A nonlinear model detection method and system for electrochemical sensor

Abstract

The present invention relates to the field of electrochemical detection technology, specifically to a nonlinear model detection method and system for an electrochemical sensor. The method comprises: first, obtaining electrochemical data of a signal tower steel structure using an electrochemical sensor; second, using a corrosion nonlinear model machine for corrosion detection; the model realizes accurate judgment of corrosion by combining a hyperbolic tangent function and an exponential decay function; for a corrosion point, using a corrosion degree assessment method to analyze the electrochemical data of the corrosion point, obtain relevant features and relevant feature importance coefficients, and calculate the corrosion degree assessment value according to the corrosion degree assessment function; judging the corrosion level by a corrosion degree assessment threshold, so as to timely reflect the corrosion situation; finally, using a corrosion rate calculation method to obtain the influence of relevant factors on the corrosion rate from the difference between the environmental data and the corrosion features at multiple detection times, so as to improve the calculation accuracy and facilitate the maintenance of the signal tower.

Description

Technical Field

[0001] The present invention relates to the technical field of electrochemical detection, and in particular to a nonlinear model detection method and system for an electrochemical sensor. Background Art

[0002] Electrochemistry is the science of studying the charged interface phenomena formed by two types of conductors and the changes that occur on them. Electrochemical sensors are made based on ion conductivity. According to the formation of their electrical characteristics, electrochemical sensors can be divided into potentiometric sensors, conductivity sensors, coulometric sensors, polarographic sensors and electrolytic sensors. Electrochemical sensors are mainly used to analyze gas, liquid or solid components dissolved in liquid, pH value of liquid, conductivity and redox potential and other parameters.

[0003] The application of electrochemistry includes many fields, such as pipeline corrosion monitoring, steel structure corrosion monitoring and metal equipment corrosion assessment. In steel corrosion detection, electrochemical technology can be used to monitor and analyze the corrosion process to help prevent and control the corrosion of steel structures.

[0004] In terms of steel structure corrosion detection, signal towers, as an important application scenario of steel structures, have special corrosion risks. These structures are usually exposed to various environmental conditions, such as wind, rain, sunlight, and pollutants, all of which may accelerate the corrosion process. In order to effectively deal with the corrosion risk of signal towers, it is crucial to adopt advanced corrosion detection technology.

[0005] In the existing technology, corrosion detection of steel structures of signal towers mainly relies on drone technology, and experts make judgments by collecting surface images of steel structures; however, this method has certain defects. The stability of the drone's flight cannot be guaranteed at high altitudes, and the complex steel structure may cause reflection and occlusion of the image, which will interfere with image acquisition. This makes it impossible to obtain effective data for accurate analysis, which is not conducive to effective maintenance of signal towers when corrosion occurs.

[0006] To this end, the present invention proposes a nonlinear model detection method and system for an electrochemical sensor. Summary of the invention

[0007] The purpose of the present invention is to provide a nonlinear model detection method and system of an electrochemical sensor for detecting the steel corrosion of a signal tower; it mainly starts from the following points: first, the corresponding electrochemical data is obtained by an electrochemical sensor installed on the steel surface of the signal tower, and a corrosion detection data set is constructed according to the detection time, detection position and electrochemical data; secondly, the present invention proposes a corrosion nonlinear model to perform corrosion detection on the preprocessed corrosion detection data set; the model combines the hyperbolic tangent function and the exponential decay function to establish a nonlinear model, which can identify small corrosion changes, thereby improving the sensitivity of detection and reducing the possibility of missed detection; wherein, the hyperbolic tangent function It can effectively capture the nonlinear relationship in the data, while the exponential decay function can simulate the dynamic changes of the corrosion process. The combination of the two can more accurately identify and describe complex corrosion characteristics. Finally, the present invention adopts a corrosion degree assessment method to evaluate the corrosion points on the steel surface for the detected corrosion positions. The method performs relevant feature analysis on the collected electrochemical data to obtain time domain features, frequency domain features, polarization impedance features, statistical features and correlation features, and evaluates the importance of each feature. The corrosion degree of each corrosion point is calculated in combination with the corrosion features and the importance of the features. In addition, the present invention calculates the corrosion rate based on the collected environmental data in combination with the corrosion degree.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] A nonlinear model detection method for an electrochemical sensor includes: installing the electrochemical sensor at different positions on the steel surface of a signal tower;

[0010] Further, collecting electrochemical data of the electrochemical sensor at multiple detection times;

[0011] Further, constructing a corrosion detection data set according to the detection time, the detection position of the electrochemical sensor and the electrochemical data;

[0012] Furthermore, data preprocessing is performed on the data in the corrosion detection data set to obtain a standard corrosion detection data set; wherein the data preprocessing includes: data cleaning of the electrochemical data to obtain first electrochemical data; wherein the data cleaning includes: denoising, missing value filling and outlier processing; normalizing the first electrochemical data to obtain second electrochemical data; extracting features from the second electrochemical data to obtain third electrochemical data; performing time series analysis on the third electrochemical data using a sliding window technique to obtain fourth electrochemical data; data encoding the installation position of the electrochemical sensor to obtain position coding information; aligning the fourth electrochemical data and the position coding information to obtain the standard corrosion detection data set.

[0013] Furthermore, corrosion detection is performed on the standard corrosion detection data set using a corrosion nonlinear model; wherein the function of the corrosion nonlinear model is expressed as:

[0014] ;

[0015] in, Expressed as Detection time detection position Probability of corrosion detection; It is expressed as a hyperbolic tangent function to capture the nonlinear characteristics of electrochemical data; It is expressed as an exponential function to handle the decay effect of electrochemical data; Represented as an electrochemical sensor after data preprocessing The acquired data vector; The weights are expressed as hyperbolic tangent functions; Expressed as the slope of the hyperbolic tangent function; The phase shift expressed as a hyperbolic tangent function; The magnitude offset expressed as a hyperbolic tangent function; It is expressed as the impact strength of the attenuated part; It is expressed as the decay rate; Represented as a scaling factor that adjusts the input feature in exponential decay;

[0016] Further, non-corrosion points are removed according to the detection results of the corrosion nonlinear model to obtain a set of corrosion locations;

[0017] Furthermore, the corrosion degree of the detection position in the corrosion position set is evaluated; wherein the specific evaluation process includes: selecting data corresponding to the corrosion position from the standard corrosion detection data set according to the corrosion position set;

[0018] Furthermore, feature engineering technology is used to extract relevant features from the selected data; wherein the relevant features include: time domain features, frequency domain features, polarization impedance features, statistical features and correlation features;

[0019] Further, the importance of the relevant features is evaluated using a machine learning model to obtain the importance coefficient of the relevant features;

[0020] Furthermore, the relevant features and the importance coefficients of the relevant features are substituted into a corrosion degree evaluation function to obtain the corrosion degree; the corrosion degree evaluation function is expressed as: ;in, Expressed as Corrosion points during the inspection time The degree of corrosion; Represents a nonlinear activation function when outputting results; Represented as the conversion A nonlinear function of the relevant features; Expressed as the Analytical model of relevant features; Represented as conversion The relevant feature importance coefficient of the relevant features; Indicated as corrosion point No. relevant features;

[0021] Furthermore, the corrosion level of the corrosion degree is determined according to a corrosion degree assessment threshold; wherein the corrosion level includes: mild, moderate and severe.

[0022] Further, the corrosion rate of the detection position in the corrosion position set is calculated; wherein the corrosion rate calculation process includes: obtaining environmental data within N detection times; wherein the environmental data includes: temperature, humidity, precipitation, wind speed, atmospheric pollutants and light intensity;

[0023] Further, the environmental data is subjected to the data preprocessing to obtain standard environmental data;

[0024] Further, the data difference of the corrosion point within the detection time is calculated; wherein the data difference includes: detection time difference, environmental data difference, corrosion degree difference and corrosion characteristic difference;

[0025] Further, the corrosion rate is obtained according to the data difference, and the calculation formula of the corrosion rate is:

[0026] ;

[0027] in, Expressed as Corrosion points at detection time The corrosion rate; Expressed as The difference in corrosion degree at each detection time; It is expressed as the detection time difference; Expressed as a corrosion rate influence function; Expressed as The difference of environmental data at each detection time; Expressed as Corrosion points at detection time The corrosion characteristic difference of

[0028] Wherein, the corrosion rate influence function is expressed as:

[0029] ;

[0030] in, Expressed as the difference of the environmental data; Expressed as the corrosion characteristic difference; Expressed as temperature difference; Expressed as humidity difference; It is expressed as precipitation difference; Expressed as wind speed difference; Expressed as atmospheric pollutant difference; Expressed as atmospheric pollutant difference; and Expressed as the regression coefficient of the corrosion rate influence function; and Expressed as the coefficient of the corrosion rate influence function.

[0031] Furthermore, feedback is provided on the corrosion condition of the signal tower.

[0032] A nonlinear model detection system for an electrochemical sensor, the system comprising: a detection plan formulation unit, a sensor unit, a data acquisition unit, a data preprocessing unit, an environmental monitoring unit, a nonlinear model unit, a corrosion degree assessment unit, a corrosion rate calculation unit and a feedback unit;

[0033] Wherein, the detection plan formulation unit is used to formulate a corrosion detection plan for the steel surface of the signal tower, including: detection time and detection location;

[0034] The sensor unit is used to install electrochemical sensors at different positions of the signal tower steel structure to monitor corrosion-related electrochemical data in real time;

[0035] The data acquisition unit is used to read data from the electrochemical sensor;

[0036] The data preprocessing unit is used to perform data preprocessing on the collected electrochemical data;

[0037] The environmental monitoring unit is used to collect environmental factor data related to corrosion;

[0038] The nonlinear model unit is used to construct a model to detect corrosion on the steel surface of the signal tower; wherein the nonlinear model unit uses a corrosion nonlinear model to detect whether the steel of the signal tower is corroded, and the specific process includes: the sensor unit collects electrochemical data of the electrochemical sensor within multiple detection times;

[0039] Further, constructing a corrosion detection data set according to the detection time, the detection position of the electrochemical sensor and the electrochemical data;

[0040] Further, data preprocessing is performed on the data in the corrosion detection data set to obtain a standard corrosion detection data set;

[0041] Furthermore, the standard corrosion detection data set is input into the corrosion nonlinear model to obtain the corrosion probability of the detection point; wherein the function of the corrosion nonlinear model is expressed as:

[0042] ;

[0043] in, Expressed as Detection time detection position Probability of corrosion detection; It is expressed as a hyperbolic tangent function to capture the nonlinear characteristics of electrochemical data; It is expressed as an exponential function to handle the decay effect of electrochemical data; Represented as an electrochemical sensor after data preprocessing The acquired data vector; The weights are expressed as hyperbolic tangent functions; Expressed as the slope of the hyperbolic tangent function; The phase shift expressed as a hyperbolic tangent function; The magnitude offset expressed as a hyperbolic tangent function; It is expressed as the impact strength of the attenuated part; It is expressed as the decay rate; It is expressed as a scaling factor that adjusts the input features in exponential decay.

[0044] The corrosion degree evaluation unit is used to calculate the corrosion degree of the corrosion position; wherein the evaluation process of the corrosion degree evaluation unit includes: selecting data corresponding to the corrosion position from the standard corrosion detection data set according to the detection result of the nonlinear model unit;

[0045] Furthermore, feature engineering technology is used to extract relevant features from the selected data; wherein the relevant features include: time domain features, frequency domain features, polarization impedance features, statistical features and correlation features;

[0046] Further, the importance of the relevant features is evaluated using a machine learning model to obtain the importance coefficient of the relevant features;

[0047] Furthermore, the relevant features and the importance coefficients of the relevant features are substituted into a corrosion degree evaluation function to obtain the corrosion degree; the corrosion degree evaluation function is expressed as: ;in, Expressed as Corrosion points during the inspection time The degree of corrosion; Represents a nonlinear activation function when outputting results; Represented as the conversion A nonlinear function of the relevant features; Expressed as the Analytical model of relevant features; Represented as conversion The relevant feature importance coefficient of the relevant features; Indicated as corrosion point No. relevant features;

[0048] Further, the corrosion level of the corrosion level is determined according to the corrosion level assessment threshold; wherein the corrosion level includes: mild, moderate and severe;

[0049] The corrosion rate calculation unit is used to calculate the corrosion rate; wherein the corrosion rate calculation process of the corrosion rate calculation unit includes: obtaining environmental data within N detection times; wherein the environmental data includes: temperature, humidity, precipitation, wind speed, atmospheric pollutants and light intensity;

[0050] Further, the environmental data is subjected to the data preprocessing to obtain standard environmental data;

[0051] Further, the data difference of the corrosion point within the detection time is calculated; wherein the data difference includes: detection time difference, environmental data difference, corrosion degree difference and corrosion characteristic difference;

[0052] Furthermore, the corrosion rate is obtained according to the data difference, and the calculation formula of the corrosion rate is:

[0053] ;

[0054] in, Expressed as Corrosion points at detection time The corrosion rate; Expressed as The difference in corrosion degree at each detection time; It is expressed as the detection time difference; Expressed as a corrosion rate influence function; Expressed as The difference of environmental data at each detection time; Expressed as Corrosion points at detection time The corrosion characteristic difference of

[0055] Wherein, the corrosion rate influence function is expressed as:

[0056] ;

[0057] in, Expressed as the difference of the environmental data; Expressed as the corrosion characteristic difference; Expressed as temperature difference; Expressed as humidity difference; It is expressed as precipitation difference; Expressed as wind speed difference; Expressed as atmospheric pollutant difference; Expressed as atmospheric pollutant difference; and Expressed as the regression coefficient of the corrosion rate influence function; and Expressed as the coefficient of the corrosion rate influence function.

[0058] The feedback unit is used to feedback the corrosion condition of the signal tower.

[0059] Compared with the prior art, the present invention has the following beneficial effects:

[0060] 1. The present invention proposes a corrosion linear model for corrosion detection of steel structures of signal towers; the model captures the nonlinear relationship of corrosion characteristics in electrochemical data by combining the hyperbolic tangent function and the exponential decay function; wherein the hyperbolic tangent function can effectively capture the nonlinear relationship in the data, and the exponential decay function can simulate the dynamic changes of the corrosion process; the combination of the two can identify tiny corrosion changes, thereby improving the sensitivity of detection, which is beneficial for timely maintenance of signal towers.

[0061] 2. The present invention proposes a method for calculating the corrosion degree of corrosion points of signal towers; the method analyzes the characteristics related to the corrosion degree from the electrochemical data of the corrosion points, including: time domain characteristics, frequency domain characteristics, polarization impedance characteristics, statistical characteristics and correlation characteristics; and evaluates the importance of relevant characteristics through a machine learning model; and uses a corrosion degree evaluation function to accurately evaluate the corrosion degree of the corrosion points, providing an important data basis for the subsequent maintenance of signal towers.

[0062] 3. The present invention proposes a method for calculating the corrosion rate of a signal tower; the method calculates the detection time difference, environmental data difference, corrosion degree difference and corrosion characteristic difference of multiple detection times to obtain the difference of different detection time points, and uses the corrosion rate influence function to analyze this difference to obtain the corrosion change rate at different detection times; the corrosion rate influence function analyzes the influence of environmental characteristics and corrosion characteristics on corrosion through nonlinear analysis; this method can accurately reflect the changes in corrosion at different detection times, thereby improving the maintenance efficiency of signal towers. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 A flow chart of a nonlinear model detection method for an electrochemical sensor provided by an embodiment of the present invention;

[0064] Figure 2 A structural diagram of a nonlinear model detection system for an electrochemical sensor provided by an embodiment of the present invention;

[0065] Figure 3 A schematic diagram of corrosion detection in Example 1 provided by an embodiment of the present invention;

[0066] Figure 4 A schematic diagram of the corrosion rate of the detection point of Example D1 provided by an embodiment of the present invention;

[0067] Figure 5 A schematic diagram of the corrosion rate of Example 2 provided for an embodiment of the present invention.

[0068] In the figure: 1, electrochemical sensor No. 1; 2, electrochemical sensor No. 2; 3, electrochemical sensor No. 3; 4, electrochemical sensor No. 4; 5, electrochemical sensor No. 5. DETAILED DESCRIPTION

[0069] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0070] Signal towers are structures that receive and transmit signals and are usually used for communications, broadcasting, and data transmission. They can support different types of communication equipment, such as mobile phone base stations, radio transmitters, and satellite antennas. In order to ensure that the signal can cover a wider area and avoid obstacles that interfere with signal propagation, signal towers are generally built outdoors and at very high heights. Signal towers usually use steel as a supporting structure to ensure their stability.

[0071] During the use of signal towers, due to long-term exposure to the outdoors, the supporting steel structure will rust and break, which has a significant impact on the safety of the signal tower; therefore, it is necessary to conduct corrosion inspections on the steel structure of the signal tower from time to time and maintain the signal tower based on the inspection results.

[0072] In the prior art, corrosion detection of the steel structure of the signal tower mainly relies on UAV technology, and experts make judgments by collecting surface images of the steel structure; however, this method has certain defects. The flight stability of the UAV cannot be guaranteed at high altitudes, and the complex steel structure may cause reflection and occlusion of the image, which will interfere with image acquisition, thus making it impossible to obtain effective data for accurate analysis; therefore, the present invention proposes a nonlinear model detection method and system of an electrochemical sensor to perform corrosion detection on the steel structure of the signal tower; the present invention arranges electrochemical sensors on the surface of the steel structure of the signal tower and analyzes the acquired electrochemical data to achieve accurate corrosion detection of the steel structure of the signal tower; the present invention will be described in detail from two embodiments below.

[0073] Embodiment 1:

[0074] In this embodiment, the signal tower A is described by combining the method and system proposed in the present invention; the signal tower A is a signal tower, mainly used for mobile communication, wireless Internet and broadcast information transmission; its height is about 120 meters; the tower body is a self-supporting steel pipe tower, composed of multiple steel pipes. Figure 1 and Figure 2 of common realization; among which, Figure 1 The steps in Figure 2 Corresponding to some units in; Figure 1 The steps include: S10. installing an electrochemical sensor; S20. collecting electrochemical data at multiple detection times; S30. constructing a corrosion detection data set; S40. performing data preprocessing on the corrosion detection data set to obtain a standard corrosion detection data set; S50. performing corrosion detection; S60. evaluating the degree of corrosion; S70. calculating the corrosion rate; S80. providing feedback on the corrosion situation. Figure 2 The structure of the system includes: a detection plan making unit, a sensor unit, a data acquisition unit, a data preprocessing unit, an environmental monitoring unit, a nonlinear model unit, a corrosion degree evaluation unit, a corrosion rate calculation unit and a feedback unit. According to the description of the above-mentioned embodiment.

[0075] According to the above step S10, electrochemical sensors are installed at different positions on the steel surface of the signal tower A; Figure 3 ,exist Figure 3A schematic diagram of the distribution of electrochemical sensors is given in FIG. 1 ; wherein, signal tower A is equipped with 5 electrochemical sensors; they are: electrochemical sensor No. 1 1, electrochemical sensor No. 2 2, electrochemical sensor No. 3 3, electrochemical sensor No. 4 4 and electrochemical sensor No. 5 5; for the convenience of description, D1, D2, D3, D4 and D5 are used to replace the description; refer to the location information of the electrochemical sensors in Table 1;

[0076] Table 1 Electrochemical sensor location information

[0077]

[0078] In Table 1, the positions of the electrochemical sensors are represented by x, y and z respectively; x represents the horizontal distance from the tower base; y represents the horizontal distance from the center of the tower base to the sensor; and z represents the height from the ground.

[0079] Furthermore, the detection plan formulation unit of the system formulates a corrosion detection plan for the steel surface of the signal tower A, including: detection time and detection position; see Table 2;

[0080] Table 2 Corrosion inspection plan

[0081]

[0082] Further, monitoring is performed by the sensor unit of the system according to the corrosion detection plan;

[0083] Further, the electrochemical data monitored by the electrochemical sensor during the detection time is acquired by using the data acquisition unit of the system, corresponding to the above-mentioned step S20; wherein the electrochemical data include: open circuit potential, polarization curve change rate, current density, polarization resistance, AC impedance spectrum, electrochemical noise, conductivity and local potential, etc., which are not limited to the electrochemical data listed above; refer to Table 3, which illustrates the data structure collected by D1 when the detection time is 2024-07-04;

[0084] Table 3 Electrochemical data collected by D1 on 2024-07-04

[0085]

[0086] Furthermore, the detection time, the detection position of the electrochemical sensor and the electrochemical data are used to construct a corrosion detection data set, corresponding to Figure 1 Step S30;

[0087] Furthermore, the data in the corrosion detection data set is preprocessed by the data preprocessing unit of the system, corresponding to the above-mentioned step S40; wherein the data preprocessing includes: performing data cleaning on the electrochemical data to obtain first electrochemical data; wherein the data cleaning includes: denoising, missing value filling and outlier processing; normalizing the first electrochemical data to obtain second electrochemical data; performing feature extraction on the second electrochemical data to obtain third electrochemical data; performing time series analysis on the third electrochemical data using a sliding window technique to obtain fourth electrochemical data; performing data encoding on the installation position of the electrochemical sensor to obtain position encoding information; aligning the fourth electrochemical data and the position encoding information to obtain the standard corrosion detection data set.

[0088] Refer to Table 4, where examples after data preprocessing are given;

[0089] Table 4 D1 data with detection time 2024-07-04 after data preprocessing

[0090]

[0091] In the data preprocessing process, the D1 electrochemical data with the detection time of 2024-07-04 were analyzed at intervals of 10 min using the sliding window technology;

[0092] In the embodiment of the present application, the electrochemical data collected by the electrochemical sensor is preprocessed; wherein, data cleaning (including denoising, missing value filling and outlier processing) ensures that the data is accurate and reliable, and reduces errors introduced by data quality issues; normalization processing standardizes data of different ranges so that the data in subsequent analysis are consistent and easy to compare and analyze; feature extraction helps to extract meaningful variables, simplify data, highlight key factors affecting corrosion, and improve the accuracy of corrosion detection; sliding window technology allows for more detailed analysis of time series data and captures the changing trends and periodicity of corrosion rates; through position encoding and data alignment, data at different locations can be processed uniformly and analyzed comprehensively to obtain a more comprehensive corrosion status assessment.

[0093] Furthermore, the nonlinear model unit of the system performs corrosion detection on the preprocessed data, corresponding to the above step S50; wherein the nonlinear model unit adopts a corrosion nonlinear model, and its model function is expressed as:

[0094] ;

[0095] in, Expressed as Detection time detection position Probability of corrosion detection; It is expressed as a hyperbolic tangent function to capture the nonlinear characteristics of electrochemical data; It is expressed as an exponential function to handle the decay effect of electrochemical data; Represented as an electrochemical sensor after data preprocessing The acquired data vector; The weights are expressed as hyperbolic tangent functions; Expressed as the slope of the hyperbolic tangent function; The phase shift expressed as a hyperbolic tangent function; The magnitude offset expressed as a hyperbolic tangent function; It is expressed as the impact strength of the attenuated part; It is expressed as the decay rate; It is expressed as a scaling factor that adjusts the input features in exponential decay.

[0096] For example, according to Table 4, the corrosion detection is expressed as:

[0097] ;

[0098] In the embodiment of the present application, the historical data is analyzed according to the big data technology, and the parameters of the corrosion nonlinear model are set as follows: , , , , , , According to the results, the corrosion probability of the D1 detection point with the detection time of 2024-07-04 is: ; From the above results, it can be seen that corrosion occurs at the D1 detection point;

[0099] Furthermore, according to the detection result of the nonlinear model unit, the corrosion points of the signal tower A are used to establish a corrosion point set; from the nonlinear model unit, it can be known that the corrosion points of the signal tower A are: ;

[0100] Further, the corrosion degree of the corrosion points in the corrosion point set is evaluated; wherein the corrosion degree evaluation is to evaluate the corrosion degree evaluation unit of the system at multiple detection time points, corresponding to Figure 1 The S60 step includes:

[0101] Select corresponding data from the collected standard corrosion detection data set according to the corrosion points in the corrosion point set;

[0102] Using feature engineering technology to extract relevant features from the selected data; wherein the relevant features include: time domain features, frequency domain features, polarization impedance features, statistical features and correlation features;

[0103] Among them, the specific implementation process of feature engineering technology is as follows: according to the corrosion detection plan, the distribution of time domain features is: 3 days, 5 days, 7 days, 6 days, 7 days, 8 days, 7 days, 7 days and 11 days; for different time domain distributions, the change rate, mean, variance, standard deviation of relevant features and the correlation between features can be obtained;

[0104] Further, the importance of the relevant features is evaluated using a machine learning model to obtain the importance coefficient of the relevant features; wherein the importance coefficient of the relevant features refers to the contribution of the characteristic parameters in the relevant features to the evaluation results during the corrosion degree evaluation process; such as using a random forest model for evaluation; refer to the evaluation results of the importance coefficient of the relevant features given in Table 5;

[0105] Table 5 Evaluation results of relevant feature importance

[0106]

[0107] Table 5 gives the relevant feature importance coefficients for corrosion degree assessment, for example: current density is 0.35; polarization resistance is 0.25; conductivity is 0.15; open circuit potential is 0.10; polarization curve change rate is 0.08; local potential is 0.05; and electrochemical noise is 0.02.

[0108] Furthermore, the relevant features and the importance coefficients of the relevant features are substituted into a corrosion degree evaluation function to obtain the corrosion degree; the corrosion degree evaluation function is expressed as: ;in, Expressed as Corrosion points during the inspection time The degree of corrosion; Represents a nonlinear activation function when outputting results; Represented as the conversion A nonlinear function of the relevant features; Expressed as the Analytical model of relevant features; Represented as conversion The relevant feature importance coefficient of the relevant features; Indicated as corrosion point No. relevant features;

[0109] For example, the degree of the D1 detection point with the detection time of 2024-07-04 is:

[0110] ;

[0111] Furthermore, the corrosion level of the corrosion level is determined according to the corrosion level assessment threshold; wherein the corrosion level includes: mild, moderate and severe. wherein the corrosion level assessment threshold is set to: ; Refer to Table 6, which shows the corrosion evaluation of the D1 detection point;

[0112] Table 6 Corrosion evaluation of D1 test point at different test times

[0113]

[0114] In Table 6, the corrosion degree increases from 0.920 to 0.980, showing an increasing trend overall; on 2024-07-22 and 2024-07-29, the corrosion degree fluctuates slightly (for example, it rises from 0.940 to 0.955, but then falls back to 0.940), which indicates that the corrosion degree fluctuates to a certain extent at different time points. This fluctuation may be caused by environmental factors, changes in material properties, or detection errors. In practical applications, the fluctuation of corrosion degree can help analyze and predict the laws and trends of corrosion behavior.

[0115] In the implementation of this application, a corrosion degree assessment method is used to evaluate the detection points with corrosion in the signal tower A; this method analyzes the relevant features in the electrochemical data, and uses a machine learning model to evaluate the importance of the relevant features to obtain the relevant feature importance coefficients; finally, the corrosion degree assessment function is used to comprehensively consider the relevant features and coefficients, which not only improves the accuracy and interpretability of the assessment, but also enhances the decision-making support capabilities, and has great practical application value.

[0116] Furthermore, the corrosion rate calculation unit of the system calculates the change rate of the corrosion point during multiple detection times, corresponding to step S70; the specific process includes:

[0117] The environmental monitoring unit of the system acquires environmental data within 10 detection times; wherein the environmental data includes: temperature, humidity, precipitation, wind speed, atmospheric pollutants and light intensity;

[0118] Further, the environmental data is subjected to the data preprocessing to obtain standard environmental data;

[0119] Further, the data difference of the corrosion point within the detection time is calculated; wherein the data difference includes: detection time difference, environmental data difference, corrosion degree difference and corrosion characteristic difference;

[0120] Furthermore, the corrosion rate is obtained according to the data difference, and the calculation formula of the corrosion rate is:

[0121] ;

[0122] in, Expressed as Corrosion points at detection time The corrosion rate; Expressed as The difference in corrosion degree at each detection time; It is expressed as the detection time difference; Expressed as a corrosion rate influence function; Expressed as The difference of environmental data at each detection time; Expressed as Corrosion points at detection time The corrosion characteristic difference of

[0123] Wherein, the corrosion rate influence function is expressed as:

[0124] ;

[0125] in, Expressed as the difference of the environmental data; Expressed as the corrosion characteristic difference; Expressed as temperature difference; Expressed as humidity difference; It is expressed as precipitation difference; Expressed as wind speed difference; Expressed as atmospheric pollutant difference; Expressed as atmospheric pollutant difference; and Expressed as the regression coefficient of the corrosion rate influence function; and Expressed as the coefficient of the corrosion rate influence function;

[0126] In the embodiment of the present application, the parameters in the corrosion rate influence function are set as ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ;

[0127] In the embodiment of the present application, in order to accurately grasp the corrosion change rate of the signal tower; a corrosion rate influence function is established based on the difference between environmental data and corrosion characteristics; this function often uses nonlinear methods to explore the relationship between various influencing factors, and can provide more accurate corrosion rate calculations.

[0128] For the corrosion condition of the D1 detection point, refer to Table 7;

[0129] Table 7 Corrosion of D1 test point

[0130]

[0131] The data in Table 7 show that the corrosion rate varies in different time periods. For example, the rate from 2024-07-09 to 2024-07-16 is higher (0.00214 units / day), while the rate from 2024-08-20 to 2024-08-31 is lower (0.00045 units / day). Figure 4 As shown in the figure, the rate from 2024-07-16 to 2024-07-22 is negative (-0.00083 units / day), indicating that the degree of corrosion actually decreased during this time period. This may be due to data errors or changes in environmental factors. Although the corrosion rate is lower in some time periods, the overall trend can help understand the acceleration or slowdown of corrosion. This helps to formulate maintenance and protection measures. Table 7 helps to gain a deeper understanding of the dynamic changes in the corrosion process of signal tower A at the D1 detection point and helps predict future corrosion conditions.

[0132] Furthermore, the feedback unit of the system provides feedback on the corrosion condition of the signal tower A, corresponding to step S80;

[0133] In the embodiment of the present application, by combining the nonlinear model detection method and system of the electrochemical sensor proposed in the present invention, corrosion detection, corrosion degree assessment and corrosion rate calculation of the steel structure of the signal tower A are realized; mainly including the following points: first, the electrochemical sensor arranged on the surface of the steel is used to obtain electrochemical data within the detection time; second, a corrosion nonlinear model is used to analyze the collected electrochemical data to determine whether the signal tower A is corroded; the model improves the detection accuracy by combining the tangent hyperbolic function and the exponential decay function; for the detected corrosion points, a corrosion degree assessment method is adopted, which obtains the importance coefficient of the relevant features for the corrosion degree assessment by analyzing the electrochemical data; and the corrosion degree assessment function is used to comprehensively consider the relevant features and coefficients to achieve accurate assessment; finally, a corrosion rate calculation method is used to calculate the corrosion rate of the signal tower A; the method uses environmental characteristics and corrosion characteristics to establish a corrosion rate influence function, which improves the accuracy of the calculation by nonlinearly analyzing the relationship between the influencing factors; the above methods can be combined to realize the detection of the steel structure corrosion of the signal tower A, which is conducive to the staff to carry out timely maintenance of the signal tower A.

[0134] Embodiment 2:

[0135] In Example 1, the signal tower A was subjected to nonlinear corrosion detection of the electrochemical sensor model. In the process of studying the corrosion detection of the signal tower, it was found that the corrosion condition of the signal tower would change due to changes in the environment. For example, the height of the corrosion point in the signal tower was different, and its corrosion state was greatly affected. Or the different geographical locations of the signal tower also had different effects on corrosion. Usually, signal towers erected in areas close to the sea are more prone to corrosion, and the corrosion rate is faster than that in other areas. The main reason is that there is a large amount of salt in the marine environment, which forms salt mist in the air and adheres to the surface of the signal tower. These salt mists react with metals, accelerating the oxidation and corrosion of metals.

[0136] In order to further illustrate the effectiveness of the present invention in detecting corrosion of steel structures of signal towers, in the embodiment of the present application, corrosion detection is performed on a signal tower B erected at the same time as the signal tower A; the signal tower B is erected near a coastal city, about 500 meters from the sea, and 80 meters high. It is a self-supporting steel tower (also called a self-supporting communication tower) with good stability and load-bearing capacity; the specific corrosion detection process of the signal tower B includes:

[0137] Obtain electrochemical data collected by electrochemical sensors at multiple detection times; wherein the electrochemical sensors are installed at multiple steel surface locations of signal tower B; wherein 5 electrochemical sensors are installed on signal tower B, and the specific installation locations are: Sensor 1: the top position of signal tower B, which is usually exposed to strong environmental conditions such as wind and precipitation; Sensor 2: the horizontal surface in the middle of signal tower B, which may be exposed to different climate and environmental conditions; Sensor 3: the bottom of signal tower B, close to the ground, which may be affected by moisture and pollution; Sensor 4: one side of signal tower B, so that corrosion data on different directions of the signal tower can be obtained. Sensor 5: the other side of signal tower B, opposite to sensor 4, so as to compare the corrosion conditions of the two sides, and the side installed faces the sea.

[0138] In addition, the salt concentration on signal tower B can be obtained through the electrochemical sensor;

[0139] Further, the acquired electrochemical data is preprocessed to obtain a standard corrosion detection data set;

[0140] Furthermore, the corrosion nonlinear model is used to perform corrosion detection on the location where the sensor is installed; wherein the formula of the corrosion nonlinear model is: ;in, Expressed as Detection time detection position Probability of corrosion detection; It is expressed as a hyperbolic tangent function to capture the nonlinear characteristics of electrochemical data; It is expressed as an exponential function to handle the decay effect of electrochemical data; Represented as an electrochemical sensor after data preprocessing The acquired data vector; The weights are expressed as hyperbolic tangent functions; Expressed as the slope of the hyperbolic tangent function; The phase shift expressed as a hyperbolic tangent function; The magnitude offset expressed as a hyperbolic tangent function; It is expressed as the impact strength of the attenuated part; It is expressed as the decay rate; Represented as a scaling factor that adjusts the input feature in exponential decay;

[0141] Refer to Table 8, which shows the detection status of the detection points;

[0142] Table 8 Corrosion detection of signal tower B

[0143]

[0144] From the data in Table 8, it can be seen that corrosion occurred at all five test points.

[0145] Further, non-corrosion points are removed according to the detection results of the corrosion nonlinear model to obtain a set of corrosion locations;

[0146] Further, data corresponding to the corrosion position is selected from the standard corrosion detection data set according to the corrosion position set; wherein the height information and salt concentration of the detection point are added to the selected data;

[0147] Furthermore, feature engineering techniques are used to extract relevant features from the selected data;

[0148] Further, the importance of the relevant features is evaluated using a machine learning model to obtain the importance coefficient of the relevant features;

[0149] Further, the relevant features and the importance coefficients of the relevant features are substituted into a corrosion degree evaluation function to obtain the corrosion degree;

[0150] Further, the corrosion level of the corrosion level is determined according to the corrosion level assessment threshold; wherein the corrosion level includes: mild, moderate and severe. Referring to Table 9, Table 9 shows the corrosion assessment of the detection point in signal tower B at the detection time of 2024-08-31;

[0151] Table 9 Corrosion assessment of signal tower B

[0152]

[0153] According to the above-mentioned electrochemical sensor positions and their environmental influences, detection point 5 (side, facing the sea): Due to long-term exposure to salt spray, the corrosion degree at this position is usually more serious, so it is likely to be the most seriously corroded area; detection point 1 (top position): Exposed to strong winds and precipitation, these conditions will aggravate corrosion, but the impact of salt spray is usually not as serious as the sea side; detection point 3 (bottom position): Affected by ground moisture and pollutants, the corrosion situation is serious, but with good protective measures, the corrosion level may be lower than the top; detection point 4 (side position): Compared with the top and the sea side, the corrosion situation is usually lighter, but it is still affected by wind, rain and other climatic conditions; detection point 2 (middle horizontal plane): It is relatively evenly affected by climatic conditions, and the corrosion level is usually relatively light.

[0154] Further, see Figure 5 , calculating the corrosion rate of the detection position in the corrosion position set; wherein the corrosion rate calculation process includes:

[0155] Acquire environmental data within multiple detection times; wherein the environmental data includes: temperature, humidity, precipitation, wind speed, atmospheric pollutants, and light intensity;

[0156] Further, the environmental data is subjected to the data preprocessing to obtain standard environmental data;

[0157] Further, the data difference of the corrosion point within the detection time is calculated; wherein the data difference includes: detection time difference, environmental data difference, corrosion degree difference and corrosion characteristic difference;

[0158] Furthermore, the corrosion rate is obtained according to the data difference; the calculation formula of the corrosion rate is:

[0159] ;

[0160] in, Expressed as Corrosion points at detection time The corrosion rate; Expressed as The difference in corrosion degree at each detection time; It is expressed as the detection time difference; Expressed as a corrosion rate influence function; Expressed as The difference of environmental data at each detection time; Expressed as Corrosion points at detection time The corrosion characteristic difference of

[0161] Wherein, the corrosion rate influence function is expressed as:

[0162] ;

[0163] in, Expressed as the difference of the environmental data; Expressed as the corrosion characteristic difference; Expressed as temperature difference; Expressed as humidity difference; It is expressed as precipitation difference; Expressed as wind speed difference; Expressed as atmospheric pollutant difference; Expressed as atmospheric pollutant difference; and Expressed as the regression coefficient of the corrosion rate influence function; and Expressed as the coefficient of the corrosion rate influence function.

[0164] Furthermore, feedback is provided on the corrosion condition of the signal tower.

[0165] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A nonlinear model detection method for an electrochemical sensor, characterized in that: include: Install the electrochemical sensor at different positions on the steel surface of the signal tower; Collecting electrochemical data of the electrochemical sensor during multiple detection times; constructing a corrosion detection data set according to the detection time, the detection position of the electrochemical sensor and the electrochemical data; Data preprocessing is performed on the data in the corrosion detection data set to obtain a standard corrosion detection data set; corrosion detection is performed on the standard corrosion detection data set using a corrosion nonlinear model; wherein the function of the corrosion nonlinear model is expressed as: in, It is expressed as the corrosion detection probability of the detection position j in the tth detection time; tanh() is expressed as the hyperbolic tangent function, which is used to capture the nonlinear characteristics of electrochemical data; exp() is expressed as an exponential function, which is used to process the attenuation effect of electrochemical data; represents the data vector obtained by the electrochemical sensor j after data preprocessing; α represents the weight of the hyperbolic tangent function; β represents the slope of the hyperbolic tangent function; γ represents the phase offset of the hyperbolic tangent function; δ represents the amplitude offset of the hyperbolic tangent function; ε represents the influence intensity of the attenuation part; ζ represents the attenuation rate; η represents the scaling factor for adjusting the input feature in the exponential attenuation; According to the detection results of the corrosion nonlinear model, non-corroded points are removed to obtain a set of corrosion positions; the corrosion degree of the detection positions in the set of corrosion positions is evaluated; the corrosion rate of the detection positions in the set of corrosion positions is calculated; and the corrosion condition of the signal tower is fed back.

2. The nonlinear model detection method of an electrochemical sensor according to claim 1, characterized in that: The data preprocessing includes: performing data cleaning on the electrochemical data to obtain first electrochemical data; wherein the data cleaning includes: denoising, missing value filling and outlier processing; normalizing the first electrochemical data to obtain second electrochemical data; performing feature extraction on the second electrochemical data to obtain third electrochemical data; performing time series analysis on the third electrochemical data using a sliding window technique to obtain fourth electrochemical data; performing data encoding on the installation position of the electrochemical sensor to obtain position encoding information; and aligning the fourth electrochemical data with the position encoding information to obtain the standard corrosion detection data set.

3. The nonlinear model detection method of an electrochemical sensor according to claim 1, characterized in that: The process of evaluating the corrosion degree in the corrosion position set includes: selecting data corresponding to the corrosion position from the standard corrosion detection data set according to the corrosion position set; extracting relevant features from the selected data using feature engineering technology; wherein the relevant features include: time domain features, frequency domain features, polarization impedance features, statistical features and correlation features; using a machine learning model to evaluate the importance of the relevant features to obtain the importance coefficient of the relevant features; substituting the relevant features and the importance coefficient of the relevant features into a corrosion degree evaluation function to obtain the corrosion degree; the corrosion degree evaluation function is expressed as: Among them, E j (t) represents the corrosion degree of corrosion point j at the tth detection time; σ represents the nonlinear activation function when outputting the result; φ k () represents the nonlinear function used to transform the kth related feature; M k () represents the analysis model for the kth related feature; w k Expressed as the relevant feature importance coefficient of the converted k-th relevant feature; f jk It is represented as the kth relevant feature of corrosion point j; The corrosion level of the corrosion degree is determined according to a corrosion degree assessment threshold; wherein the corrosion level includes: mild, moderate and severe.

4. The nonlinear model detection method of an electrochemical sensor according to claim 1, characterized in that: The corrosion rate calculation process includes: obtaining environmental data within N detection times; wherein the environmental data include: temperature, humidity, precipitation, wind speed, atmospheric pollutants and light intensity; performing data preprocessing on the environmental data to obtain standard environmental data; calculating the data difference of the corrosion point within the detection time; wherein the data difference includes: detection time difference, environmental data difference, corrosion degree difference and corrosion characteristic difference; obtaining the corrosion rate according to the data difference; Among them, R j (t) represents the corrosion rate of corrosion point j at the tth detection time; ΔE j (t) represents the difference in corrosion degree at the tth detection time; Δt represents the difference in detection time; g() represents the corrosion rate influence function; ΔEnv(t) represents the difference in environmental data at the tth detection time; Δf j (t) represents the corrosion characteristic difference of corrosion point j at the tth detection time.

5. The nonlinear model detection method of an electrochemical sensor according to claim 4, characterized in that: The corrosion rate influence function is expressed as: Wherein, ΔEnv represents the difference in environmental data; Δf represents the difference in corrosion characteristics; ΔTem represents the temperature difference; ΔHum represents the humidity difference; ΔPre represents the precipitation difference; ΔWin represents the wind speed difference; ΔPol represents the atmospheric pollutant difference; Δlig represents the atmospheric pollutant difference; β0, β1, β2, β3, β4, β5, β6, β7, β8, β9 and β 10 It is expressed as the regression coefficient of the corrosion rate influence function; α1, α2, α3, α4 and α5 are expressed as the coefficients of the corrosion rate influence function.

6. A nonlinear model detection system for an electrochemical sensor, characterized in that: The system comprises: a detection plan formulation unit, a sensor unit, a data acquisition unit, a data preprocessing unit, an environmental monitoring unit, a nonlinear model unit, a corrosion degree assessment unit, a corrosion rate calculation unit and a feedback unit; The detection plan formulation unit is used to formulate a corrosion detection plan for the steel surface of the signal tower, including: detection time and detection position; the sensor unit is used to install electrochemical sensors at different positions of the steel structure of the signal tower to monitor electrochemical data related to corrosion in real time; the data acquisition unit is used to read data from the electrochemical sensor; the data preprocessing unit is used to perform data preprocessing on the collected electrochemical data; the environmental monitoring unit is used to collect environmental factor data related to corrosion; the nonlinear model unit is used to construct a model to perform corrosion detection on the steel surface of the signal tower; The nonlinear model unit uses a corrosion nonlinear model to detect whether the steel of the signal tower is corroded. The specific process includes: the sensor unit collects electrochemical data of the electrochemical sensor within multiple detection times; constructs a corrosion detection data set according to the detection time, the detection position of the electrochemical sensor and the electrochemical data; performs data preprocessing on the data in the corrosion detection data set to obtain a standard corrosion detection data set; inputs the standard corrosion detection data set into the corrosion nonlinear model to obtain the corrosion probability of the detection point; wherein the function of the corrosion nonlinear model is expressed as: in, It is expressed as the corrosion detection probability of the detection position j in the tth detection time; tanh() is expressed as the hyperbolic tangent function, which is used to capture the nonlinear characteristics of electrochemical data; exp() is expressed as an exponential function, which is used to process the attenuation effect of electrochemical data; represents the data vector obtained by the electrochemical sensor j after data preprocessing; α represents the weight of the hyperbolic tangent function; β represents the slope of the hyperbolic tangent function; γ represents the phase offset of the hyperbolic tangent function; δ represents the amplitude offset of the hyperbolic tangent function; ε represents the influence intensity of the attenuation part; ζ represents the attenuation rate; η represents the scaling factor for adjusting the input feature in the exponential attenuation; The corrosion degree evaluation unit is used to calculate the corrosion degree of the corrosion position; the corrosion rate calculation unit is used to calculate the corrosion rate; and the feedback unit is used to feedback the corrosion condition of the signal tower.

7. The nonlinear model detection system of an electrochemical sensor according to claim 6, characterized in that: The corrosion degree evaluation unit comprises: selecting data corresponding to the corrosion position from the standard corrosion detection data set according to the detection result of the nonlinear model unit; extracting relevant features from the selected data by using feature engineering technology; wherein the relevant features include: time domain features, frequency domain features, polarization impedance features, statistical features and correlation features; using a machine learning model to evaluate the importance of the relevant features to obtain the importance coefficient of the relevant features; substituting the relevant features and the importance coefficient of the relevant features into the corrosion degree evaluation function to obtain the corrosion degree; the corrosion degree evaluation function is expressed as: Among them, E j (t) represents the corrosion degree of corrosion point j at the tth detection time; σ represents the nonlinear activation function when outputting the result; φ k () represents the nonlinear function used to transform the kth related feature; M k () represents the analysis model for the kth related feature; w k Expressed as the relevant feature importance coefficient of the converted k-th relevant feature; f jk It is represented as the kth relevant feature of corrosion point j; The corrosion level of the corrosion degree is determined according to a corrosion degree assessment threshold; wherein the corrosion level includes: mild, moderate and severe.

8. The nonlinear model detection system of an electrochemical sensor according to claim 6, characterized in that: The corrosion rate calculation unit comprises: obtaining environmental data within N detection times; wherein the environmental data comprises: temperature, humidity, precipitation, wind speed, atmospheric pollutants and light intensity; performing data preprocessing on the environmental data to obtain standard environmental data; calculating the data difference of the corrosion point within the detection time; wherein the data difference comprises: detection time difference, environmental data difference, corrosion degree difference and corrosion characteristic difference; obtaining the corrosion rate according to the data difference; Among them, R j (t) represents the corrosion rate of corrosion point j at the tth detection time; ΔE j (t) represents the difference in corrosion degree at the tth detection time; Δt represents the difference in detection time; g() represents the corrosion rate influence function; ΔEnv(t) represents the difference in environmental data at the tth detection time; Δf j (t) represents the corrosion characteristic difference of corrosion point j at the tth detection time.

9. The nonlinear model detection system of an electrochemical sensor according to claim 8, characterized in that: The corrosion rate influence function is expressed as: Wherein, ΔEnv represents the difference in environmental data; Δf represents the difference in corrosion characteristics; ΔTem represents the temperature difference; ΔHum represents the humidity difference; ΔPre represents the precipitation difference; ΔWin represents the wind speed difference; ΔPol represents the atmospheric pollutant difference; Δlig represents the atmospheric pollutant difference; β0, β1, β2, β3, β4, β5, β6, β7, β8, β9 and β 10 It is expressed as the regression coefficient of the corrosion rate influence function; α1, α2, α3, α4 and α5 are expressed as the coefficients of the corrosion rate influence function.

Images