collect iteratively characteristic-stable noise, continuously detect noise with relatively stable characteristics, and after reaching a certain frequency, enter it into a specific noise database;

[0058] Remove the noise with stable characteristics. Detect the noise with relatively stable characteristics in the original signal through an algorithm, compare it with the specific noise database, and remove the noise in the specific noise database;

[0059] Algorithm selection: Determine the calculation time of the time-domain convolution algorithm and the moving average filtering algorithm when processing the same data based on the amount of data, and select the algorithm with shorter time. The specific steps are as follows:

[0060] S1: Set a threshold. When the amount of data is lower than the threshold, select the moving average filtering algorithm;

[0061] S2: Set a threshold. When the amount of data is higher than the threshold, select the time-domain convolution algorithm and convert the time-domain signal to the frequency-domain signal;

[0062] S3: Process the frequency-domain signal using the time-domain convolution algorithm. The algorithm formula is as follows:

[0063] Among them, Y(k) is the frequency-domain representation of the output signal, X(k) is the frequency-domain representation of the input signal, and H(k) is the frequency-domain representation of the filter;

[0064] Substitute data:

[0065] N = 4;

[0066] X(0) = 1 (X(0) can be any real number. Assume it is 1 as an example);

[0067] X(1) = -2 + 2j (obtained from X(3) = X(1));

[0068] H(k) = 1;

[0069] Now, we can calculate Y(k):

[0070] Y(0) = X(0) × H(0);

[0071] = 1 × 1;

[0072] = 1;

[0073] Y(1) = X(1) × H(1);

[0074] = (-2 + 2j) × 1;

[0075] = -2 + 2j;

[0076] Y(2) = X(2) × H(2);

[0077] = (-6j) × 1;

[0078] = -6j;

[0079] Y(3) = X(3) × H(3);

[0080] = (-2 - 2j) × 1;

[0081] = -2 - 2j;

[0082] Therefore, the frequency - domain representation of the output signal is obtained.

[0083] S4: After the calculation is completed, convert the frequency - domain signal back into the time - domain signal;

[0084] Simplify the algorithm iteration and the stop condition of the algorithm iteration process, so that the algorithm ends the loop on the premise of meeting the accuracy requirements and the number of iterations.

[0085] The recognition and removal of data pre - processes the input original signal, removes the non - relevant part through detrending processing, trims the irrelevant part of the signal, and reduces the computational amount.

[0086] It should be noted that the pre - processing of the input original signal to identify and remove non - relevant data is achieved by eliminating the linear or non - linear trend components in the signal through detrending processing, and then subtracting these trend components from the original signal. Analyze the signal to determine and trim the parts that are obviously irrelevant to the target analysis, such as silent segments, background noise, or invalid data within a specific time period. This step helps to reduce the amount of data and complexity of subsequent calculations. In addition, applying a window function at both ends of the signal makes the signal edge transition more smoothly to zero, thus effectively reducing spectral leakage, significantly reducing the unnecessary computational burden, and also improving the quality of subsequent signal processing and analysis, ensuring that the final result is more accurate and reliable, and can greatly improve work efficiency and analysis accuracy.

[0087] The enhancement of noise suppression ability sets the initial state estimate value and its covariance matrix, predicts the state at the current moment based on the state estimate value at the previous moment, corrects the predicted value through signal data, adjusts the filter parameters, and dynamically suppresses the noise.

[0088] It should be noted that based on the predicted value of the state at the previous moment, the state at the current moment is predicted through the system model, and the predicted value is corrected using the actual measurement data to obtain a more accurate state estimate. The specific steps are as follows: Set the initial state and its covariance matrix, and then at each time step, use the state transition model to predict the current state and its covariance. Then, calculate the gain based on the actual measurement values and update the state estimate and covariance. In addition, evaluate the noise level in the current environment according to the real-time collected data, and dynamically adjust the process noise covariance and measurement noise covariance, so as to optimize the filtering effect, achieve efficient suppression of noise, reduce noise interference, and improve the signal quality.

[0089] The steps for the pure useful signal are as follows:

[0090] L1: Collect the main signal, which consists of the useful signal and the noise signal;

[0091] L2: Set the reference signal, and the calculation formula for the reference signal is as follows:

[0092]

[0093] Substitute the data for calculation, and there are the following two simple discrete-time reference signals and filters:

[0094] r1(n) = {1, 0, 0, 0} (a unit impulse signal);

[0095] r2(n) = {1, 1, 1, 1} (a constant signal);

[0096] h(n) = {21, 21} (a simple average filter);

[0097] For r1(n) and h(n), we can calculate:

[0098] y1(0) = 1 × 21 = 21;

[0099] y1(1) = 0 × 21 + 1 × 21 = 21;

[0100] y1(2) = 0 for n > 1;

[0101] Where r is the i-th reference signal, h(t) is the filter used to adjust the i-th reference signal, y(t) is the adjusted i-th reference signal, and Fori is the value range of the indexing variable i;

[0102] L3: Perform weighted summation on all the adjusted reference signals to obtain an estimate of the noise signal, and the formula is as follows:

[0103]

[0104] Substitute the data,

[0105] w1 = 1,

[0106] y1(0) = 21;

[0107] y1(1) = 21;

[0108] n(0) = w1y1(0);

[0109] = 1×2;

[0110] = 2;

[0111] n(1) = w1y1(1);

[0112] = 1×2;

[0113] = 2;

[0114] where w is the weighting coefficient of the i-th adjusted reference signal, and n(t) is the estimated value of the noise signal;

[0115] L4: Extraction of useful signal. Subtract the estimated noise signal from the main signal to obtain the estimated value of the pure useful signal. The calculation formula is as follows:

[0116]

[0117] where S(t) is the estimated value of the pure useful signal, x(t) is the main signal, and n(t) is the estimated value of the noise signal;

[0118] Substitute the data,

[0119] n(0) = 2;

[0120] n(1) = 2;

[0121] x(0) = 5;

[0122] x(1) = 7;

[0123] Now, we can calculate s(t):

[0124] s(0) = x(0) - n(0);

[0125] = 5 - 2;

[0126] = 3;

[0127] s(1) = x(1) - n(1);

[0128] = 7 - 2;

[0129] = 5;

[0130] Therefore, at time points t = 0 and t = 1, the estimated values of the pure useful signal are s(0) = 3 and s(1) = 5 respectively.

[0131] The collecting iteration characteristic stabilizes the noise, continuously detects the relatively stable noise. When the same relatively stable noise reaches the set number of times, it is recorded in the specific noise database for the calculation of the spectral subtraction data model, reducing the computational burden.

[0132] The relatively stable noise in the main signal is removed through an algorithm. The algorithm formula is as follows:

[0133] Among them, S(f) represents the estimated value of the spectrum of the pure useful signal, which is the processed result, X(f) represents the spectrum of the main signal, α represents the adjustment factor, which is a scalar value used to control the intensity of spectral subtraction, and N(f) represents the estimated value of the noise spectrum;

[0134] Substitute the following data:

[0135] The value of the spectrum X(f) of the main signal at a certain frequency point f0 is X(f0) = 10;

[0136] The estimated value N(f) of the noise spectrum at the same frequency point f0 is N(f0) = 2;

[0137] The adjustment factor α = 0.5.

[0138] Substitute these values into the formula, and we get:

[0139] S(f0) = max{10 - 0.5×2};

[0140] S(f0) = max{10 - 1};

[0141] S(f0) = max{9};

[0142] S(f0) = 9;

[0143] Therefore, at the frequency point f0, the estimated value S(f0) of the spectrum of the pure useful signal is 9.

[0144] When the data volume is lower than the threshold, the moving average filtering algorithm is selected. The algorithm formula is as follows:

[0145] Among them: y(n) is the output signal, that is, the signal after moving average filtering processing, 1 / M represents the average coefficient of the signal samples within the moving window, ∑ M-1 K=0 represents the summation operation, x represents the input signal, and (n - k) represents the time index or sample index of the input signal;

[0146] Suppose we have the following sample values of the input signal x(n):

[0147] x(0) = 1;

[0148] x(1) = 2;

[0149] x(2) = 3;

[0150] x(3) = 4;

[0151] x(4) = 5;

[0152] And we choose a sliding window size M = 3;

[0153] However, for the sake of clarity of the example, we start the calculation from n = 2 and calculate until n = 4;

[0154] When n = 2:

[0155] y(2) = 1 / 3[x(2) + x(1) + x(0)];

[0156] y(2) = 1 / 3[3 + 2 + 1];

[0157] y(2) = 1 / 3 × 6;

[0158] y(2) = 2;

[0159] When n = 3:

[0160] y(3) = 1 / 3[x(3) + x(2) + x(1)];

[0161] y(3) = 1 / 3[4 + 3 + 2];

[0162] y(3) = 1 / 3 × 9;

[0163] y(3) = 3;

[0164] When n = 4:

[0165] y(4) = 1 / 3[x(4) + x(3) + x(2)];

[0166] y(4) = 1 / 3[5 + 4 + 3];

[0167] y(4) = 1 / 3 × 12;

[0168] y(4) = 4;

[0169] Thus, the output signal is obtained.

[0170] When the data volume is higher than the threshold, the time-domain convolution algorithm is selected to convert the time-domain signal to the frequency-domain signal. The conversion formula is as follows:

[0171] Where: X(k) is the frequency-domain representation of the input signal x(n), x(n) is the original input signal in the time domain, and e is the complex exponential function;

[0172] Substitute the sample values of the time-domain signal x(n):

[0173] x(0) = 1;

[0174] x(1) = 2;

[0175] x(2) = 3;

[0176] x(3) = 4;

[0177] The number of samples N = 4;

[0178] Calculate the sample values of the frequency-domain signal X(k);

[0179] When k = 0:

[0180] X(0) = x(0) + x(1) + x(2) + x(3);

[0181] X(0) = 1 + 2 + 3 + 4;

[0182] X(0) = 10;

[0183] (Note: When k = 0, the complex exponential term is 1, so the summation is actually the sum of all sample values of the time-domain signal.)

[0184] When k = 1:

[0185] X(1) = 1×(1) + 2×(-j) + 3×(-1) + 4×(j);

[0186] X(1) = 1 - 2j - 3 + 4j;

[0187] X(1) = -2 + 2j;

[0188] (Note: Here we use the properties of the complex exponential function, specifically e / 2 = -j and e / 2 = j.)

[0189] Similarly, calculate the values of X(k) for k = 2 and k = 3. Due to the symmetry of the DFT (for real input signals), X(N - k) is the conjugate complex number of X(k) (i.e., the real parts are the same and the imaginary parts have opposite signs). Therefore, if X(1) has been calculated, then X(3) is the conjugate complex number of X(1). Similarly, X(2) will be a pure imaginary number (for real input signals and even-length DFTs), and the absolute value of its imaginary part is related to X(0) and X(N / 2).

[0190] Give the calculation results of X(2) and X(3) (using the same properties of the complex exponential function):

[0191] When k = 2:

[0192] X(2) = -6j;

[0193] When k = 3:

[0194] X(3) = X(1)* = -2 - 2j;

[0195] Here, X(1)* represents the conjugate complex number of X(1).

[0196] The conversion of the frequency-domain signal into the time-domain signal is as follows:

[0197] where x(n) is the finally obtained time-domain signal, X(k) is the processed frequency-domain signal, e is the conjugate form of the complex exponential function, and 1 / N is the normalization factor;

[0198] Substitute the data,

[0199] Y(0) = 1;

[0200] Y(1) = -2 + 2j;

[0201] Y(2) = -6j;

[0202] Y(3) = -2 - 2j;

[0203] N = 4 (because there are four frequency-domain samples);

[0204] Now, substitute these data into the IDFT formula to calculate the time-domain signal x(n):

[0205] For n = 0:

[0206] x(0) = 1 / 4[Y(0) + Y(1) + Y(2) + Y(3)];

[0207] x(0) = 1 / 4[1 + (-2 + 2j) + (-6j) + (-2 - 2j)];

[0208] x(0) = 1 / 4[-3 - 6j];

[0209] x(0) = -3 / 4 - 3 / 2j;

[0210] For n = 1:

[0211] x(1) = 1 / 4[Y(0)×e + Y(1)×ej + Y(2)×e + Y(3)×e];

[0212] x(1)=1 / 4[1+(-2+2j)×i+(-6j)×(-1)+(-2-2j)×(-i)];

[0213] (where \(i = \sqrt{-1}\) is the imaginary unit);

[0214] x(1)=1 / 4[1 - 2i + 2j + 6j + 2i + 2j];

[0215] x(1)=1 / 4[1 + 8j];

[0216] x(1)=1 / 4 + 2j;

[0217] For \(n = 2\):

[0218] x(2)=1 / 4[1 - 2 - 6j + 2];

[0219] =-(3 / 2)j;

[0220] For \(n = 3\):

[0221] x(3)=1 / 4[1 + 2i - 6j - 2 - 2i];

[0222] =-3 / 4-(3 / 2)j;

[0223] The final time-domain signal is calculated.

[0224] Example 2:

[0225] As Figures 1-7 shown, the present invention provides a technical solution: a data processing method for an accelerometer driving and control system based on optoelectronic signals, the method comprising:

[0226] Obtain the signal to be processed, and obtain the original signal collected by the optoelectronic sensor;

[0227] Identify and remove data, perform identification processing on the input original signal, and remove the irrelevant data therein;

[0228] Enhance the noise suppression ability, estimate the state of the system through the data containing noise, and adjust the filtering parameters according to the current environmental conditions to achieve the best noise suppression effect;

[0229] Purify the useful signal, use the reference signal to estimate and remove the noise components in the main signal to obtain a pure useful signal;

[0230] Collect and iterate on the characteristics to stabilize the noise, continuously detect the noise with relatively stable characteristics, and after reaching a certain frequency, record it in a specific noise database;

[0231] Remove the noise with stable characteristics. Detect the noise with relatively stable characteristics in the original signal through an algorithm, compare it with a specific noise database, and remove the noise in the specific noise database;

[0232] Algorithm selection: Based on the amount of data, judge the computing time of the time-domain convolution algorithm and the moving average filtering algorithm when processing the same data, and select the algorithm with shorter computing time. The specific steps are as follows:

[0233] S1: Select the time-domain convolution algorithm and convert the time-domain signal to the frequency-domain signal;

[0234] S2: Process the frequency-domain signal using the time-domain convolution algorithm. The algorithm formula is as follows:

[0235] Among them, Y(k) is the frequency-domain representation of the output signal, X(k) is the frequency-domain representation of the input signal, and H(k) is the frequency-domain representation of the filter;

[0236] Substitute the data:

[0237] N = 4;

[0238] x(0) = 1 (x(0) can be any real number. Assume it is 1 as an example);

[0239] x(1) = -2 + 2j (obtained from x(3) = x(1));

[0240] H(k) = 1;

[0241] Now, we can calculate Y(k):

[0242] Y(0) = x(0) × H(0);

[0243] = 1 × 1;

[0244] = 1;

[0245] Y(1) = x(1) × H(1);

[0246] = (-2 + 2j) × 1;

[0247] = -2 + 2j;

[0248] Y(2) = x(2) × H(2);

[0249] = (-6j) × 1;

[0250] = -6j;

[0251] Y(3) = x(3) × H(3);

[0252] = (-2 - 2j) × 1;

[0253] =-2 - 2j;

[0254] Therefore, the frequency-domain representation of the output signal is obtained.

[0255] S4: After the calculation is completed, convert the frequency-domain signal back into a time-domain signal;

[0256] Simplify the algorithm iteration and the stop condition of the algorithm iteration process, so that the algorithm ends the loop on the premise of meeting the accuracy requirements and the number of iterations.

[0257] The identified removal data preprocesses the input original signal, removes the irrelevant part through detrending processing, trims the irrelevant part of the signal, and reduces the amount of calculation.

[0258] It should be noted that the input original signal is preprocessed to identify and remove irrelevant data. This is achieved by eliminating the linear or non-linear trend components in the signal through detrending processing, and then subtracting these trend components from the original signal. Analyze the signal to determine and trim the parts that are clearly irrelevant to the target analysis, such as silent segments, background noise, or invalid data within a specific time period. This step helps to reduce the amount of data and complexity of subsequent calculations. In addition, apply a window function at both ends of the signal, which makes the signal edge transition more smoothly to zero, thereby effectively reducing spectral leakage, significantly reducing the unnecessary computational burden, and improving the quality of subsequent signal processing and analysis, ensuring that the final result is more accurate and reliable, and can greatly improve work efficiency and analysis accuracy.

[0259] The enhanced noise suppression ability sets the initial state estimate and its covariance matrix, predicts the state at the current moment based on the state estimate at the previous moment, corrects the predicted value through signal data, adjusts the filter parameters, and dynamically suppresses the noise.

[0260] It should be noted that based on the state estimate at the previous moment, the state at the current moment is predicted through the system model, and the predicted value is corrected using the actual measurement data to obtain a more accurate state estimate. The specific steps are as follows: Set the initial state and its covariance matrix. Then, at each time step, use the state transition model to predict the current state and its covariance, and then calculate the gain based on the actual measurement value and update the state estimate and covariance. In addition, evaluate the noise level in the current environment based on the real-time collected data, and dynamically adjust the process noise covariance and measurement noise covariance, thereby optimizing the filtering effect, achieving efficient suppression of noise, reducing noise interference, and improving the signal quality.

[0261] The steps for the pure useful signal are as follows:

[0262] L1: Collect the main signal, which consists of the useful signal and the noise signal;

[0263] L2: Set the reference signal. The calculation formula for the reference signal is as follows:

[0264] Substitute the data for calculation. There are the following two simple discrete-time reference signals and filters:

[0265] r1(n) = {1, 0, 0, 0} (a unit impulse signal);

[0266] r2(n) = {1, 1, 1, 1} (a constant signal);

[0267] h(n) = {21, 21} (a simple averaging filter);

[0268] For r1(n) and h(n), we can calculate:

[0269] y1(0) = 1 × 21 = 21;

[0270] y1(1) = 0 × 21 + 1 × 21 = 21;

[0271] y1(2) = 0 for n > 1;

[0272] Where r is the i-th reference signal, h(t) is the filter used to adjust the i-th reference signal, y(t) is the adjusted i-th reference signal, and Fori is the range of values of the index variable i;

[0273] L3: Perform a weighted sum of all the adjusted reference signals to obtain an estimate of the noise signal. The formula is as follows:

[0274] Substitute the data,

[0275] w1 = 1,

[0276] y1(0) = 21;

[0277] y1(1) = 21;

[0278] n(0) = w1y1(0);

[0279] = 1 × 2;

[0280] = 2;

[0281] n(1) = w1y1(1);

[0282] = 1 × 2;

[0283] = 2;

[0284] Where w is the weighting coefficient of the i-th adjusted reference signal, and n(t) is the estimated value of the noise signal;

[0285] L4: Useful signal extraction. Subtract the estimated noise signal from the main signal to obtain an estimate of the pure useful signal. The calculation formula is as follows:

[0286] Where S(t) is the estimate of the pure useful signal, x(t) is the main signal, and n(t) is the estimate of the noise signal;

[0287] Substitute the data,

[0288] n(0) = 2;

[0289] n(1) = 2;

[0290] x(0) = 5;

[0291] x(1) = 7;

[0292] Now, we can calculate s(t):

[0293] s(0) = x(0) - n(0);

[0294] = 5 - 2;

[0295] = 3;

[0296] s(1) = x(1) - n(1);

[0297] = 7 - 2;

[0298] = 5;

[0299] Therefore, at time points t = 0 and t = 1, the estimates of the pure useful signal are s(0) = 3 and s(1) = 5 respectively.

[0300] The above-mentioned collection iteration feature stabilizes the noise, continuously detects the relatively stable noise. When the same relatively stable noise reaches the set number of times, it is recorded in the specific noise database for spectrum subtraction data model calculation to reduce the computational burden.

[0301] The above-mentioned algorithm is used to remove the relatively stable noise in the main signal. The algorithm formula is as follows:

[0302]

[0303] Where S(f) represents the estimated spectrum of the pure useful signal, X(f) represents the spectrum of the main signal, α represents the adjustment factor, and N(f) represents the estimated spectrum of the noise;

[0304] Substitute the following data:

[0305] The value of the spectrum X(f) of the main signal at a certain frequency point f0 is X(f0) = 10;

[0306] The estimated value of the noise spectrum N(f) at the same frequency point f0 is N(f0) = 2;

[0307] The adjustment factor α = 0.5.

[0308] Substituting these values into the formula, we get:

[0309] S(f0) = max{10 - 0.5×2};

[0310] S(f0) = max{10 - 1};

[0311] S(f0) = max{9};

[0312] S(f0) = 9;

[0313] Therefore, at the frequency point f0, the estimated value of the spectrum of the pure useful signal S(f0) is 9.

[0314] The time-domain convolution algorithm is selected to convert the time-domain signal into a frequency-domain signal, and the conversion formula is as follows:

[0315] Where: X(k) is the frequency-domain representation of the input signal x(n), x(n) is the original input signal in the time domain, and e is the complex exponential function;

[0316] Substitute the sample values of the time-domain signal x(n):

[0317] x(0) = 1;

[0318] x(1) = 2;

[0319] x(2) = 3;

[0320] x(3) = 4;

[0321] The number of samples N = 4;

[0322] Calculate the sample values of the frequency-domain signal X(k);

[0323] When k = 0:

[0324] X(0) = x(0) + x(1) + x(2) + x(3);

[0325] X(0) = 1 + 2 + 3 + 4;

[0326] X(0) = 10;

[0327] (Note: When k = 0, the complex exponential term is 1, so the summation is actually the sum of all sample values of the time-domain signal.)

[0328] When k = 1:

[0329] X(1) = 1×(1) + 2×(-j) + 3×(-1) + 4×(j);

[0330] X(1) = 1 - 2j - 3 + 4j; X(1) = -2 + 2j;

[0331] (Note: Here we used the properties of the complex exponential function, specifically, e / 2 = -j and e / 2 = j.)

[0332] Similarly, calculate the values of X(k) for k = 2 and k = 3. Due to the symmetry of the DFT (for real input signals), X(N - k) is the conjugate complex of X(k) (i.e., the real parts are the same and the imaginary parts have opposite signs). Therefore, if X(1) has been calculated, then X(3) is the conjugate complex of X(1). Similarly, X(2) will be a pure imaginary number (for real input signals and DFTs of even length), and the absolute value of its imaginary part is related to X(0) and X(N / 2).

[0333] Give the calculation results of X(2) and X(3) (using the same properties of the complex exponential function):

[0334] When k = 2:

[0335] X(2) = -6j;

[0336] When k = 3:

[0337] X(3) = X(1)* = -2 - 2j;

[0338] Here X(1)* represents the conjugate complex of X(1).

[0339] The conversion of the frequency-domain signal back to the time-domain signal is as follows:

[0340] where x(n) is the finally obtained time-domain signal, X(k) is the processed frequency-domain signal, e is the conjugate form of the complex exponential function, and 1 / N is the normalization factor;

[0341] Substitute the data,

[0342] Y(0) = 1;

[0343] Y(1) = -2 + 2j;

[0344] Y(2) = -6j;

[0345] Y(3) = -2 - 2j;

[0346] N = 4 (because there are four frequency-domain samples);

[0347] Now, substitute these data into the IDFT formula to calculate the time-domain signal x(n):

[0348] For n = 0:

[0349] x(0) = 1 / 4[Y(0) + Y(1) + Y(2) + Y(3)];

[0350] x(0) = 1 / 4[1 + (-2 + 2j) + (-6j) + (-2 - 2j)];

[0351] x(0) = 1 / 4[-3 - 6j];

[0352] x(0) = -3 / 4 - 3 / 2j;

[0353] For n = 1:

[0354] x(1) = 1 / 4[Y(0)×e + Y(1)×ej + Y(2)×e + Y(3)×e];

[0355] x(1) = 1 / 4[1 + (-2 + 2j)×i + (-6j)×(-1) + (-2 - 2j)×(-i)];

[0356] (where i = -1 is the imaginary unit);

[0357] x(1) = 1 / 4[1 - 2i + 2j + 6j + 2i + 2j];

[0358] x(1) = 1 / 4[1 + 8j];

[0359] x(1) = 1 / 4 + 2j;

[0360] For n = 2:

[0361] x(2) = 1 / 4[1 - 2 - 6j + 2];

[0362] = -(3 / 2)j;

[0363] For n = 3:

[0364] x(3) = 1 / 4[1 + 2i - 6j - 2 - 2i];

[0365] = -3 / 4 - (3 / 2)j;

[0366] The final time-domain signal is calculated.

[0367] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended embodiments and their equivalents.

Claims

1. A data processing method for an accelerometer drive and control system based on photoelectric signals, characterized in that: The method comprises: Obtain the signal to be processed and the original signal collected by the photoelectric sensor; Identify and remove data, identify and process the input original signal and remove irrelevant data; Enhance noise suppression capability, estimate the system status through noisy data, and adjust the filter parameters according to the current environmental conditions to achieve the best noise suppression effect; Pure useful signal, using the reference signal to estimate and remove the noise components in the main signal to obtain a pure useful signal; Collect iterative characteristic stable noise, continuously detect relatively stable noise, and enter it into the specific noise database after reaching a certain frequency; Remove noise with stable characteristics. Use an algorithm to detect relatively stable noise in the original signal, compare it with a specific noise database, and remove the noise in the specific noise database. Algorithm selection: Based on the amount of data, determine the calculation time of the time domain convolution algorithm and the moving average filter algorithm when processing the same data, and select the algorithm with shorter calculation time. The specific steps are as follows: S1: Set the threshold. When the data volume is lower than the threshold, select the sliding average filtering algorithm. S2: Set a threshold. When the data volume is higher than the threshold, select a time domain convolution algorithm to convert the time domain signal into a frequency domain signal. S3: Use the time domain convolution algorithm to process the frequency domain signal. The algorithm formula is as follows: Among them, Y(k) is the frequency domain representation of the output signal, X(k) is the frequency domain representation of the input signal, and H(k) is the frequency domain representation of the filter; S4: After the calculation is completed, the frequency domain signal is reversely converted into a time domain signal; Simplify the algorithm iteration, simplify the stop condition of the algorithm iteration process, so that the algorithm ends the cycle under the premise of meeting the accuracy requirements and the number of iterations; the identification and removal of data, preprocessing of the input original signal, removing the non-correlated parts through detrending processing, cutting out the irrelevant parts of the signal, and reducing the amount of calculation; the enhanced noise suppression capability, setting the initial state estimate and its covariance matrix, predicting the state at the current moment based on the state estimate at the previous moment, correcting the predicted value through signal data, adjusting the filter parameters, and dynamically suppressing the noise; the steps of purifying useful signals are as follows: L1: collect the main signal, which is composed of useful signal and noise signal; L2: Set the reference signal. The reference signal calculation formula is as follows: Where r(t) is the i-th reference signal, h(t) is the filter used to adjust the i-th reference signal, y(t) is the adjusted i-th reference signal, and Fori is the value range of the index variable i; L3: All adjusted reference signals are weighted and summed to obtain an estimate of the noise signal. The formula is as follows: Among them, w i is the weighting coefficient of the i-th adjusted reference signal, and n(t) is the estimated value of the noise signal; L4: Useful signal extraction, subtract the estimated noise signal from the main signal to obtain the estimated value of the pure useful signal. The calculation formula is as follows: Among them, S(t) is the estimated value of the pure useful signal, x(t) is the main signal, and n(t) is the estimated value of the noise signal; The method collects stable noise with iterative characteristics and continuously detects noise with relatively stable characteristics. When the same noise with relatively stable characteristics is detected for a set number of times, it is entered into a specific noise database for spectrum subtraction data model calculation to reduce the calculation burden. The method removes the noise with relatively stable characteristics in the main signal through an algorithm, and the algorithm formula is as follows: Wherein, S(f) represents the estimated value of the spectrum of the pure useful signal, X(f) represents the spectrum of the main signal, α represents the adjustment factor, and N(f) represents the estimated value of the noise spectrum.

2. The data processing method of the accelerometer driving and control system based on photoelectric signals according to claim 1, characterized in that: When the data volume is lower than the threshold, the sliding average filtering algorithm is selected, and the algorithm formula is as follows: Where: y(n) is the output signal, that is, the signal after sliding average filtering, 1 / M represents the average coefficient of the signal samples in the sliding window, ∑ M-1 k=0 represents a summation operation, x(nk) represents the input signal, n represents the current time index or sample index, and (nk) represents the time index of different samples in the current window.

3. The data processing method of the accelerometer driving and control system based on photoelectric signals according to claim 1, characterized in that: When the data volume is higher than the threshold, the time domain convolution algorithm is selected to convert the time domain signal to the frequency domain signal. The conversion formula is as follows: Where: X(k) is the frequency domain representation of the input signal x(n), x(n) is the original input signal in the time domain, and e is a complex exponential function.

4. The data processing method of the accelerometer driving and control system based on photoelectric signals according to claim 1, characterized in that: The frequency domain signal is reversely converted into the time domain signal, and the conversion formula is as follows: Among them, x(n) is the final time domain signal, X(k) is the processed frequency domain signal, e is the conjugate form of the complex exponential function, 1 / N is the normalization factor, ∑ N-1 k=0 Represents a sum operation.

5. An accelerometer drive and control system based on photoelectric signals, characterized in that: A data processing method for an accelerometer drive and control system based on photoelectric signals as described in any one of claims 1 to 4 is used.

Images