Gathering pipeline liquid accumulation state recognition method and system based on time-frequency feature screening

Abstract

The present application relates to pipeline liquid accumulation detection technical field, especially in a kind of based on time-frequency feature screening's gathering and transportation pipeline liquid accumulation state identification method and system, comprising the following steps: S1, obtains the modeling data of several liquid accumulation states of gathering and transportation pipeline, calibrates the state label of the modeling data;S2, extract time-frequency feature and mark index bit number;S3, use Lasso regression method to screen time-frequency feature and calculate the characteristic coefficient of time-frequency feature, retain the time-frequency feature of characteristic coefficient not 0 and record corresponding index bit number as sample index bit number;S4, construct random forest model;S5, obtain the real-time data of gathering and transportation pipeline, extract the time-frequency feature of real-time data, according to sample index bit number extraction time-frequency feature, input random forest model to the time-frequency feature extracted, to obtain the liquid accumulation state of gathering and transportation pipeline.The present application improves the recognition accuracy of liquid accumulation state.

Description

Technical Field

[0001] The present invention relates to the technical field of pipeline fluid accumulation detection, and in particular to a method and system for identifying the fluid accumulation state of a gathering and transportation pipeline based on time-frequency feature screening. Background Art

[0002] Existing pipeline liquid accumulation identification methods primarily focus on simulation-based research into the formation patterns of liquid accumulation and empirically based monitoring and early warning methods. These methods lack data-driven online identification research and in-depth analysis of sensor data from gathering and transportation pipelines, thus failing to capture sufficient information.

[0003] The data-driven recognition method involves three key steps: data collection, feature extraction, and classification model building. During the online recognition process, by inputting real-time data into the model, the prediction results can be quickly obtained. In terms of feature extraction, since limited sensors are needed in the gathering and transportation pipeline to achieve accurate identification of the effusion state, the features extracted solely by relying on time domain signals or frequency domain signals are often insufficient and incomplete in terms of information. Therefore, the time-frequency feature fusion method can obtain more comprehensive information. Although the existing technology can extract high-dimensional time-frequency features, it has not been effectively screened, which results in a large amount of useless information, thereby reducing the accuracy of classification and recognition. Summary of the Invention

[0004] The present invention aims to solve at least one of the technical problems existing in the related art. To this end, the present invention provides a method and system for identifying the accumulation state of a gathering and transportation pipeline based on time-frequency feature screening. This method solves the technical problem of low accuracy in identifying accumulation states in existing data processing methods. It obtains more comprehensive time-frequency features, retains key information in the time-frequency features that is highly relevant to the label, and improves the accuracy of identifying accumulation states.

[0005] The present invention provides a method for identifying the liquid accumulation state of a gathering and transportation pipeline based on time-frequency feature screening, comprising the following steps:

[0006] S1, obtaining modeling data of several liquid accumulation states of a gathering and transportation pipeline, and calibrating state labels of the modeling data;

[0007] S2, extracting the time-frequency features of the modeling data and marking the index digit of each time-frequency feature;

[0008] S3, use the Lasso regression method to screen the time-frequency features and calculate the characteristic coefficients of the time-frequency features, retain the time-frequency features whose characteristic coefficients are not 0 and record the corresponding index digits as the sample index digits;

[0009] S4, using the filtered time-frequency features as modeling samples, and the state labels corresponding to the time-frequency features in the modeling samples as supervision to build a random forest model;

[0010] S5, acquiring real-time data of the gathering and transportation pipeline, and extracting time-frequency characteristics of the real-time data;

[0011] S6, extracting part of the time-frequency features of the real-time data according to the number of sample index bits, and inputting the extracted time-frequency features into the random forest model to obtain the liquid accumulation state of the gathering and transportation pipeline.

[0012] A further improvement of the method for identifying the accumulation state of a gathering and transportation pipeline based on time-frequency feature screening of the present invention is that step S1 includes the following steps:

[0013] S11: collecting detection signals of sensors of the gathering and transportation pipeline to construct a data set, and determining the liquid accumulation state corresponding to each data point in the data set;

[0014] S12: using a sliding window method to select some data points in the data set as modeling data and construct a two-dimensional matrix;

[0015] Set the modeling data by Sliding windows are composed, and the length of each sliding window is , then each sliding window signal is , each sliding window signal is combined to form modeling data, and the modeling data is a two-dimensional matrix , is the field of real numbers;

[0016] S13: calibrating the state label of the modeling data according to the state of the effusion, wherein the value of the state label is an integer and is in direct proportion to the height of the effusion level corresponding to the effusion state.

[0017] Modeling data Corresponding dimensional label vector It can be expressed as:

[0018]

[0019] in, Indicates the state label corresponding to the first sliding window signal, Indicates the state label corresponding to the second sliding window signal, Indicates the The state label corresponding to the sliding window signal.

[0020] A further improvement of the method for identifying the accumulation state of a gathering and transportation pipeline based on time-frequency feature screening of the present invention is that extracting the time-frequency features of the modeling data in step S2 includes the following steps:

[0021] S21, determining a time domain signal of each sliding window signal in the modeling data, and performing Fourier transform on the time domain signal to obtain a corresponding frequency domain signal;

[0022] S22, extracting time domain features from the time domain signal and extracting frequency domain features from the frequency domain signal;

[0023] S23, fusing the time domain features and the frequency domain features to obtain fused time-frequency features;

[0024] S24, normalizing the fused time-frequency features to obtain time-frequency features.

[0025] A further improvement of the method for identifying the accumulation state of a gathering and transportation pipeline based on time-frequency feature screening of the present invention is that the time domain feature is extracted in the following manner:

[0026]

[0027] in, Indicates the time domain signal at The value of the moment, Represents the total number of points of the time domain signal, represents the first characteristic of the time domain signal, represents the second characteristic of the time domain signal, represents the third characteristic of the time domain signal, represents the fourth characteristic of the time domain signal, represents the fifth characteristic of the time domain signal, represents the sixth characteristic of the time domain signal, represents the seventh characteristic of the time domain signal, represents the eighth characteristic of the time domain signal, represents the ninth characteristic of the time domain signal, represents the tenth characteristic of the time domain signal, represents the eleventh characteristic of the time domain signal, represents the twelfth characteristic of the time domain signal, represents the thirteenth characteristic of the time domain signal, represents the fourteenth characteristic of the time domain signal, Represents the fifteenth feature of the time domain signal.

[0028] A further improvement of the method for identifying the accumulation state of a gathering and transportation pipeline based on time-frequency feature screening of the present invention is that the frequency domain features are extracted in the following manner:

[0029]

[0030] in, Represents the total number of points of the frequency domain signal, Indicates the The amplitude or energy value of a frequency domain signal, Indicates the The frequency corresponding to the frequency domain signal, represents the first characteristic of the frequency domain signal, represents the second characteristic of the frequency domain signal, represents the third characteristic of the frequency domain signal, represents the fourth characteristic of the frequency domain signal, represents the fifth characteristic of the frequency domain signal, represents the sixth characteristic of the frequency domain signal, represents the seventh characteristic of the frequency domain signal, represents the eighth characteristic of the frequency domain signal, represents the ninth characteristic of the frequency domain signal, represents the tenth characteristic of the frequency domain signal, represents the eleventh characteristic of the frequency domain signal, represents the twelfth characteristic of the frequency domain signal, Represents the thirteenth feature of the frequency domain signal.

[0031] The present invention further improves the method for identifying the accumulation state of the gathering and transportation pipeline based on time-frequency feature screening, and integrates the time-frequency features into , the two-dimensional matrix composed of the fusion time-frequency features of several sliding window signals is ,

[0032] The standardization method is:

[0033]

[0034] in, Represents a two-dimensional matrix Middle Rank Column data, Indicates the The average value of the column data, Indicates the The standard deviation of the column data, express Standardized data;

[0035] After normalization processing, each sliding window signal obtains the normalized time-frequency feature: , the two-dimensional matrix composed of the standardized time-frequency features of each sliding window signal is .

[0036] A further improvement of the method for identifying the accumulation state of a gathering and transportation pipeline based on time-frequency feature screening of the present invention is that step S3 specifically includes:

[0037] The regression model is set using the Lasso regression method.

[0038] in, represents the characteristic coefficient vector, represents random error, Represents the characteristic coefficient corresponding to the first time-frequency feature of each sliding window signal in the modeling data, Represents the characteristic coefficient corresponding to the second time-frequency feature of each sliding window signal in the modeling data, Represents the characteristic coefficient corresponding to the 28th time-frequency feature of each sliding window signal in the modeling data;

[0039] The regression model is estimated in the least squares regularization is expressed as:

[0040]

[0041] in, represents the penalty coefficient, Indicates the The standardized features corresponding to the sliding window signal, The value range is , Indicates the The state label corresponding to the sliding window signal;

[0042] The characteristic coefficient vector is calculated by the least squares regularization estimation method ;

[0043] Record No. The characteristic coefficients are not 0, and the standardized time-frequency features are retained. The standardized time-frequency features corresponding to the feature coefficients are expressed as ; The matrix formed by the standardized time-frequency features after screening is , which is specifically expressed as:

[0044]

[0045] in, Represents the filtered and standardized time-frequency features corresponding to the first sliding window signal in the modeling data, Represents the filtered and standardized time-frequency features corresponding to the second sliding window signal in the modeling data, Indicates the first The filtered and normalized time-frequency features corresponding to the sliding window signal.

[0046] A further improvement of the method for identifying the accumulation state of a gathering and transportation pipeline based on time-frequency feature screening of the present invention is that step S5 includes extracting the time-frequency features of real-time data. ;

[0047] Step S6 includes:

[0048] S61, normalizing the time-frequency features of the real-time data to obtain normalized time-frequency features. The normalization method is:

[0049]

[0050] in, Indicates the real-time data Features, Represents the standardized Features

[0051] S62, retaining the normalized time-frequency features of the real-time data corresponding to the number of sample index bits, recorded as ;

[0052] S63, the filtered standardized time-frequency features are input into the constructed random forest model to obtain the liquid accumulation status of the gathering and transportation pipeline.

[0053] The present invention also provides a system for identifying the accumulation state of a gathering and transportation pipeline based on time-frequency feature screening. The system is used to perform the above-mentioned identification method, and the system includes:

[0054] A time-frequency feature extraction unit, a feature screening unit, a random forest model construction unit, and a state recognition unit, wherein the time-frequency feature extraction unit and the feature screening unit are connected, the time-frequency feature extraction unit and the state recognition unit are connected, the feature screening unit, the random forest model construction unit, and the state recognition unit are connected, and the random forest model construction unit is connected to the state recognition unit;

[0055] A time-frequency feature extraction unit, used to extract time-frequency features and perform standardization processing;

[0056] A feature screening unit is used to screen time-frequency features using the Lasso regression method and calculate the characteristic coefficients of the time-frequency features, retain the time-frequency features whose characteristic coefficients are not 0 and record the corresponding index digits as the sample index digits;

[0057] Random forest model building unit, used to build a random forest model;

[0058] The state recognition unit is used to extract the time-frequency features of real-time data and perform normalized processing.

[0059] The present invention effectively realizes the identification of the accumulation state of the gathering and transportation pipeline by combining time-frequency feature extraction with Lasso regression feature screening and applying the random forest model. Through formula calculation, the present invention obtains comprehensive time-frequency features, including 15 time-domain features and 13 frequency-domain features, and fully mines the information in the signal. Lasso regression is used for feature screening to retain the key information of time-frequency features that are highly correlated with the label, thereby effectively improving the recognition accuracy of the model. The present invention adopts random forest as a classification model to enhance the performance and robustness of the gathering and transportation pipeline status identification. In addition, the present invention is based on a data-driven recognition method, which breaks through the limitations of traditional methods and has the advantages of automation and simplified workflow. It can be applied to actual industrial environments to provide real-time prediction results of the gathering and transportation pipeline status to assist staff in making judgments.

[0060] Additional aspects and advantages of the present invention will be set forth in part in the description which follows and, in part, will be obvious from the description which follows, or may be learned by practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0062] Figure 1 This is a flow chart of a method and system for identifying the accumulation state of liquid in a gathering and transportation pipeline based on time-frequency feature screening provided by an embodiment of the present invention.

[0063] Figure 2 This is a confusion matrix for identifying the liquid accumulation state of a gathering and transportation pipeline under the conditions of a safe liquid level and a warning liquid level provided by an embodiment of the present invention.

[0064] Figure 3 This is a confusion matrix for identifying the liquid accumulation state of a gathering and transportation pipeline under the conditions of safe liquid level, low liquid level, and high liquid level provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0065] To make the purpose, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the present invention. Obviously, the embodiments described are 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 ordinary technicians in this field without making creative work are within the scope of protection of the present invention. The following embodiments are used to illustrate the present invention, but are not used to limit the scope of the present invention.

[0066] The following combination Figure 1 The present invention describes a method for identifying the accumulation state of a gathering and transportation pipeline based on time-frequency feature screening, comprising the following steps:

[0067] S1, obtaining modeling data of several liquid accumulation states of a gathering and transportation pipeline, and calibrating state labels of the modeling data;

[0068] S2, extracting the time-frequency features of the modeling data and marking the index digit of each time-frequency feature;

[0069] S3, use the Lasso regression method to screen the time-frequency features and calculate the characteristic coefficients of the time-frequency features, retain the time-frequency features whose characteristic coefficients are not 0 and record the corresponding index digits as the sample index digits;

[0070] S4, using the filtered time-frequency features as modeling samples, and the state labels corresponding to the time-frequency features in the modeling samples as supervision to build a random forest model;

[0071] S5, acquiring real-time data of the gathering and transportation pipeline, and extracting time-frequency characteristics of the real-time data;

[0072] S6, extracting part of the time-frequency features of the real-time data according to the number of sample index bits, and inputting the extracted time-frequency features into the random forest model to obtain the liquid accumulation state of the gathering and transportation pipeline.

[0073] Through precise time-frequency feature extraction and Lasso regression screening, this recognition method can efficiently identify key features related to the warning liquid level, thereby constructing a high-precision random forest model, improving the accuracy of liquid accumulation status identification, reducing the risk of false alarms and missed alarms, and providing strong guarantees for the safe operation of gathering and transportation pipelines.

[0074] This identification method enables real-time monitoring and early warning of liquid accumulation in gathering and transportation pipelines. By acquiring real-time data and inputting it into a constructed random forest model, it can quickly determine whether the accumulation is approaching or reaching the warning level, thereby triggering an alarm mechanism and notifying relevant personnel to take appropriate measures. This real-time monitoring and early warning capability helps to promptly identify and address potential safety hazards and prevent accidents.

[0075] This method also has the potential for continuous optimization and improvement. By continuously accumulating real-time data and model outputs, a rich historical dataset can be formed for subsequent model training and optimization. This not only improves the accuracy and reliability of the model's identification of effusion states, but also allows it to adapt to different operating conditions and changes in effusion states, enhancing the method's versatility and adaptability.

[0076] In a preferred embodiment of the method for identifying the accumulation state of a gathering and transportation pipeline based on time-frequency feature screening of the present invention, step S1 includes the following steps:

[0077] S11: Collecting detection signals from sensors on the gathering and transportation pipeline to construct a data set, and determining the fluid accumulation state corresponding to each data point in the data set. By collecting detection signals from sensors on the gathering and transportation pipeline and constructing the data set, changes in the fluid accumulation state can be comprehensively and accurately recorded. This step provides a solid foundation for subsequent data analysis and model building, ensuring the accuracy and reliability of fluid accumulation state identification.

[0078] S12: using a sliding window method to select some data points in the data set as modeling data and construct a two-dimensional matrix;

[0079] Set the modeling data by Sliding windows are composed, and the length of each sliding window is , then each sliding window signal is , each sliding window signal is combined to form modeling data, and the modeling data is a two-dimensional matrix , is the field of real numbers;

[0080] A sliding window method was used to select a subset of data points from the dataset as modeling data and construct a two-dimensional matrix. This step effectively reduced the data dimension and computational complexity while retaining key information about the effusion state. This not only improved the efficiency of model construction but also enhanced the model's generalization capabilities, enabling it to better adapt to different operating conditions and changes in effusion state. The construction of the two-dimensional matrix also facilitated subsequent time-frequency feature extraction and feature screening, further improving the accuracy and robustness of effusion state identification.

[0081] S13: calibrating the state label of the modeling data according to the state of the effusion, wherein the value of the state label is an integer and is in direct proportion to the height of the effusion level corresponding to the effusion state.

[0082] Modeling data Corresponding dimensional label vector It can be expressed as:

[0083]

[0084] in, Indicates the state label corresponding to the first sliding window signal, Indicates the state label corresponding to the second sliding window signal, Indicates the The state label corresponding to the sliding window signal.

[0085] Preferably, the plurality of liquid accumulation states include a safe liquid level and a warning liquid level, the modeling data corresponding to the safe liquid level is calibrated as a first label, and the modeling data corresponding to the warning liquid level is calibrated as a second label.

[0086] Furthermore, extracting the time-frequency features of the modeling data in step S2 includes the following steps: S21, determining a time domain signal of each sliding window signal in the modeling data, and performing Fourier transform on the time domain signal to obtain a corresponding frequency domain signal;

[0087] S22, extract time domain features from time domain signals and frequency domain features from frequency domain signals; time domain features such as mean, variance, and peak value reflect the changing trend of the signal over time, while frequency domain features such as main frequency and bandwidth reveal the distribution pattern of the signal in the frequency domain;

[0088] S23, fusing the time domain features and the frequency domain features to obtain fused time-frequency features; the fusion process ensures the complementarity of the time domain and frequency domain information, providing a more comprehensive feature description for subsequent effusion status identification;

[0089] S24, normalizes the fused time-frequency features to obtain time-frequency features; the normalization process eliminates the dimensional differences between different features, thereby improving the comparability of the features and the stability of the model.

[0090] Furthermore, the time domain features are extracted as follows:

[0091]

[0092] in, Indicates the time domain signal at The value of the moment, Represents the total number of points of the time domain signal, Represents the first characteristic of the time domain signal, reflecting the DC component of the time domain signal, Represents the second characteristic of the time domain signal, characterizing the average amplitude of the time domain signal. It represents the third characteristic of the time domain signal, reflecting the average level of the time domain signal energy. It represents the fourth characteristic of the time domain signal, which measures the degree of fluctuation of the time domain signal from the average value. It represents the fifth characteristic of the time domain signal, reflecting a special weighted average characteristic of the time domain signal amplitude. It represents the sixth characteristic of the time domain signal and the effective amplitude of the time domain signal. Represents the seventh characteristic of the time domain signal, reflecting the maximum amplitude, Represents the eighth characteristic of the time domain signal, the maximum value of the time domain signal, Represents the ninth characteristic of the time domain signal, the minimum value of the time domain signal, It represents the tenth characteristic of the time domain signal, reflecting the waveform characteristics of the time domain signal. Represents the eleventh characteristic of the time domain signal, measuring the symmetry of the time domain signal, The twelfth characteristic of the time domain signal reflects the relative relationship between the AC component and the DC component. The thirteenth characteristic of the time domain signal is the ratio of the effective value to the average absolute value. The skewness of the time domain signal is the fourteenth characteristic of the time domain signal, which measures the symmetry of the probability distribution. The fifteenth characteristic of the time domain signal, the kurtosis of the time domain signal, measures the steepness of the amplitude distribution.

[0093] Furthermore, the frequency domain features are extracted as follows:

[0094]

[0095] in, Represents the total number of points of the frequency domain signal, Indicates the The amplitude or energy value of a frequency domain signal, Indicates the The frequency corresponding to the frequency domain signal, Represents the first characteristic of the frequency domain signal, reflecting its DC component, Represents the second characteristic of the frequency domain signal, measuring its deviation from the mean The degree of fluctuation, Represents the third characteristic of the frequency domain signal, measuring the symmetry of the probability distribution, Represents the fourth characteristic of the frequency domain signal, which measures the steepness of the amplitude distribution. Represents the fifth characteristic of the frequency domain signal, reflecting The average characteristics of Represents the sixth characteristic of the frequency domain signal, reflecting Effective amplitude, Represents the seventh characteristic of the frequency domain signal, measuring Deviation Fluctuations, Represents the eighth feature of the frequency domain signal, the normalized amplitude feature measure, Represents the ninth characteristic of the frequency domain signal, the normalized metric, Represents the tenth characteristic of the frequency domain signal, reflecting The ratio of the fluctuation to its effective amplitude, Represents the eleventh characteristic of the frequency domain signal, measuring The symmetry of the distribution, Represents the twelfth characteristic of the frequency domain signal, measuring The steepness of the distribution, Represents the thirteenth characteristic of frequency domain signals, a special weighted metric.

[0096] Furthermore, the fusion time-frequency features are , the two-dimensional matrix composed of the fusion time-frequency features of several sliding window signals is ,Right now

[0097] in, The meaning of the representation is the real number field The previous one A matrix with 28 rows and 28 columns;

[0098] The standardization method is:

[0099]

[0100] in, Represents a two-dimensional matrix Middle Rank Column data, Indicates the The average value of the column data, Indicates the The standard deviation of the column data, express Standardized data;

[0101] After normalization processing, each sliding window signal obtains the normalized time-frequency feature: , the two-dimensional matrix composed of the standardized time-frequency features of each sliding window signal is .

[0102] Preferably, step S3 specifically includes:

[0103] The regression model is set using the Lasso regression method.

[0104] in, represents the characteristic coefficient vector, represents random error, Represents the characteristic coefficient corresponding to the first time-frequency feature of each sliding window signal in the modeling data, Represents the characteristic coefficient corresponding to the second time-frequency feature of each sliding window signal in the modeling data, Represents the characteristic coefficient corresponding to the 28th time-frequency feature of each sliding window signal in the modeling data;

[0105] The regression model is estimated in the least squares regularization is expressed as:

[0106]

[0107] in, represents the penalty coefficient, Indicates the The standardized features corresponding to the sliding window signal, The value range is , Indicates the The state label corresponding to the sliding window signal;

[0108] The characteristic coefficient vector is calculated by the least squares regularization estimation method ;

[0109] Record No. characteristic coefficient is not 0, then we know that there is The characteristic coefficients of the standardized time-frequency features are not 0, and the standardized time-frequency features with The standardized time-frequency features corresponding to the feature coefficients are expressed as ; The matrix formed by the standardized time-frequency features after screening is , which is specifically expressed as:

[0110]

[0111] in, Represents the filtered and standardized time-frequency features corresponding to the first sliding window signal in the modeling data, Represents the filtered and standardized time-frequency features corresponding to the second sliding window signal in the modeling data, Indicates the first The filtered and normalized time-frequency features corresponding to the sliding window signal.

[0112] Specifically, step S5 includes extracting the time-frequency features of real-time data ;

[0113] Step S6 includes:

[0114] S61, normalizing the time-frequency features of the real-time data to obtain normalized time-frequency features. The normalization method is:

[0115]

[0116] in, Indicates the real-time data Features, Represents the standardized Features

[0117] S62, retaining the normalized time-frequency features of the real-time data corresponding to the number of sample index bits, recorded as ;

[0118] S63, the filtered standardized time-frequency features are input into the constructed random forest model to obtain the liquid accumulation status of the gathering and transportation pipeline.

[0119] The present invention also provides a system for identifying the accumulation state of a gathering and transportation pipeline based on time-frequency feature screening. The system is used to perform the above-mentioned identification method, and the system includes:

[0120] A time-frequency feature extraction unit, a feature screening unit, a random forest model construction unit, and a state recognition unit, wherein the time-frequency feature extraction unit and the feature screening unit are connected, the time-frequency feature extraction unit and the state recognition unit are connected, the feature screening unit, the random forest model construction unit, and the state recognition unit are connected, and the random forest model construction unit is connected to the state recognition unit;

[0121] A time-frequency feature extraction unit, used to extract time-frequency features and perform standardization processing;

[0122] A feature screening unit is used to screen time-frequency features using the Lasso regression method and calculate the characteristic coefficients of the time-frequency features, retain the time-frequency features whose characteristic coefficients are not 0 and record the corresponding index digits as the sample index digits;

[0123] Random forest model building unit, used to build a random forest model;

[0124] The state recognition unit is used to extract the time-frequency features of real-time data and perform normalized processing.

[0125] The present invention effectively realizes the identification of the accumulation state of the gathering and transportation pipeline by combining time-frequency feature extraction with Lasso regression feature screening and applying the random forest model. Through formula calculation, the present invention obtains comprehensive time-frequency features, including 15 time-domain features and 13 frequency-domain features, and fully mines the information in the signal. Lasso regression is used for feature screening to retain the key information of time-frequency features that are highly correlated with the label, thereby effectively improving the recognition accuracy of the model. The present invention adopts random forest as a classification model to enhance the performance and robustness of the gathering and transportation pipeline status identification. In addition, the present invention is based on a data-driven recognition method, which breaks through the limitations of traditional methods and has the advantages of automation and simplified workflow. It can be applied to actual industrial environments to provide real-time prediction results of the gathering and transportation pipeline status to assist staff in making judgments.

[0126] In a specific embodiment, an ultrasonic sensor is used to obtain a one-dimensional signal of the ultrasonic sensor as an initial sensor detection signal. If the types of sensors in the gathering and transportation pipeline are expanded to multiple types, the measurement variable will be expanded to a higher dimension.

[0127] The state division of the effusion state is divided according to the actual situation. Its label is an integer and increases as the effusion level of each state increases. In a specific embodiment, the effusion state is divided into two categories: safe liquid level and warning liquid level. The safe liquid level label is 0 and the warning liquid level label is 1. Then the first label represents the safe liquid level label, and the second label represents the warning liquid level label. In another specific implementation case, the effusion state is divided into three categories: safe liquid level, low liquid level and high liquid level. The safe liquid level label is 0, the low liquid level label is 1, and the high liquid level label is 2. Then the first label represents the safe liquid level label, and the second label represents the low liquid level label and the high liquid level label.

[0128] In order to verify the effectiveness of the identification method of the present invention, the following experiments were conducted.

[0129] The inner diameter of the gathering pipeline is 110 mm. The liquid level in the pipeline is artificially set to simulate the amount of liquid accumulation in actual industry. When the labels used are 0 for the safety liquid level and 1 for the warning liquid level (hereinafter referred to as the two-category case), samples corresponding to a liquid accumulation level of 0 mm to 10 mm in the pipeline are considered to be safe liquid levels, and samples corresponding to a liquid accumulation level of 15 mm to 100 mm are considered to be warning liquid levels. When the labels used are 0 for the safety liquid level, 1 for the low liquid level, and 2 for the high liquid level (hereinafter referred to as the three-category case), samples corresponding to a liquid accumulation level of 0 mm to 10 mm in the pipeline are considered to be safe liquid levels, samples corresponding to a liquid accumulation level of 15 mm to 50 mm are considered to be low liquid levels, and samples corresponding to a liquid accumulation level of 55 mm to 100 mm are considered to be high liquid levels.

[0130] In the case of binary classification, the modeling samples contain 2 states, each state includes 80 samples, a total of 160 samples, including 28 features before screening, that is, the standardized time-frequency feature matrix is ; After screening, 3 features are retained, that is, the standardized time-frequency feature matrix after screening is In the case of three classifications, the modeling samples contain three states, each state includes 80 samples, a total of 240 samples, including 28 features before screening, that is, the standardized time-frequency feature matrix is , after screening, 5 features are retained, that is, the standardized time-frequency feature matrix after screening is .

[0131] The confusion matrix for the binary classification case is as follows Figure 2 As shown, the confusion matrix for the three-class case is as follows Figure 3As shown, in the two-classification case, 20 safety liquid level samples and 20 warning liquid level samples are included; in the three-classification case, 20 safety liquid level samples, 20 low liquid level samples and 20 high liquid level samples are included. The accuracy of the gathering and transportation pipeline status recognition in the two-classification and three-classification cases can reach 100% and 90% respectively, indicating the effectiveness of the recognition method of the present invention.

[0132] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for identifying the accumulation state of a gathering and transportation pipeline based on time-frequency feature screening, characterized in that: The steps include: S1, obtaining modeling data of several liquid accumulation states of a gathering and transportation pipeline, and calibrating state labels of the modeling data; S2, extracting the time-frequency features of the modeling data and marking the index digit of each time-frequency feature; S3, use the Lasso regression method to screen the time-frequency features and calculate the characteristic coefficients of the time-frequency features, retain the time-frequency features whose characteristic coefficients are not 0 and record the corresponding index digits as the sample index digits; S4, using the filtered time-frequency features as modeling samples, and the state labels corresponding to the time-frequency features in the modeling samples as supervision to build a random forest model; S5, acquiring real-time data of the gathering and transportation pipeline, and extracting time-frequency characteristics of the real-time data; S6, extracting part of the time-frequency features of the real-time data according to the number of sample index bits, and inputting the extracted time-frequency features into the random forest model to obtain the liquid accumulation state of the gathering and transportation pipeline.

2. The method for identifying the accumulation state of a gathering and transportation pipeline based on time-frequency feature screening according to claim 1 is characterized in that: Step S1 includes the following steps: S11: collecting detection signals of sensors of the gathering and transportation pipeline to construct a data set, and determining the liquid accumulation state corresponding to each data point in the data set; S12: using a sliding window method to select some data points in the data set as modeling data and construct a two-dimensional matrix; Set the modeling data by Sliding windows are composed, and the length of each sliding window is , then each sliding window signal is , each sliding window signal is combined to form modeling data, and the modeling data is a two-dimensional matrix , is the field of real numbers; S13: calibrating the state label of the modeling data according to the state of the effusion, wherein the value of the state label is an integer and is in direct proportion to the height of the effusion level corresponding to the effusion state. Modeling data Corresponding dimensional label vector It can be expressed as: in, Indicates the state label corresponding to the first sliding window signal, Indicates the state label corresponding to the second sliding window signal, Indicates the The state label corresponding to the sliding window signal.

3. The method for identifying the accumulation state of a gathering and transportation pipeline based on time-frequency feature screening according to claim 2 is characterized in that: Extracting the time-frequency features of the modeling data in step S2 includes the following steps: S21, determining a time domain signal of each sliding window signal in the modeling data, and performing Fourier transform on the time domain signal to obtain a corresponding frequency domain signal; S22, extracting time domain features from the time domain signal and extracting frequency domain features from the frequency domain signal; S23, fusing the time domain features and the frequency domain features to obtain fused time-frequency features; S24, normalizing the fused time-frequency features to obtain time-frequency features.

4. The method for identifying the accumulation state of a gathering and transportation pipeline based on time-frequency feature screening according to claim 3 is characterized in that: The time domain features are extracted as follows: in, Indicates the time domain signal at The value of the moment, Represents the total number of points of the time domain signal, represents the first characteristic of the time domain signal, represents the second characteristic of the time domain signal, represents the third characteristic of the time domain signal, represents the fourth characteristic of the time domain signal, represents the fifth characteristic of the time domain signal, represents the sixth characteristic of the time domain signal, represents the seventh characteristic of the time domain signal, represents the eighth characteristic of the time domain signal, represents the ninth characteristic of the time domain signal, represents the tenth characteristic of the time domain signal, represents the eleventh characteristic of the time domain signal, represents the twelfth characteristic of the time domain signal, represents the thirteenth characteristic of the time domain signal, represents the fourteenth characteristic of the time domain signal, Represents the fifteenth feature of the time domain signal.

5. The method for identifying the accumulation state of a gathering and transportation pipeline based on time-frequency feature screening according to claim 4 is characterized in that: The frequency domain features are extracted as follows: in, Represents the total number of points of the frequency domain signal, Indicates the The amplitude or energy value of a frequency domain signal, Indicates the The frequency corresponding to the frequency domain signal, represents the first characteristic of the frequency domain signal, represents the second characteristic of the frequency domain signal, represents the third characteristic of the frequency domain signal, represents the fourth characteristic of the frequency domain signal, represents the fifth characteristic of the frequency domain signal, represents the sixth characteristic of the frequency domain signal, represents the seventh characteristic of the frequency domain signal, represents the eighth characteristic of the frequency domain signal, represents the ninth characteristic of the frequency domain signal, represents the tenth characteristic of the frequency domain signal, represents the eleventh characteristic of the frequency domain signal, represents the twelfth characteristic of the frequency domain signal, Represents the thirteenth feature of the frequency domain signal.

6. The method for identifying the accumulation state of a gathering and transportation pipeline based on time-frequency feature screening according to claim 5 is characterized in that: The fused time-frequency features are , the two-dimensional matrix composed of the fusion time-frequency features of several sliding window signals is , The standardization method is: in, Represents a two-dimensional matrix Middle Rank Column data, Indicates the The average value of the column data, Indicates the The standard deviation of the column data, express Standardized data; After normalization processing, each sliding window signal obtains the normalized time-frequency feature: , the two-dimensional matrix composed of the standardized time-frequency features of each sliding window signal is .

7. The method for identifying the accumulation state of a gathering and transportation pipeline based on time-frequency feature screening according to claim 6 is characterized in that: Step S3 specifically includes: The regression model is set using the Lasso regression method. in, represents the characteristic coefficient vector, represents random error, Represents the characteristic coefficient corresponding to the first time-frequency feature of each sliding window signal in the modeling data, Represents the characteristic coefficient corresponding to the second time-frequency feature of each sliding window signal in the modeling data, Represents the characteristic coefficient corresponding to the 28th time-frequency feature of each sliding window signal in the modeling data; The regression model is estimated in the least squares regularization is expressed as: in, represents the penalty coefficient, Indicates the The standardized features corresponding to the sliding window signal, The value range is , Indicates the The state label corresponding to the sliding window signal; The characteristic coefficient vector is calculated by the least squares regularization estimation method ; Record No. The characteristic coefficients are not 0, and the standardized time-frequency features are retained. The standardized time-frequency features corresponding to the feature coefficients are expressed as ; The matrix formed by the standardized time-frequency features after screening is , which is specifically expressed as: in, Represents the filtered and standardized time-frequency features corresponding to the first sliding window signal in the modeling data, Represents the filtered and standardized time-frequency features corresponding to the second sliding window signal in the modeling data, Indicates the first The filtered and normalized time-frequency features corresponding to the sliding window signal.

8. The method for identifying the accumulation state of a gathering and transportation pipeline based on time-frequency feature screening according to claim 7 is characterized in that: Step S5 includes extracting the time-frequency features of real-time data ; Step S6 includes: S61, normalizing the time-frequency features of the real-time data to obtain normalized time-frequency features. The normalization method is: in, Indicates the real-time data Features, Represents the normalized Features S62, retaining the normalized time-frequency features of the real-time data corresponding to the number of sample index bits, recorded as ； S63, the filtered standardized time-frequency features are input into the constructed random forest model to obtain the liquid accumulation status of the gathering and transportation pipeline.

9. A system for identifying the accumulation state of a gathering and transportation pipeline based on time-frequency feature screening, characterized in that: The identification system is used to perform the identification method according to any one of claims 1 to 8, and the identification system includes: A time-frequency feature extraction unit, a feature screening unit, a random forest model construction unit, and a state recognition unit, wherein the time-frequency feature extraction unit and the feature screening unit are connected, the time-frequency feature extraction unit and the state recognition unit are connected, the feature screening unit, the random forest model construction unit, and the state recognition unit are connected, and the random forest model construction unit is connected to the state recognition unit; A time-frequency feature extraction unit, used to extract time-frequency features and perform standardization processing; A feature screening unit is used to screen time-frequency features using the Lasso regression method and calculate the characteristic coefficients of the time-frequency features, retain the time-frequency features whose characteristic coefficients are not 0 and record the corresponding index digits as the sample index digits; Random forest model building unit, used to build a random forest model; The state recognition unit is used to extract the time-frequency features of real-time data and perform normalized processing.

Images