CN116682224B - Coal mine fire detection method and system based on power line network - Google Patents
Coal mine fire detection method and system based on power line network Download PDFInfo
- Publication number
- CN116682224B CN116682224B CN202310968822.8A CN202310968822A CN116682224B CN 116682224 B CN116682224 B CN 116682224B CN 202310968822 A CN202310968822 A CN 202310968822A CN 116682224 B CN116682224 B CN 116682224B
- Authority
- CN
- China
- Prior art keywords
- fire
- signal
- microwave radar
- time
- metal interference
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention relates to a coal mine fire detection method and system based on a power line network. Constructing a metal interference filter according to the characteristics of the historical microwave radar reflected signals and the metal interference signals; performing filtering processing on the microwave radar reflected signal through convolution operation; calculating correlation coefficients between all the extracted features and fire conditions, and screening out features with high correlation degree; performing fuzzy logic analysis on the fire related characteristics to judge whether fire exists or not; if a fire exists, analyzing the change trend of the fire feature in time according to the fire related feature and through time sequence analysis, accurately filtering metal interference and retaining reliable signals through construction of a specific filter and filtering aiming at metal interference signals, then reasoning through fuzzy logic to obtain a reliable fire judgment result, and analyzing the persistence of the fire through time sequence analysis so as to better solve the fire development and evolution process and provide good guidance for fire prevention control.
Description
Technical Field
The invention belongs to the technical field of coal mine fire detection, and particularly relates to a coal mine fire detection method and system based on a power line network.
Background
In a coal mine fire detection environment, metal interference is prevalent. Coal mines often contain a large number of metal structures, equipment and tools that reflect and scatter microwave radar signals, thereby introducing metal interference signals.
The metal interference has a great influence on the radar detection of fire. Microwave radar typically relies on the received reflected signal to detect fire. The metal interference signal may overlap or cover the fire signal such that the fire signal becomes obscured or masked, resulting in a decrease in the accuracy of fire detection.
Although the common filtering pretreatment operation can filter metal interference signals to a certain extent, the effect is limited. Metal interference signals typically have a high strength and wide bandwidth, so specific filters need to be designed to effectively filter out these disturbances.
Disclosure of Invention
The invention provides a coal mine fire detection system and device based on a power line network, and aims to solve the problems mentioned in the background art.
The invention is realized in this way, and provides a coal mine fire detection method based on a power line network, which installs a microwave radar at a wellhead or a loading and unloading area of a coal mine, and comprises the following steps:
Acquiring a historical microwave radar reflected signal, and constructing a metal interference filter according to characteristics of the historical microwave radar reflected signal and a metal interference signal, wherein the method specifically comprises the following steps of:
preprocessing the historical microwave radar reflected signal, including removing direct current component and normalizing,
converting the preprocessed microwave radar reflected signal from the time domain to the frequency domain through Fourier transformation, calculating correlation coefficients between each frequency component and other frequency components in the frequency spectrum,
comparing and analyzing the correlation coefficient between each frequency component and other frequency components with a preset correlation coefficient threshold, if the correlation coefficient exceeds the preset correlation coefficient threshold, judging that the corresponding frequency component is related to the metal interference signal, setting the area corresponding to the frequency component as a metal interference area,
analyzing the signal characteristics of the metal interference region to determine the frequency range and energy distribution to be filtered or reduced, and constructing a desired spectrogram or magnitude spectrum based on the analysis result,
converts the desired spectrogram or amplitude spectrum into a time domain signal to obtain a time domain waveform of the desired output signal,
the historical microwave radar reflected signal is input into the adaptive filter as an input signal to obtain an actual output signal, wherein the weight of the adaptive filter is initialized when the microwave radar reflected signal is input for the first time,
Calculating the difference between the actual output signal and the desired output signal to obtain an error, e (n) =Σ { [ y ] actual (n)-y expected (n)] 2 N is the number of data points, N is the data points, y actual (n) time domain value of actual output signal for each data point, y expected (n) time domain values of the desired output signal for each data point, y actual (n)-y expected (n) is the difference between the time domain values of the actual output signal and the desired output signal for the same data point,
the weight of the adaptive filter is adjusted according to the error and the input signal, and the adjustment formula of the weight is as follows: w (n+1) =w (n) +μ×e (n) ×e (n), where w (n+1) is an adjusted weight, w (n) is a current weight, μ is a step size factor, e (n) is an error, x (n) is an input signal, filtering is performed with an adjusted adaptive filter,
repeating the steps until the error gradually converges to be smaller than or equal to a preset error threshold value, setting the error as a final error, and adjusting the weight of the adaptive filter according to the final error to obtain a metal interference filter;
collecting real-time microwave radar reflected signals, inputting the real-time microwave radar reflected signals as input signals into a metal interference filter, and performing filtering processing on the microwave radar reflected signals through convolution operation to obtain target signals for filtering the metal interference signals;
Extracting features from the target signal, calculating correlation coefficients between all the extracted features and fire, screening out features with high correlation degree, and setting the features as fire correlation features;
performing fuzzy logic analysis on the fire related characteristics to judge whether fire exists or not;
if the fire exists, analyzing the change trend of the fire characteristics in time according to the fire related characteristics and through time sequence analysis to obtain the persistence of the fire.
Further, the step of performing fuzzy logic analysis on the fire related features to determine whether a fire exists includes:
taking each fire related characteristic as a fuzzy variable, and determining the variable range of the fire related characteristic to be fuzzified;
dividing each fuzzy variable into a plurality of fuzzy sets according to the property and variable range of the fire related characteristics;
setting a membership function for each fuzzy set of fire related features, and substituting real-time feature values of the fire related features into corresponding membership functions to calculate membership degrees of the fire related features in each fuzzy set;
combining membership degrees of the real-time characteristic values in different fuzzy sets to form a fuzzy output set of fire related characteristics;
Calculating a correlation coefficient matrix among all fire related features, and calculating the weight of each fire related feature according to the correlation coefficient matrix among all fire related features;
and accumulating the products of the weights of the fire related features and the maximum membership degree in the fuzzy output set, and comparing the accumulated value with a fire threshold value to make decision judgment on whether fire exists in the coal mine.
Further, the step of calculating the correlation coefficient matrix between all the fire related features and calculating the weight of each fire related feature according to the correlation coefficient matrix between all the fire related features includes:
calculating pearson correlation coefficients between fire related features to obtain an m x m correlation coefficient matrix, wherein m is the number of fire related features;
calculating the global average value of each fire related feature, subtracting the global average value of the corresponding feature from the real-time feature value of each fire related feature to obtain data after centralized processing and forming a data matrix;
covariance calculation is carried out on the data matrix to obtain covariance among relevant characteristics of each fire condition and form a covariance matrix;
Performing eigenvalue decomposition on the covariance matrix to obtain eigenvalues and corresponding eigenvectors;
sorting the eigenvalues obtained by decomposing the eigenvalues according to the order from big to small, and sorting the eigenvectors according to the eigenvalue sorting;
selecting feature vectors corresponding to the top k feature values which are ranked at the front as main components;
taking the variance interpretation proportion of each main component as the weight of the corresponding fire related characteristics, wherein the variance interpretation proportion of the main component is the sum of the characteristic value corresponding to the main component divided by the total characteristic value;
the weights of the fire related features were normalized so that the sum of the weights was 1.
Further, the step of analyzing the trend of the fire feature over time according to the fire related feature and through time sequence analysis to obtain the persistence of the fire comprises:
time sequence data of relevant characteristics of each fire condition are respectively obtained, wherein the time sequence data comprise time points and corresponding characteristic values.
Carrying out centralization treatment on the time series data, namely subtracting the initial time from the time point;
fitting each polynomial function and time sequence data through a least square method to obtain each fitted curve, wherein each polynomial function is a function of gradually increasing the order from a simple first-order polynomial function;
Calculating residual errors of each fitting curve and actual data points to obtain fitting errors, screening out a fitting curve with the smallest fitting error, and determining the fitting curve as the best fitting curve of relevant characteristics of corresponding fire;
determining the slope of a trend line of the relevant characteristic of the corresponding fire according to the coefficient of a polynomial function corresponding to the best fit curve of the relevant characteristic of the corresponding fire;
and determining the curvature of the trend line of the relevant characteristic of the corresponding fire according to the second derivative of the polynomial function corresponding to the best fit curve of the relevant characteristic of the corresponding fire.
Furthermore, the step of inputting the real-time microwave radar reflected signal as an input signal into the metal interference filter and performing filtering processing on the microwave radar reflected signal through convolution operation to obtain a target signal for filtering the metal interference signal includes:
converting the metal interference filter into a frequency domain, and multiplying the metal interference filter with an input signal in the frequency domain, wherein the calculation formula is as follows:
X(f)＝FFT(x(t))，
Y(f)＝X(f)*H(f)，
wherein X (f) is the frequency domain representation of the input signal, FFT (X (t)) is the fourier transform of the input signal, Y (f) is the filtered frequency domain signal, and H (f) is the frequency domain representation of the filter;
the result of the multiplication is inverse transformed to convert the input signal back to the time domain, which represents Y (t) =ifft (Y (f)), where IFFT (Y (f)) is the inverse fourier transform of the filtered frequency domain signal.
The invention also provides a coal mine fire detection system based on the power line network, which is used for executing a coal mine fire detection method based on the power line network, and installing a microwave radar at a wellhead or a loading and unloading area of a coal mine, and comprises the following steps:
and a filter construction module: the method is used for acquiring historical microwave radar reflected signals and constructing a metal interference filter according to the characteristics of the historical microwave radar reflected signals and metal interference signals, and specifically comprises the following steps:
preprocessing the historical microwave radar reflected signal, including removing direct current component and normalizing,
converting the preprocessed microwave radar reflected signal from the time domain to the frequency domain through Fourier transformation, calculating correlation coefficients between each frequency component and other frequency components in the frequency spectrum,
comparing and analyzing the correlation coefficient between each frequency component and other frequency components with a preset correlation coefficient threshold, if the correlation coefficient exceeds the preset correlation coefficient threshold, judging that the corresponding frequency component is related to the metal interference signal, setting the area corresponding to the frequency component as a metal interference area,
analyzing the signal characteristics of the metal interference region to determine the frequency range and energy distribution to be filtered or reduced, and constructing a desired spectrogram or magnitude spectrum based on the analysis result,
Converts the desired spectrogram or amplitude spectrum into a time domain signal to obtain a time domain waveform of the desired output signal,
the historical microwave radar reflected signal is input into the adaptive filter as an input signal to obtain an actual output signal, wherein the weight of the adaptive filter is initialized when the microwave radar reflected signal is input for the first time,
calculating the difference between the actual output signal and the desired output signal to obtain an error, e (n) =Σ { [ y ] actual (n)-y expected (n)] 2 N is the number of data points, N is the data points, y actual (n) time domain value of actual output signal for each data point, y expected (n) time domain values of the desired output signal for each data point, y actual (n)-y expected (n) is the difference between the time domain values of the actual output signal and the desired output signal for the same data point,
the weight of the adaptive filter is adjusted according to the error and the input signal, and the adjustment formula of the weight is as follows: w (n+1) =w (n) +μ×e (n) ×e (n), where w (n+1) is an adjusted weight, w (n) is a current weight, μ is a step size factor, e (n) is an error, x (n) is an input signal, filtering is performed with an adjusted adaptive filter,
repeating the steps until the error gradually converges to be smaller than or equal to a preset error threshold value, setting the error as a final error, and adjusting the weight of the adaptive filter according to the final error to obtain a metal interference filter;
And a filtering module: the method comprises the steps of collecting real-time microwave radar reflected signals, inputting the real-time microwave radar reflected signals serving as input signals into a metal interference filter, and performing filtering processing on the microwave radar reflected signals through convolution operation to obtain target signals for filtering the metal interference signals;
and the feature extraction module is used for: the method comprises the steps of extracting features from a target signal, calculating correlation coefficients between all the extracted features and fire, screening out features with high correlation degree, and setting the features as fire correlation features;
the fire judgment module is used for: the fuzzy logic analysis is used for carrying out fuzzy logic analysis on the fire related characteristics so as to judge whether fire exists or not;
fire analysis module: and if the fire exists, analyzing the time change trend of the fire characteristics according to the fire related characteristics and through time sequence analysis to obtain the persistence of the fire.
Still further, the detection system also includes a 380 volt AC power line for forming the power line network to provide power supply for each mining power device of the coal mine.
Compared with the prior art, the metal interference filter is constructed by utilizing the characteristic of the metal interference signal in the microwave radar signal, and the constructed metal interference filter is used for filtering the real-time microwave radar signal, so that the metal interference signal in the microwave radar signal can be well filtered, the metal interference can be more accurately filtered, and a reliable target signal can be reserved; the fire related characteristics are analyzed by using a fuzzy logic analysis method, and reasoning and judgment are carried out by using a fuzzy rule and a membership function, so that the uncertainty and the ambiguity of the fire can be well processed, and the accuracy and the reliability of the fire judgment are improved; according to the time variation trend of the fire characteristics, the duration of the fire is obtained, and the development and evolution process of the fire can be more comprehensively understood by a time sequence analysis-based method, so that deeper fire information is provided, and a good enlightenment effect is achieved for formulating a fire prevention and control strategy of a coal mine.
When the metal interference filter is constructed, the frequency components related to the metal interference signal are judged by calculating the correlation coefficient between different frequency components in the frequency spectrum, and the components with the phase relation number exceeding the preset threshold value are determined as metal interference areas, so that the metal interference areas can be more accurately positioned, and the subjective judgment only by observation is avoided; the adaptive filter is adopted to filter the historical microwave radar reflected signals, and the performance of the adaptive filter is gradually optimized through error calculation and weight adjustment so as to meet the requirement of expected output signals, and the adaptive filter can flexibly adapt to the change of signal characteristics, so that metal interference signals can be filtered more effectively; and gradually converging the error to a preset error threshold value to ensure that the metal interference filter can reach the expected performance, so that the filter can gradually approach the optimal solution, and the filtering effect and the convergence speed are improved.
Drawings
FIG. 1 is a schematic flow chart of a coal mine fire detection method based on a power line network;
fig. 2 is a system block diagram of a coal mine fire detection system based on a power line network.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
Example 1
Referring to fig. 1, a first embodiment provides a method for detecting fire conditions in a coal mine based on a power line network, wherein a microwave radar is installed at a wellhead or a loading and unloading area of the coal mine, and the method comprises steps S101 to S105:
s101, acquiring a historical microwave radar reflected signal, and constructing a metal interference filter according to characteristics of the historical microwave radar reflected signal and a metal interference signal, wherein the method specifically comprises the following steps:
preprocessing the historical microwave radar reflected signal, including removing direct current component and normalizing,
the preprocessed microwave radar reflected signal is converted from the time domain to the frequency domain through fourier transformation, and correlation coefficients between each frequency component and other frequency components in the frequency spectrum are calculated, which can be calculated through pearson correlation coefficients, spearman correlation coefficients and the like,
the correlation coefficient between each frequency component and other frequency components is compared with a preset correlation coefficient threshold value, if the correlation coefficient threshold value exceeds the preset correlation coefficient threshold value, the corresponding frequency component is judged to be related to the metal interference signal, the area corresponding to the frequency component is set as a metal interference area, in general, the frequency component with higher correlation coefficient indicates that stronger linear correlation exists between the frequency components, possibly the frequency component where the metal interference signal exists,
Analyzing the signal characteristics of the metal interference region to determine the frequency range and energy distribution to be filtered or reduced, and constructing a desired spectrogram or magnitude spectrum based on the analysis result,
converts the desired spectrogram or amplitude spectrum into a time domain signal to obtain a time domain waveform of the desired output signal,
the historical microwave radar reflected signal is input into the adaptive filter as an input signal to obtain an actual output signal, wherein the weight of the adaptive filter is initialized when the microwave radar reflected signal is input for the first time,
calculating the difference between the actual output signal and the desired output signal to obtain an error, e (n) =Σ { [ y ] actual (n)-y expected (n)] 2 N is the number of data points, N is the data points, y actual (n) time domain value of actual output signal for each data point, y expected (n) time domain values of the desired output signal for each data point, y actual (n)-y expected (n) is the difference between the time domain values of the actual output signal and the desired output signal for the same data point,
the weight of the adaptive filter is adjusted according to the error and the input signal, and the adjustment formula of the weight is as follows: w (n+1) =w (n) +μ×e (n) ×e (n), where w (n+1) is an adjusted weight, w (n) is a current weight, μ is a step size factor, e (n) is an error, x (n) is an input signal, filtering is performed with an adjusted adaptive filter,
Repeating the steps until the error gradually converges to be smaller than or equal to a preset error threshold value, setting the error as a final error, and adjusting the weight of the adaptive filter according to the final error to obtain the metal interference filter.
It should be noted that, through preprocessing the historical microwave radar reflected signal, such as removing direct current component and normalization, and fourier transformation to convert the signal from time domain to frequency domain, various signal processing methods are comprehensively utilized, and the combination can more comprehensively analyze the frequency spectrum characteristics of the signal, and provide accurate frequency information for the subsequent design of the metal interference filter.
The statistical-based correlation measurement method can more accurately position the metal interference area, and avoids subjective judgment only by observation.
The adaptive filter is adopted to filter the historical microwave radar reflected signals, and the performance of the adaptive filter is gradually optimized through error calculation and weight adjustment so as to meet the requirement of expected output signals, and the adaptive filter can be flexibly adapted to the change of signal characteristics by application, so that metal interference signals can be filtered more effectively.
The adaptive filter is adjusted through error convergence, and as the weight is adjusted, the actual output signal gradually approaches the expected output signal, so that the difference between the actual output signal and the expected output signal is reduced to an acceptable range, and finally the actual output signal is stabilized. When the error reaches a satisfactorily small value or approaches a steady state, it is interpreted that the error of the filter has converged. In the learning process of the adaptive filter, the weight is adjusted according to the error signal, and the error is reduced by continuously observing the difference between the actual output signal and the expected output signal and updating the weight until the error converges. The convergence of the error means that the filter has found the optimal weight setting that will produce the actual output closest to the desired output signal for a given input signal.
S102, acquiring real-time microwave radar reflected signals, inputting the real-time microwave radar reflected signals as input signals into a metal interference filter, and performing filtering processing on the microwave radar reflected signals through convolution operation to obtain target signals for filtering the metal interference signals.
The step of inputting the real-time microwave radar reflected signal as an input signal into a metal interference filter and performing filtering processing on the microwave radar reflected signal through convolution operation to obtain a target signal for filtering the metal interference signal comprises the following steps:
Converting the metal interference filter into a frequency domain, and multiplying the metal interference filter with an input signal in the frequency domain, wherein the calculation formula is as follows:
X(f)＝FFT(x(t))，
Y(f)＝X(f)*H(f)，
wherein X (f) is the frequency domain representation of the input signal, FFT (X (t)) is the fourier transform of the input signal, Y (f) is the filtered frequency domain signal, and H (f) is the frequency domain representation of the filter;
the result of the multiplication is inverse transformed to convert the input signal back to the time domain, which represents Y (t) =ifft (Y (f)), where IFFT (Y (f)) is the inverse fourier transform of the filtered frequency domain signal.
And S103, extracting features from the target signal, calculating correlation coefficients between all the extracted features and fire, screening out features with high correlation degree, and setting the features as fire correlation features.
S104, performing fuzzy logic analysis on the fire related characteristics to judge whether fire exists.
The step of performing fuzzy logic analysis on the fire related features to determine whether a fire exists includes:
taking each fire related characteristic as a fuzzy variable, and determining the variable range of the fire related characteristic to be fuzzified;
dividing each fuzzy variable into a plurality of fuzzy sets according to the property and variable range of the fire related characteristics;
Setting a membership function for each fuzzy set of fire related features, and substituting real-time feature values of the fire related features into corresponding membership functions to calculate membership degrees of the fire related features in each fuzzy set;
combining membership degrees of the real-time characteristic values in different fuzzy sets to form a fuzzy output set of fire related characteristics, wherein the fuzzy output can be represented by vectors or matrixes, and each element represents the membership degree of the corresponding fuzzy set;
calculating a correlation coefficient matrix among all fire related features, and calculating the weight of each fire related feature according to the correlation coefficient matrix among all fire related features;
and accumulating the products of the weights of the fire related features and the maximum membership degree in the fuzzy output set, and comparing the accumulated value with a fire threshold value to make decision judgment on whether fire exists in the coal mine.
It should be noted that, the membership function is used to describe membership degrees between the feature value and the fuzzy set, and the common membership functions include a triangular membership function, a gaussian membership function, a trapezoidal membership function, and the like, and appropriate membership function shapes and parameters are selected so as to accurately represent the membership degrees of the feature value in the fuzzy set. If the feature value exceeds the range of the fuzzy set, fuzzy boundary processing can be considered. For example, out-of-range feature values may be mapped to nearest fuzzy set boundaries and assigned a lower membership.
If the fire related features are intensity features, the fire related features can be divided into fuzzy sets such as low intensity, medium intensity and high intensity, and the real-time intensity value is assumed to be 20, and the membership degrees of the fire related features in the fuzzy sets such as low intensity, medium intensity and high intensity are calculated to be 0.8,0.5 and 0.2 according to a triangular membership function, so that the fuzzy output set {0.8,0.5,0.2}.
By using each fire related feature as a fuzzy variable and determining the variable range of the fire related feature to be fuzzified, the selection can be made according to the nature and actual condition of the fire related feature. The reasonable variable range and fuzzy set division can better capture the semantic information of the characteristic value and improve the accuracy of fuzzy logic analysis.
In the process of setting a membership function for each fuzzy set of fire related characteristics, membership function shapes and parameters suitable for the characteristics can be selected, membership degree relations between characteristic values and the fire can be flexibly expressed by different membership function shapes, such as triangles, trapezoids, gaussian curves and the like, and the contribution degree of the characteristic values to the fire can be better described through reasonable selection and adjustment.
Membership degrees of the real-time characteristic values in different fuzzy sets are combined to form a fuzzy output set of fire related characteristics, and comprehensive evaluation can be performed on fire through the maximum value, the weighted average value and the like of the membership degrees.
When the correlation coefficient matrix among all fire related features is calculated, the correlation among the features needs to be considered, and the correlation among different features is calculated, so that the comprehensive features of the fire can be better reflected.
The weight of each feature is calculated according to the correlation coefficient matrix among all fire related features, and the product of the weight and the maximum membership in the fuzzy output set is accumulated, so that the contribution degree of each feature to the fire can be accurately measured, and the decision judgment of whether the fire exists in the coal mine can be made by comparing the contribution degree with the fire threshold.
Further, the step of calculating the correlation coefficient matrix between all the fire related features and calculating the weight of each fire related feature according to the correlation coefficient matrix between all the fire related features includes:
calculating pearson correlation coefficients between fire related features to obtain an m x m correlation coefficient matrix, wherein m is the number of fire related features;
calculating the global average value of each fire related feature, subtracting the global average value of the corresponding feature from the real-time feature value of each fire related feature to obtain data after centralized processing and forming a data matrix; to ensure that the mean value of each feature is 0.
Covariance calculation is carried out on the data matrix to obtain covariance among relevant characteristics of each fire condition and form a covariance matrix;
performing eigenvalue decomposition on the covariance matrix to obtain eigenvalues and corresponding eigenvectors; eigenvalue decomposition is a form of representing a covariance matrix as the product of eigenvalues and corresponding eigenvectors, covariance matrix c=pdp T Wherein C is covariance matrix, P is eigenvector matrix, D is eigenvalue diagonal matrix, P T Is the transpose of the eigenvector matrix.
Sorting the eigenvalues obtained by decomposing the eigenvalues according to the order from big to small, and sorting the eigenvectors according to the eigenvalue sorting; thus, feature vectors with larger feature values correspond to more important variance interpretations in the data.
Selecting feature vectors corresponding to the top k feature values which are ranked at the front as main components;
taking the variance interpretation proportion of each main component as the weight of the corresponding fire related characteristics, wherein the variance interpretation proportion of the main component is the sum of the characteristic value corresponding to the main component divided by the total characteristic value;
the weights of the fire related features were normalized so that the sum of the weights was 1.
And S105, if the fire exists, analyzing the time change trend of the fire features according to the fire related features and through time sequence analysis to obtain the persistence of the fire.
The step of analyzing the time variation trend of the fire characteristics according to the fire related characteristics and through time sequence analysis to obtain the persistence of the fire comprises the following steps:
time sequence data of relevant characteristics of each fire condition are respectively obtained, wherein the time sequence data comprise time points and corresponding characteristic values.
The time series data is subjected to a centering process, i.e. the starting time is subtracted from the time point.
Fitting each polynomial function and the time series data by a least square method to obtain each fitting curve, wherein each polynomial function is a function with gradually increased order from a simple first-order polynomial function (linear fitting), such as a quadratic polynomial, a cubic polynomial and the like.
And calculating residual errors of each fitting curve and the actual data points to obtain fitting errors, screening out the fitting curve with the minimum fitting error, and determining the fitting curve as the best fitting curve of the relevant characteristics of the corresponding fire.
Determining the slope of a trend line of the relevant characteristic of the corresponding fire according to the coefficient of a polynomial function corresponding to the best fit curve of the relevant characteristic of the corresponding fire; the positive and negative of the slope represent the increasing or decreasing trend of the fire characteristics, the absolute value of the slope represents the changing speed, and the sign and the size of the slope are used for primarily judging whether the fire is gradually increased or gradually decreased.
Determining the curvature of a trend line of the relevant characteristic of the corresponding fire according to the second derivative of the polynomial function corresponding to the best fit curve of the relevant characteristic of the corresponding fire; wherein the curvature describes the extent of bending of the trend line, which can help determine whether the change in the fire characteristics is accelerating or slowing down. Positive curvature indicates that the trend line is convex upward and that the increase in fire characteristics may be accelerating; negative curvature indicates that the trend line is concave downward and the reduction in fire characteristics may be accelerating. The change speed of the change trend of the fire can be further analyzed through the change of the curvature.
It should be noted that, based on the fitted trend line, a polynomial function may be used to predict the characteristic value of the fire at the future time point. The change trend of the predicted value can be used for judging the future change direction of the fire characteristics. For example, if the predicted value exhibits a gradual upward trend, it may suggest that the fire characteristics continue to grow in the future; if the predicted value exhibits a gradual decrease trend, it may suggest that the fire characteristics continue to decrease in the future.
The metal interference filter is constructed by utilizing the characteristic of the metal interference signal in the microwave radar signal, and the real-time microwave radar signal is filtered through the constructed metal interference filter, so that the metal interference signal in the microwave radar signal can be well filtered, the metal interference is more accurately filtered, and a reliable target signal is reserved; the fire related characteristics are analyzed by using a fuzzy logic analysis method, and reasoning and judgment are carried out by using a fuzzy rule and a membership function, so that the uncertainty and the ambiguity of the fire can be well processed, and the accuracy and the reliability of the fire judgment are improved; according to the time variation trend of the fire characteristics, the duration of the fire is obtained, and the development and evolution process of the fire can be more comprehensively understood by a time sequence analysis-based method, so that deeper fire information is provided, and a good enlightenment effect is achieved for formulating a fire prevention and control strategy of a coal mine.
When the metal interference filter is constructed, the frequency components related to the metal interference signal are judged by calculating the correlation coefficient between different frequency components in the frequency spectrum, and the components with the phase relation number exceeding the preset threshold value are determined as metal interference areas, so that the metal interference areas can be more accurately positioned, and the subjective judgment only by observation is avoided; the adaptive filter is adopted to filter the historical microwave radar reflected signals, and the performance of the adaptive filter is gradually optimized through error calculation and weight adjustment so as to meet the requirement of expected output signals, and the adaptive filter can flexibly adapt to the change of signal characteristics, so that metal interference signals can be filtered more effectively; and gradually converging the error to a preset error threshold value to ensure that the metal interference filter can reach the expected performance, so that the filter can gradually approach the optimal solution, and the filtering effect and the convergence speed are improved.
Example two
Referring to fig. 2, a second embodiment provides a fire detection system for a coal mine based on a power line network, in which a microwave radar is installed at a wellhead or a loading and unloading area of the coal mine, including:
and a filter construction module: the method is used for acquiring historical microwave radar reflected signals and constructing a metal interference filter according to the characteristics of the historical microwave radar reflected signals and metal interference signals, and specifically comprises the following steps:
Preprocessing the historical microwave radar reflected signal, including removing direct current component and normalizing,
converting the preprocessed microwave radar reflected signal from the time domain to the frequency domain through Fourier transformation, calculating correlation coefficients between each frequency component and other frequency components in the frequency spectrum,
comparing and analyzing the correlation coefficient between each frequency component and other frequency components with a preset correlation coefficient threshold, if the correlation coefficient exceeds the preset correlation coefficient threshold, judging that the corresponding frequency component is related to the metal interference signal, setting the area corresponding to the frequency component as a metal interference area,
analyzing the signal characteristics of the metal interference region to determine the frequency range and energy distribution to be filtered or reduced, and constructing a desired spectrogram or magnitude spectrum based on the analysis result,
converts the desired spectrogram or amplitude spectrum into a time domain signal to obtain a time domain waveform of the desired output signal,
the historical microwave radar reflected signal is input into the adaptive filter as an input signal to obtain an actual output signal, wherein the weight of the adaptive filter is initialized when the microwave radar reflected signal is input for the first time,
Calculating the difference between the actual output signal and the desired output signal to obtain an error, e (n) =Σ { [ y ] actual (n)-y expected (n)] 2 N is the number of data points, N is the data points, y actual (n) time domain value of actual output signal for each data point, y expected (n) time domain values of the desired output signal for each data point, y actual (n)-y expected (n) is the difference between the time domain values of the actual output signal and the desired output signal for the same data point,
the weight of the adaptive filter is adjusted according to the error and the input signal, and the adjustment formula of the weight is as follows: w (n+1) =w (n) +μ×e (n) ×e (n), where w (n+1) is an adjusted weight, w (n) is a current weight, μ is a step size factor, e (n) is an error, x (n) is an input signal, filtering is performed with an adjusted adaptive filter,
repeating the steps until the error gradually converges to be smaller than or equal to a preset error threshold value, setting the error as a final error, and adjusting the weight of the adaptive filter according to the final error to obtain a metal interference filter;
the filter construction module is further configured to:
carrying out frequency spectrum analysis on the microwave radar reflected signal through fast Fourier transform to obtain the energy distribution of the signal on a frequency domain;
Searching and identifying a metal interference frequency band from the spectrogram according to the characteristics of the metal interference signal;
and separating the corresponding frequency domain signals according to the identified metal interference frequency bands to extract metal interference signals.
And a filtering module: the method is used for collecting real-time microwave radar reflected signals, inputting the real-time microwave radar reflected signals serving as input signals into a metal interference filter, and performing filtering processing on the microwave radar reflected signals through convolution operation to obtain target signals for filtering the metal interference signals.
The filtering module is further configured to:
converting the metal interference filter into a frequency domain, and multiplying the metal interference filter with an input signal in the frequency domain, wherein the calculation formula is as follows:
X(f)＝FFT(x(t))，
Y(f)＝X(f)*H(f)，
wherein X (f) is the frequency domain representation of the input signal, FFT (X (t)) is the fourier transform of the input signal, Y (f) is the filtered frequency domain signal, and H (f) is the frequency domain representation of the filter;
the result of the multiplication is inverse transformed to convert the input signal back to the time domain, which represents Y (t) =ifft (Y (f)), where IFFT (Y (f)) is the inverse fourier transform of the filtered frequency domain signal.
And the feature extraction module is used for: the method is used for extracting features from the target signal, calculating correlation coefficients between all the extracted features and fire, screening out features with high correlation degree, and setting the features as fire correlation features.
The fire judgment module is used for: and the fuzzy logic analysis is used for carrying out fuzzy logic analysis on the fire related characteristics to judge whether fire exists.
The fire judgment module is also used for:
taking each fire related characteristic as a fuzzy variable, and determining the variable range of the fire related characteristic to be fuzzified;
dividing each fuzzy variable into a plurality of fuzzy sets according to the property and variable range of the fire related characteristics;
setting a membership function for each fuzzy set of fire related features, and substituting real-time feature values of the fire related features into corresponding membership functions to calculate membership degrees of the fire related features in each fuzzy set;
combining membership degrees of the real-time characteristic values in different fuzzy sets to form a fuzzy output set of fire related characteristics;
calculating a correlation coefficient matrix among all fire related features, and calculating the weight of each fire related feature according to the correlation coefficient matrix among all fire related features;
and accumulating the products of the weights of the fire related features and the maximum membership degree in the fuzzy output set, and comparing the accumulated value with a fire threshold value to make decision judgment on whether fire exists in the coal mine.
The fire judgment module is also used for:
calculating pearson correlation coefficients between fire related features to obtain an m x m correlation coefficient matrix, wherein m is the number of fire related features;
calculating the global average value of each fire related feature, subtracting the global average value of the corresponding feature from the real-time feature value of each fire related feature to obtain data after centralized processing and forming a data matrix;
covariance calculation is carried out on the data matrix to obtain covariance among relevant characteristics of each fire condition and form a covariance matrix;
performing eigenvalue decomposition on the covariance matrix to obtain eigenvalues and corresponding eigenvectors;
sorting the eigenvalues obtained by decomposing the eigenvalues according to the order from big to small, and sorting the eigenvectors according to the eigenvalue sorting;
selecting feature vectors corresponding to the top k feature values which are ranked at the front as main components;
taking the variance interpretation proportion of each main component as the weight of the corresponding fire related characteristics, wherein the variance interpretation proportion of the main component is the sum of the characteristic value corresponding to the main component divided by the total characteristic value;
the weights of the fire related features were normalized so that the sum of the weights was 1.
Fire analysis module: and if the fire exists, analyzing the time change trend of the fire characteristics according to the fire related characteristics and through time sequence analysis to obtain the persistence of the fire.
The fire analysis module is also used for:
time sequence data of relevant characteristics of each fire condition are respectively obtained, wherein the time sequence data comprise time points and corresponding characteristic values.
Carrying out centralization treatment on the time series data, namely subtracting the initial time from the time point;
fitting each polynomial function and time sequence data through a least square method to obtain each fitted curve, wherein each polynomial function is a function of gradually increasing the order from a simple first-order polynomial function;
calculating residual errors of each fitting curve and actual data points to obtain fitting errors, screening out a fitting curve with the smallest fitting error, and determining the fitting curve as the best fitting curve of relevant characteristics of corresponding fire;
determining the slope of a trend line of the relevant characteristic of the corresponding fire according to the coefficient of a polynomial function corresponding to the best fit curve of the relevant characteristic of the corresponding fire;
and determining the curvature of the trend line of the relevant characteristic of the corresponding fire according to the second derivative of the polynomial function corresponding to the best fit curve of the relevant characteristic of the corresponding fire.
The detection system also comprises a 380 volt AC power line which is used for forming the power line network and providing power supply for each mining power device of the coal mine.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, and alternatives falling within the spirit and principles of the invention.
Claims (7)
1. The method for detecting the fire condition of the coal mine based on the power line network is characterized in that a microwave radar is installed at a wellhead or a loading and unloading area of the coal mine, and the method comprises the following steps:
acquiring a historical microwave radar reflected signal, and constructing a metal interference filter according to characteristics of the historical microwave radar reflected signal and a metal interference signal, wherein the method specifically comprises the following steps of:
preprocessing the historical microwave radar reflected signal, including removing direct current component and normalizing,
converting the preprocessed microwave radar reflected signal from the time domain to the frequency domain through Fourier transformation, calculating correlation coefficients between each frequency component and other frequency components in the frequency spectrum,
comparing and analyzing the correlation coefficient between each frequency component and other frequency components with a preset correlation coefficient threshold, if the correlation coefficient exceeds the preset correlation coefficient threshold, judging that the corresponding frequency component is related to the metal interference signal, setting the area corresponding to the frequency component as a metal interference area,
Analyzing the signal characteristics of the metal interference region to determine the frequency range and energy distribution to be filtered or reduced, and constructing a desired spectrogram or magnitude spectrum based on the analysis result,
converts the desired spectrogram or amplitude spectrum into a time domain signal to obtain a time domain waveform of the desired output signal,
the historical microwave radar reflected signal is input into the adaptive filter as an input signal to obtain an actual output signal, wherein the weight of the adaptive filter is initialized when the microwave radar reflected signal is input for the first time,
calculating the difference between the actual output signal and the desired output signal to obtain an error, e (n) =Σ { [ y ] actual （n）-y expected （n）] 2 N is the number of data points, N is the data points, y actual (n) time domain value of actual output signal for each data point, y expected (n) time domain values of the desired output signal for each data point, y actual （n）-y expected (n) is the difference between the time domain values of the actual output signal and the desired output signal for the same data point,
the weight of the adaptive filter is adjusted according to the error and the input signal, and the adjustment formula of the weight is as follows: w (n+1) =w (n) +μ×e (n) ×e (n), where w (n+1) is an adjusted weight, w (n) is a current weight, μ is a step size factor, e (n) is an error, x (n) is an input signal, filtering is performed with an adjusted adaptive filter,
Repeating the steps until the error gradually converges to be smaller than or equal to a preset error threshold value, setting the error as a final error, and adjusting the weight of the adaptive filter according to the final error to obtain a metal interference filter;
collecting real-time microwave radar reflected signals, inputting the real-time microwave radar reflected signals as input signals into a metal interference filter, and performing filtering processing on the microwave radar reflected signals through convolution operation to obtain target signals for filtering the metal interference signals;
extracting features from the target signal, calculating correlation coefficients between all the extracted features and fire, screening out features with high correlation degree, and setting the features as fire correlation features;
performing fuzzy logic analysis on the fire related characteristics to judge whether fire exists or not;
if the fire exists, analyzing the change trend of the fire characteristics in time according to the fire related characteristics and through time sequence analysis to obtain the persistence of the fire.
2. The method for detecting fire in a coal mine based on a power line network according to claim 1, wherein the step of performing fuzzy logic analysis on fire related features to determine whether fire exists comprises:
Taking each fire related characteristic as a fuzzy variable, and determining the variable range of the fire related characteristic to be fuzzified;
dividing each fuzzy variable into a plurality of fuzzy sets according to the property and variable range of the fire related characteristics;
setting a membership function for each fuzzy set of fire related features, and substituting real-time feature values of the fire related features into corresponding membership functions to calculate membership degrees of the fire related features in each fuzzy set;
combining membership degrees of the real-time characteristic values in different fuzzy sets to form a fuzzy output set of fire related characteristics;
calculating a correlation coefficient matrix among all fire related features, and calculating the weight of each fire related feature according to the correlation coefficient matrix among all fire related features;
and accumulating the products of the weights of the fire related features and the maximum membership degree in the fuzzy output set, and comparing the accumulated value with a fire threshold value to make decision judgment on whether fire exists in the coal mine.
3. The method for detecting fire in coal mine based on power line network as claimed in claim 2, wherein the step of calculating the correlation coefficient matrix between all fire correlation features and calculating the weight of each fire correlation feature according to the correlation coefficient matrix between all fire correlation features comprises:
Calculating pearson correlation coefficients between fire related features to obtain an m x m correlation coefficient matrix, wherein m is the number of fire related features;
calculating the global average value of each fire related feature, subtracting the global average value of the corresponding feature from the real-time feature value of each fire related feature to obtain data after centralized processing and forming a data matrix;
covariance calculation is carried out on the data matrix to obtain covariance among relevant characteristics of each fire condition and form a covariance matrix;
performing eigenvalue decomposition on the covariance matrix to obtain eigenvalues and corresponding eigenvectors;
sorting the eigenvalues obtained by decomposing the eigenvalues according to the order from big to small, and sorting the eigenvectors according to the eigenvalue sorting;
selecting feature vectors corresponding to the top k feature values which are ranked at the front as main components;
taking the variance interpretation proportion of each main component as the weight of the corresponding fire related characteristics, wherein the variance interpretation proportion of the main component is the sum of the characteristic value corresponding to the main component divided by the total characteristic value;
the weights of the fire related features were normalized so that the sum of the weights was 1.
4. The method for detecting fire in a coal mine based on a power line network according to claim 1, wherein the step of analyzing the trend of the fire characteristic over time according to the fire related characteristic and through time sequence analysis to obtain the persistence of the fire comprises:
Respectively acquiring time sequence data of each fire related characteristic, wherein the time sequence data comprises time points and corresponding characteristic values;
carrying out centralization treatment on the time series data, namely subtracting the initial time from the time point;
fitting each polynomial function and time sequence data through a least square method to obtain each fitted curve, wherein each polynomial function is a function of gradually increasing the order from a simple first-order polynomial function;
calculating residual errors of each fitting curve and actual data points to obtain fitting errors, screening out a fitting curve with the smallest fitting error, and determining the fitting curve as the best fitting curve of relevant characteristics of corresponding fire;
determining the slope of a trend line of the relevant characteristic of the corresponding fire according to the coefficient of a polynomial function corresponding to the best fit curve of the relevant characteristic of the corresponding fire;
and determining the curvature of the trend line of the relevant characteristic of the corresponding fire according to the second derivative of the polynomial function corresponding to the best fit curve of the relevant characteristic of the corresponding fire.
5. The method for detecting fire conditions in a coal mine based on a power line network according to claim 1, wherein the step of inputting the real-time microwave radar reflection signal as an input signal into a metal interference filter and filtering the microwave radar reflection signal by convolution operation to obtain a target signal for filtering the metal interference signal comprises the steps of:
Converting the metal interference filter into a frequency domain, and multiplying the metal interference filter with an input signal in the frequency domain, wherein the calculation formula is as follows:
X(f)=FFT(x(t))，
Y(f)=X(f)*H(f)，
wherein X (f) is the frequency domain representation of the input signal, FFT (X (t)) is the fourier transform of the input signal, Y (f) is the filtered frequency domain signal, and H (f) is the frequency domain representation of the filter;
the result of the multiplication is inverse transformed to convert the input signal back to the time domain, which represents Y (t) =ifft (Y (f)), where IFFT (Y (f)) is the inverse fourier transform of the filtered frequency domain signal.
6. A coal mine fire detection system based on a power line network, characterized in that a microwave radar is installed at a wellhead or a loading and unloading area of a coal mine, comprising:
and a filter construction module: the method is used for acquiring historical microwave radar reflected signals and constructing a metal interference filter according to the characteristics of the historical microwave radar reflected signals and metal interference signals, and specifically comprises the following steps:
preprocessing the historical microwave radar reflected signal, including removing direct current component and normalizing,
converting the preprocessed microwave radar reflected signal from the time domain to the frequency domain through Fourier transformation, calculating correlation coefficients between each frequency component and other frequency components in the frequency spectrum,
Comparing and analyzing the correlation coefficient between each frequency component and other frequency components with a preset correlation coefficient threshold, if the correlation coefficient exceeds the preset correlation coefficient threshold, judging that the corresponding frequency component is related to the metal interference signal, setting the area corresponding to the frequency component as a metal interference area,
analyzing the signal characteristics of the metal interference region to determine the frequency range and energy distribution to be filtered or reduced, and constructing a desired spectrogram or magnitude spectrum based on the analysis result,
converts the desired spectrogram or amplitude spectrum into a time domain signal to obtain a time domain waveform of the desired output signal,
the historical microwave radar reflected signal is input into the adaptive filter as an input signal to obtain an actual output signal, wherein the weight of the adaptive filter is initialized when the microwave radar reflected signal is input for the first time,
calculating the difference between the actual output signal and the desired output signal to obtain an error, e (n) =Σ { [ y ] actual （n）-y expected （n）] 2 }/N, whereN is the number of data points, N represents each data point, y actual (n) time domain value of actual output signal for each data point, y expected (n) time domain values of the desired output signal for each data point, y actual （n）-y expected (n) is the difference between the time domain values of the actual output signal and the desired output signal for the same data point,
the weight of the adaptive filter is adjusted according to the error and the input signal, and the adjustment formula of the weight is as follows: w (n+1) =w (n) +μ×e (n) ×e (n), where w (n+1) is an adjusted weight, w (n) is a current weight, μ is a step size factor, e (n) is an error, x (n) is an input signal, filtering is performed with an adjusted adaptive filter,
repeating until the error gradually converges to be smaller than or equal to a preset error threshold value, setting the error as a final error, and adjusting the weight of the adaptive filter according to the final error to obtain a metal interference filter;
and a filtering module: the method comprises the steps of collecting real-time microwave radar reflected signals, inputting the real-time microwave radar reflected signals serving as input signals into a metal interference filter, and performing filtering processing on the microwave radar reflected signals through convolution operation to obtain target signals for filtering the metal interference signals;
and the feature extraction module is used for: the method comprises the steps of extracting features from a target signal, calculating correlation coefficients between all the extracted features and fire, screening out features with high correlation degree, and setting the features as fire correlation features;
The fire judgment module is used for: the fuzzy logic analysis is used for carrying out fuzzy logic analysis on the fire related characteristics so as to judge whether fire exists or not;
fire analysis module: and if the fire exists, analyzing the time change trend of the fire characteristics according to the fire related characteristics and through time sequence analysis to obtain the persistence of the fire.
7. The system of claim 6, further comprising a 380 volt AC power line configured to form the power line network for providing power to individual mining power equipment of the coal mine.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310968822.8A CN116682224B (en) | 2023-08-03 | 2023-08-03 | Coal mine fire detection method and system based on power line network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310968822.8A CN116682224B (en) | 2023-08-03 | 2023-08-03 | Coal mine fire detection method and system based on power line network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116682224A CN116682224A (en) | 2023-09-01 |
| CN116682224B true CN116682224B (en) | 2023-10-10 |
Family
ID=87781341
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310968822.8A Active CN116682224B (en) | 2023-08-03 | 2023-08-03 | Coal mine fire detection method and system based on power line network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116682224B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5235339A (en) * | 1992-11-13 | 1993-08-10 | The United States Of America As Represented By The Secretary Of The Navy | Radar target discrimination systems using artificial neural network topology |
| KR20160032526A (en) * | 2014-09-16 | 2016-03-24 | 김진규 | Fire detector |
| KR20190061425A (en) * | 2017-11-28 | 2019-06-05 | 썬메디텍(주) | Intelligent fire prevention diagnosis system and method |
| CN111896944A (en) * | 2020-07-29 | 2020-11-06 | 西安京维智网科技有限公司 | Intelligent fire safety monitoring system and monitoring method based on millimeter wave radar |
| KR102391509B1 (en) * | 2021-04-27 | 2022-04-27 | 주식회사 대성티엠씨 | Fire detection system in road tunnel |
| CN114666426A (en) * | 2022-02-11 | 2022-06-24 | 深圳市震有智联科技有限公司 | Data analysis method, device and storage medium |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7397363B2 (en) * | 1993-06-08 | 2008-07-08 | Raymond Anthony Joao | Control and/or monitoring apparatus and method |
-
2023
- 2023-08-03 CN CN202310968822.8A patent/CN116682224B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5235339A (en) * | 1992-11-13 | 1993-08-10 | The United States Of America As Represented By The Secretary Of The Navy | Radar target discrimination systems using artificial neural network topology |
| KR20160032526A (en) * | 2014-09-16 | 2016-03-24 | 김진규 | Fire detector |
| KR20190061425A (en) * | 2017-11-28 | 2019-06-05 | 썬메디텍(주) | Intelligent fire prevention diagnosis system and method |
| CN111896944A (en) * | 2020-07-29 | 2020-11-06 | 西安京维智网科技有限公司 | Intelligent fire safety monitoring system and monitoring method based on millimeter wave radar |
| KR102391509B1 (en) * | 2021-04-27 | 2022-04-27 | 주식회사 대성티엠씨 | Fire detection system in road tunnel |
| CN114666426A (en) * | 2022-02-11 | 2022-06-24 | 深圳市震有智联科技有限公司 | Data analysis method, device and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116682224A (en) | 2023-09-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103033567B (en) | Pipeline defect signal identification method based on guided wave | |
| CN102841131B (en) | Intelligent steel cord conveyer belt defect identification method and intelligent steel cord conveyer belt defect identification system | |
| KR102014751B1 (en) | Apparatus and method for detecting electric load using modified mfcc | |
| CN111059066B (en) | Centrifugal pump cavitation state discrimination method based on autocorrelation spectrum and balanced square envelope spectrum | |
| CN110702966B (en) | Fault arc detection method, device and system based on probabilistic neural network | |
| EP2857860A1 (en) | Signal processing device | |
| CN105447464A (en) | Electric energy quality disturbance recognition and classification method based on PSO | |
| CN112380992B (en) | Method and device for evaluating and optimizing accuracy of monitoring data in machining process | |
| CN112207631B (en) | Method for generating tool detection model, method, system, device and medium for detecting tool detection model | |
| CN115600088A (en) | Distribution transformer fault diagnosis method based on vibration signals | |
| CN113567969A (en) | Illegal sand dredger automatic monitoring method and system based on underwater acoustic signals | |
| CN116682224B (en) | Coal mine fire detection method and system based on power line network | |
| Matz et al. | Automated power quality monitoring system for on-line detection and classification of disturbances | |
| CN112751633B (en) | Broadband spectrum detection method based on multi-scale window sliding | |
| CN111597873B (en) | Support vector data description-based ultrashort wave threat signal sensing method | |
| CN111339986B (en) | Device frequency law mining method and system based on time domain/frequency domain analysis | |
| CN117330906A (en) | Equipment arc fault detection method, device, equipment and storage medium | |
| CN116773961A (en) | Transmission line corrosion detection method based on vibration signal high-frequency characteristic analysis | |
| Abdoos et al. | A hybrid method based on time frequency analysis and artificial intelligence for classification of power quality events | |
| CN114638265A (en) | Milling flutter discrimination method based on signal convolution neural network | |
| CN113065498A (en) | Abnormal data detection method based on improved EMD and neural network model | |
| Viet et al. | A method for monitoring voltage disturbances based on discrete wavelet transform and adaptive linear neural network | |
| CN112731388A (en) | Target detection method based on effective scattering point energy accumulation | |
| CN117540328B (en) | Noise processing method in high-precision measurement process of coal mine noise | |
| Solis-Munoz et al. | A Novel Methodology for Power Quality Disturbances Detection and Classification in Industrial Facilities |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |