304 austenitic stainless steel corrosion resistance monitoring system associated with heat treatment temperature field

Abstract

The invention discloses a heat treatment temperature field associated 304 austenitic stainless steel corrosion resistance monitoring system, which relates to the technical field of metal material heat treatment monitoring, and comprises a temperature field monitoring network, an electrochemical noise monitoring module and a data fusion and intelligent analysis center, the temperature field monitoring network monitors temperature field distribution of a heat treatment workpiece in real time and generates temperature field data, the electrochemical noise monitoring module collects electrochemical noise signals of the workpiece in real time and generates electrochemical noise data, and the data fusion and intelligent analysis center performs time synchronization and space registration to form synchronous data after receiving the two types of data. Temperature field features and electrochemical noise features are extracted, the corrosion risk probability is calculated through an intelligent judgment model, a corrosion risk early warning signal is generated when the probability exceeds a preset threshold value, the system establishes real-time correlation between the temperature field and corrosion resistance, and early recognition and accurate positioning of the corrosion risk are achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of metal material heat treatment monitoring, in particular to a 304 austenitic stainless steel corrosion resistance monitoring system related to heat treatment temperature field. BACKGROUND

[0002] 304 austenitic stainless steel is widely used in petrochemical, food and pharmaceutical industries due to its good corrosion resistance and forming performance. During heat treatment, when the material is in the sensitization temperature range of 450℃ to 850℃, chromium carbide will precipitate at the grain boundary, resulting in the formation of chromium-depleted zone near the grain boundary, thereby significantly reducing the intergranular corrosion resistance of the material. In actual industrial production, due to uneven temperature field distribution of heat treatment furnace, complex workpiece shape and other factors, the thermal history experienced by different parts of the workpiece is different, so that the local area may be in the sensitization temperature range for a long time, resulting in a decrease in corrosion resistance, which seriously affects the safety and service life of the equipment.

[0003] The existing monitoring technology mainly has the following limitations: patent CN100373144C discloses a material heat treatment surface temperature real-time monitoring system, which realizes real-time measurement and display of temperature field by arranging a thermocouple network, but can only provide temperature data and cannot directly reflect the influence of temperature distribution on material corrosion resistance; some other patents infer the furnace tube wall corrosion state by monitoring the thermocouple temperature difference, which is simple in principle, but only suitable for rough evaluation of uniform corrosion and cannot accurately identify the intergranular corrosion risk of 304 austenitic stainless steel caused by sensitization. The existing technology generally separates temperature monitoring and corrosion evaluation, lacks in-depth analysis of the correlation between the two, and cannot realize early warning of corrosion risk.

[0004] Therefore, it is urgent to develop a monitoring system that can predict the corrosion risk of 304 austenitic stainless steel in real time and accurately from the heat treatment temperature field and realize early warning. The core technical problem that needs to be solved by the system is: how to establish a real-time correlation model between the heat treatment temperature field and the material corrosion resistance, and through comprehensive analysis of the temperature field distribution characteristics and electrochemical noise signals, to realize early identification and accurate positioning of the corrosion risk, so as to provide technical support for heat treatment process optimization and safe operation of equipment.

[0005] In summary, there is a lack of integrated solution that can organically combine heat treatment temperature field monitoring and corrosion resistance evaluation in the existing technology, which cannot meet the demand for real-time warning of material corrosion risk in industrial production. Therefore, it is of great industrial application value to develop a monitoring system that can correlate heat treatment temperature field and corrosion resistance of 304 austenitic stainless steel. SUMMARY

[0006] The purpose of the present application is to make up for the deficiencies of the prior art, and provide a 304 austenitic stainless steel corrosion resistance monitoring system associated with a heat treatment temperature field, which synchronously registers temperature field data and electrochemical noise data, extracts features, and calculates intelligent models, establishes real-time correlation between temperature field and corrosion resistance, realizes early identification and accurate positioning of corrosion risk, and makes up for the deficiencies of the prior art that separates temperature monitoring and corrosion evaluation and is difficult to early warn.

[0007] The present application provides the following technical solutions to solve the above technical problems: a 304 austenitic stainless steel corrosion resistance monitoring system associated with a heat treatment temperature field, comprising a temperature field monitoring network, an electrochemical noise monitoring module, and a data fusion and intelligent analysis center:

[0008] The temperature field monitoring network is used to monitor the temperature field distribution of the heat treated workpiece in real time and generate temperature field data;

[0009] The electrochemical noise monitoring module is used to monitor the electrochemical noise signal of the heat treated workpiece in real time and generate electrochemical noise data;

[0010] The data fusion and intelligent analysis center is connected with the temperature field monitoring network and the electrochemical noise monitoring module, and is used to receive temperature field data and electrochemical noise data, time synchronize and spatially register the temperature field data and the electrochemical noise data to form synchronous data, extract temperature field features and electrochemical noise features from the synchronous data, calculate corrosion risk probability based on the temperature field features and the electrochemical noise features through an intelligent discrimination model, and generate a corrosion risk warning signal when the corrosion risk probability exceeds a preset threshold.

[0011] Further, the temperature field monitoring network comprises a plurality of thermocouples, a temperature data acquisition unit, and a temperature field reconstruction module.

[0012] The plurality of thermocouples are K-type thermocouples that directly contact the surface of the heat treated workpiece and are arranged on the heat treated workpiece at a non-uniform density, wherein the K-type thermocouples are densely arranged in the weld area, the heat affected zone, and the area with large temperature gradient of the heat treated workpiece, and the densely arranged spacing is between 1mm and 10mm, and the spacing in other areas is between 10mm and 50mm.

[0013] The temperature data acquisition unit is configured to synchronously acquire temperature readings of all K-type thermocouples at a sampling rate not lower than 1Hz, and transmit the temperature readings to the temperature field reconstruction module through a communication interface;

[0014] The temperature field reconstruction module is configured to calculate the temperature value of each point on the surface of the heat-treated workpiece based on the temperature readings using an interpolation algorithm to generate a two-dimensional temperature field distribution map, and to calculate the local temperature value, temperature gradient vector and temperature change rate scalar in the temperature field distribution map;

[0015] The temperature gradient vector is represented as the gradient size and direction, and the temperature change rate scalar is represented as the rate of change of temperature over time. The temperature field reconstruction module is further configured to update the temperature field distribution map in real time and send the temperature field data to the data fusion and intelligent analysis center;

[0016] The temperature field monitoring network also includes a calibration module for periodically calibrating the K-type thermocouple to ensure measurement accuracy. The calibration module adjusts the offset error by comparing the standard temperature source with the K-type thermocouple readings.

[0017] Further, the electrochemical noise monitoring module includes a three-electrode system, an electrochemical noise acquisition instrument and a signal preprocessing unit;

[0018] The three-electrode system includes two 304 austenitic stainless steel working electrodes and one reference electrode. The spacing between the two working electrodes is set to be within the range of 1mm to 5mm, and the three-electrode system is connected to the surface of the heat-treated workpiece through an electrolyte.

[0019] The electrochemical noise acquisition instrument is configured to measure the current noise between the two working electrodes and the potential noise between the working electrode and the reference electrode synchronously, with a sampling rate not less than 1000Hz and a bandwidth from direct current to 1000Hz;

[0020] The signal preprocessing unit is configured to perform digital filtering on the original electrochemical noise signal to remove high-frequency noise and 50Hz power frequency interference, and to extract electrochemical noise features, including current noise standard deviation, potential noise standard deviation, noise resistance, wavelet energy entropy and marginal spectrum entropy of Hilbert-Huang transform;

[0021] The wavelet energy entropy is the Shannon entropy of the energy value of each frequency band calculated after the signal is decomposed into multiple frequency bands by wavelet transform, and the marginal spectrum entropy is the Shannon entropy of the spectrum value calculated after the marginal spectrum of the signal is obtained by Hilbert-Huang transform;

[0022] The electrochemical noise monitoring module also includes an electrode maintenance unit for monitoring the electrode state and automatically reminding to replace or clean the electrode to ensure long-term measurement stability.

[0023] Further, the data fusion and intelligent analysis center includes a data synchronization module, a feature extraction module, an intelligent discrimination module and a warning module;

[0024] The data synchronization module is configured to: stamp the temperature field data and the electrochemical noise data with a unified timestamp, the timestamp precision is not less than 1 μs, and spatial registration is performed, for associating each electrochemical noise measurement point with the nearest temperature field measurement point to form synchronized data;

[0025] The feature extraction module is configured to: extract temperature field features and electrochemical noise features from the synchronized data, wherein the temperature field features include local temperature value, temperature gradient size and temperature change rate, and the electrochemical noise features include current noise standard deviation, noise resistance, wavelet energy entropy and marginal spectrum entropy;

[0026] The intelligent discrimination module is configured to: based on the temperature field features and the electrochemical noise features, calculate the corrosion risk probability using an improved BP neural network model, the improved BP neural network model includes an input layer, at least one hidden layer and an output layer, wherein the number of input layer nodes corresponds to the number of features, and the number of output layer nodes is 1, indicating the corrosion risk probability;

[0027] The early warning module is configured to: when the corrosion risk probability exceeds a preset threshold, generate an early warning signal and trigger an audible and light alarm, and send early warning information to an external display or control system;

[0028] The data fusion and intelligent analysis center further integrates a data storage unit for archiving historical data to support model retraining and trend analysis.

[0029] Further, the data synchronization module includes a time synchronization unit and a spatial registration unit;

[0030] The time synchronization unit is configured to: use a high-precision clock source to stamp the temperature field data and the electrochemical noise data with a timestamp, the timestamp precision is 1 μs, and ensure that the temperature field data and the timestamp are aligned with the electrochemical noise data and the timestamp on the time axis;

[0031] The spatial registration unit is configured to: based on the heat treatment workpiece geometric model and the sensor position information, map the position of each electrochemical noise measurement point to the heat treatment workpiece surface coordinates, and find the nearest temperature field measurement point to establish the spatial correlation between the electrochemical noise measurement point and the temperature field measurement point;

[0032] The spatial registration unit is further configured to: when the temperature field measurement point and the electrochemical noise measurement point do not coincide, calculate the temperature field value at the electrochemical noise measurement point through an interpolation algorithm to ensure that the temperature field features and the electrochemical noise features are consistent in space;

[0033] The data synchronization module further includes a data verification subunit for detecting data loss or abnormality, and triggering reacquisition when detecting data loss or abnormality to ensure data integrity.

[0034] Furthermore, the feature extraction module includes a temperature field feature extraction unit and an electrochemical noise feature extraction unit;

[0035] The temperature field feature extraction unit is configured to extract local temperature values, temperature gradient vectors, and temperature change rate scalars from synchronous data. The local temperature values ​​are the temperature values ​​at the electrochemical noise measurement points. The temperature gradient vectors are obtained by calculating the temperature difference between adjacent points in the temperature field distribution. The temperature change rate scalars are obtained by calculating the rate of temperature change over time at the same location.

[0036] The electrochemical noise feature extraction unit is configured to extract the current noise standard deviation, potential noise standard deviation, noise resistance, wavelet energy entropy, and marginal spectral entropy from the filtered electrochemical noise signal. The current noise standard deviation and potential noise standard deviation are the standard deviations of the signal in the time domain, the noise resistance is the ratio of the potential noise standard deviation to the current noise standard deviation, the wavelet energy entropy is calculated by decomposing the electrochemical noise signal into sub-bands of multiple scales using wavelet transform, and the marginal spectral entropy is calculated by obtaining the Hilbert spectrum of the signal by performing Hilbert-Huang transform on the electrochemical noise signal, and then calculating the Shannon entropy of the marginal spectrum.

[0037] The feature extraction module is further configured to normalize the extracted features to eliminate the influence of dimensions and improve model convergence.

[0038] Furthermore, the intelligent discrimination module uses an improved BP neural network model to calculate the corrosion risk probability. This improved BP neural network model includes an input layer, a hidden layer, and an output layer. The input layer nodes receive temperature field features and electrochemical noise features, while the hidden layer nodes use the following custom activation function:

[0039]

[0040] in These are the input values ​​for the hidden layer nodes. It is the scaling factor of the activation function. It is the shape parameter of the activation function. These are the slope parameters of the activation function, which are obtained through training;

[0041] The output layer nodes of the improved BP neural network model use the sigmoid function to output the corrosion risk probability.

[0042] The improved BP neural network model is trained using the following custom loss function:

[0043]

[0044] in It is the value of the loss function. It is the number of training samples. It is the first The true corrosion risk label for each sample It is the first The predicted corrosion risk probability for each sample. It is the L2 regularization parameter. It is the first Each model weight value, It is the association regularization parameter. It is the first A temperature field characteristic value, It is the first One electrochemical noise characteristic value, It is the quantity of temperature field characteristics. It is the number of electrochemical noise features;

[0045] The intelligent discrimination module is also configured to periodically update model parameters to adapt to process changes.

[0046] Furthermore, the early warning module includes a threshold comparison unit, an alarm generation unit, a communication interface, and a log recording unit;

[0047] The threshold comparison unit is configured to compare the corrosion risk probability calculated by the intelligent discrimination module with a preset threshold. The preset threshold is set in the range of 0.5 to 0.9 according to the material and heat treatment process requirements of 304 austenitic stainless steel, and can be dynamically adjusted.

[0048] The alarm generation unit is configured to generate a corrosion risk warning signal and trigger an audible and visual alarm device when the corrosion risk probability exceeds a preset threshold. The audible and visual alarm device includes a buzzer, an LED indicator and a display screen, wherein the display screen displays the risk location and probability value.

[0049] The communication interface is configured to send the warning signal to external devices, including the heat treatment control system, the central monitoring room, or a mobile terminal, via wired or wireless means, so that operators can take corrective measures in a timely manner.

[0050] The log recording unit is configured to record the occurrence time of each warning event, the location of the heat-treated workpiece, the corrosion risk probability value, and the corresponding temperature field and electrochemical noise characteristic values, and generate a detailed warning report for historical data query and trend analysis.

[0051] Furthermore, the early warning module also integrates a feedback mechanism to automatically adjust the heat treatment process parameters based on the early warning results to reduce the risk of corrosion.

[0052] Compared with existing technologies, this corrosion resistance monitoring system for 304 austenitic stainless steel correlated with the heat treatment temperature field has the following advantages:

[0053] 1. This invention acquires real-time temperature field distribution data of heat-treated workpieces through a temperature field monitoring network, and collects electrochemical noise signals of the workpieces through an electrochemical noise monitoring module. The two types of data are synchronized in time and registered in space using a data fusion and intelligent analysis center. After extracting temperature field features and electrochemical noise features, the probability of corrosion risk is calculated through an intelligent discrimination model. This establishes a real-time correlation model between the heat treatment temperature field and the corrosion resistance of the material, enabling early identification and accurate location of corrosion risks.

[0054] 2. This invention optimizes the sensor layout of the temperature field monitoring network, combines it with an electrode maintenance unit to ensure monitoring stability, and employs an improved intelligent discrimination model to enhance the correlation between temperature field characteristics and electrochemical noise characteristics. Combined with the feedback mechanism of the early warning module and the historical data archiving function of the data storage unit, this invention not only improves the accuracy and reliability of corrosion risk assessment, but also enables dynamic adjustment of heat treatment process parameters based on early warning results.

[0055] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description

[0056] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0057] Figure 1 This is a schematic diagram of the system composition and data flow of the present invention;

[0058] Figure 2 This is a schematic diagram of the intelligent discrimination and early warning logic of the present invention;

[0059] Figure 3 This is a schematic diagram of the operation process of the present invention. Detailed Implementation

[0060] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0061] Example 1

[0062] like Figure 1 and Figure 2 As shown in the illustration, this embodiment discloses a specific implementation of a corrosion resistance monitoring system for 304 austenitic stainless steel associated with a heat treatment temperature field. The system includes a temperature field monitoring network, an electrochemical noise monitoring module, and a data fusion and intelligent analysis center. The temperature field monitoring network acquires real-time temperature field distribution data of the heat-treated workpiece, while the electrochemical noise monitoring module synchronously collects electrochemical noise signals from the workpiece. The data fusion and intelligent analysis center performs time synchronization and spatial registration of the two types of data, extracts features, and calculates the corrosion risk probability using an improved BP neural network model. When the probability exceeds a preset threshold, an early warning signal is generated. This embodiment details the composition, connection relationships, and workflow of each module, clarifies the formula definitions and parameter meanings of the core algorithm, and ensures that the system can stably achieve real-time monitoring and early warning of corrosion risk during the heat treatment of 304 austenitic stainless steel, providing technical support for process optimization and safe equipment operation in industrial production.

[0063] The corrosion resistance monitoring system for 304 austenitic stainless steel in this embodiment, which is associated with the heat treatment temperature field, consists of a temperature field monitoring network, an electrochemical noise monitoring module, and a data fusion and intelligent analysis center. Both the temperature field monitoring network and the electrochemical noise monitoring module directly act on the heat-treated workpiece. They are connected to the data fusion and intelligent analysis center via communication links to achieve real-time data transmission. The data fusion and intelligent analysis center processes, analyzes, and judges the received data, ultimately outputting a corrosion risk warning signal. The entire system forms a closed-loop monitoring link, ensuring the continuity and accuracy of the monitoring process.

[0064] The temperature field monitoring network includes multiple K-type thermocouples, temperature data acquisition units, temperature field reconstruction modules, and calibration modules.

[0065] Type K thermocouples offer excellent temperature measurement stability and a wide measurement range, suitable for the temperature range of heat treatment of 304 austenitic stainless steel. All Type K thermocouples are directly in contact with the surface of the heat-treated workpiece, and are arranged using a non-uniform density design. Denser spacing is used in the weld area, heat-affected zone, and areas with large temperature gradients, while conventional spacing is used in areas with relatively uniform temperature distribution. This arrangement allows for accurate capture of temperature changes in critical areas of the workpiece, while balancing monitoring efficiency and data integrity.

[0066] The temperature data acquisition unit establishes an electrical connection with all K-type thermocouples and synchronously acquires the temperature readings of each thermocouple according to a preset sampling rate. The acquired temperature readings are transmitted to the temperature field reconstruction module through a standard communication interface, ensuring that there is no data loss or distortion during transmission.

[0067] After receiving temperature readings from the temperature data acquisition unit, the temperature field reconstruction module uses an interpolation algorithm to calculate the temperature value at each point on the surface of the heat-treated workpiece, thereby generating a two-dimensional temperature field distribution map. The interpolation algorithm uses known temperature values ​​at thermocouple measurement points to estimate the temperature in areas of the workpiece surface without thermocouples, ensuring that the temperature field distribution map fully reflects the temperature distribution on the workpiece surface. Based on the generated two-dimensional temperature field distribution map, the temperature field reconstruction module further calculates local temperature values, temperature gradient vectors, and temperature change rate scalars. The temperature gradient vector contains information on both the magnitude and direction of the gradient; the magnitude reflects the severity of temperature change at that point, and the direction points to the direction of the fastest temperature increase. The temperature change rate scalar reflects how quickly the temperature changes at the same location over time. The temperature field reconstruction module updates the temperature field distribution map in real time according to a set update frequency and sends the temperature field data, including local temperature values, temperature gradient vectors, and temperature change rate scalars, to the data fusion and intelligent analysis center.

[0068] The calibration module periodically calibrates all K-type thermocouples. During the calibration process, a standard temperature source is introduced, and the readings of each K-type thermocouple are compared with the standard values ​​of the standard temperature source. Based on the comparison results, the offset error of the thermocouples is adjusted to ensure that the measurement accuracy of the K-type thermocouples meets the system monitoring requirements and to avoid distortion of temperature field data due to thermocouple measurement errors.

[0069] The electrochemical noise monitoring module includes a three-electrode system, an electrochemical noise acquisition instrument, a signal preprocessing unit, and an electrode maintenance unit.

[0070] The three-electrode system consists of two identical 304 austenitic stainless steel working electrodes and a reference electrode. The spacing between the two working electrodes is controlled within a predetermined range to ensure effective capture of electrochemical signal changes on the workpiece surface. The three-electrode system forms a stable electrochemical connection with the heat-treated workpiece surface through an electrolyte. The electrolyte is selected to be compatible with the heat treatment conditions of 304 austenitic stainless steel, ensuring the stability of the electrochemical reaction and the effectiveness of the signal.

[0071] The electrochemical noise acquisition instrument is electrically connected to the three-electrode system, simultaneously measuring the current noise between the two working electrodes and the potential noise between the working electrode and the reference electrode. The sampling rate and bandwidth settings of the acquisition instrument meet the acquisition requirements of electrochemical noise signals, ensuring that the effective information in the original signal can be completely captured and avoiding signal loss or distortion due to improper sampling parameter settings.

[0072] After receiving the raw electrochemical noise signal transmitted from the electrochemical noise acquisition instrument, the signal preprocessing unit first performs digital filtering. The purpose of filtering is to remove high-frequency noise and power frequency interference from the raw signal. High-frequency noise mainly comes from environmental interference during the acquisition process, while power frequency interference is a common type of power supply interference. Digital filtering can significantly improve the signal-to-noise ratio, laying the foundation for subsequent feature extraction.

[0073] After filtering, the signal preprocessing unit extracts electrochemical noise features from the processed signal, including the standard deviation of current noise, the standard deviation of potential noise, noise resistance, wavelet energy entropy, and marginal spectral entropy from the Hilbert-Huang transform. The standard deviations of current noise and potential noise are statistical characteristics of the signal in the time domain, directly reflecting the fluctuations in current and potential signals. Noise resistance is calculated by the ratio of the standard deviation of potential noise to the standard deviation of current noise, reflecting the electrochemical corrosion state of the electrode surface. Wavelet energy entropy decomposes the signal into multiple frequency bands through wavelet transform, and then calculates the Shannon entropy of the energy values ​​of each band, used to characterize the complexity of the signal's frequency distribution. Marginal spectral entropy obtains the marginal spectrum of the signal through the Hilbert-Huang transform, and then calculates the Shannon entropy of the spectral values, effectively reflecting the nonlinear characteristics of the signal.

[0074] The electrode maintenance unit monitors the working status of each electrode in the three-electrode system in real time, including the cleanliness of the electrode surface and the contact status between the electrode and the electrolyte. When the electrode condition is detected to be inconsistent with the measurement requirements, an automatic reminder signal is issued, prompting the operator to replace or clean the electrode, ensuring that the electrode maintains a good working condition over a long period of time and guaranteeing the stability and accuracy of electrochemical noise measurement.

[0075] The data fusion and intelligent analysis center includes a data synchronization module, a feature extraction module, an intelligent discrimination module, an early warning module, and a data storage unit. These modules work together to complete data processing, feature extraction, risk discrimination, and early warning output.

[0076] The data synchronization module includes a time synchronization unit, a spatial registration unit, and a data verification subunit.

[0077] The time synchronization unit uses a high-precision clock source to timestamp the temperature field data transmitted by the temperature field monitoring network and the electrochemical noise data transmitted by the electrochemical noise monitoring module. The timestamp accuracy meets the system's requirements for data time synchronization, ensuring that the temperature field data and electrochemical noise data are strictly aligned on the time axis, providing a time dimension consistency guarantee for subsequent feature correlation analysis.

[0078] The spatial registration unit maps the position of each electrochemical noise measurement point to the corresponding coordinates on the surface of the heat-treated workpiece based on a preset geometric model of the heat-treated workpiece and sensor position information. Through coordinate mapping, it finds the nearest temperature field measurement point to the electrochemical noise measurement point, establishing a spatial relationship between them. When the positions of the temperature field measurement point and the electrochemical noise measurement point do not coincide, an interpolation algorithm is used to calculate the temperature field value at the electrochemical noise measurement point, ensuring that the temperature field characteristics and electrochemical noise characteristics remain consistent in spatial location, thus achieving spatial matching between the temperature field data and the electrochemical noise data.

[0079] The data verification subunit performs integrity and validity checks on the synchronized temperature field data and electrochemical noise data, detecting any data loss or anomalies. When data loss or anomalies are detected, a re-acquisition command is automatically triggered, controlling the temperature field monitoring network and electrochemical noise monitoring module to re-acquire the corresponding data, ensuring that the data entering subsequent processing stages is complete and valid.

[0080] Through time synchronization, spatial registration, and data verification, the data synchronization module ultimately generates synchronized data that is aligned in both time and space, and then transmits the synchronized data to the feature extraction module.

[0081] The feature extraction module includes a temperature field feature extraction unit and an electrochemical noise feature extraction unit.

[0082] The temperature field feature extraction unit extracts temperature field features from the synchronous data, specifically including local temperature values, temperature gradient magnitude, and temperature change rate. Local temperature values ​​are the temperature values ​​at the electrochemical noise measurement points, directly reflecting the temperature state at that measurement location; the temperature gradient magnitude is obtained by calculating the temperature difference between adjacent points in the temperature field distribution, reflecting the severity of temperature changes at that location; and the temperature change rate is obtained by calculating the temperature change over time at the same location, reflecting the dynamic trend of temperature change.

[0083] The electrochemical noise feature extraction unit extracts electrochemical noise features from the filtered electrochemical noise signal transmitted from the signal preprocessing unit, including current noise standard deviation, noise resistance, wavelet energy entropy, and marginal spectral entropy. The extraction method for each feature is consistent with the feature definition in the signal preprocessing unit, further ensuring the accuracy and effectiveness of the features.

[0084] After feature extraction is completed, the feature extraction module normalizes all extracted temperature field features and electrochemical noise features. The purpose of normalization is to eliminate the influence of different dimensions between features, so that all features are within the same numerical range. This avoids deviations in the training and prediction of the intelligent discrimination model due to differences in feature dimensions, and can significantly improve the convergence speed of the model.

[0085] The intelligent discrimination module uses an improved BP neural network model to calculate the corrosion risk probability based on the temperature field features and electrochemical noise features output by the feature extraction module.

[0086] The improved BP neural network model includes an input layer, a hidden layer, and an output layer. The number of nodes in the input layer is the same as the number of extracted features, and each input layer node corresponds to one feature. The hidden layer is set to at least one layer, and the number of nodes in the hidden layer is determined through model training optimization. The output layer has 1 node and directly outputs the corrosion risk probability, with the probability value ranging from 0 to 1.

[0087] Hidden layer nodes use a custom activation function, the specific expression of which is as follows:

[0088]

[0089] in, The input values ​​for the hidden layer nodes are the weighted sum of the outputs of the nodes in the previous layer. This is a scaling factor for the activation function, used to adjust the overall magnitude of the function's output. Through training, the function's output is made to fit the range of the model's input data. The shape parameter of the activation function is used to adjust the shape of the function curve and enhance the function's ability to fit nonlinear features. The slope parameter of the activation function is used to adjust the rate of change of the function in different input intervals, thereby improving the model's ability to capture feature details. , , All parameters were determined through training with a large number of samples. During the training process, the parameter values ​​were continuously adjusted based on the prediction error of the model until the model achieved the best fit.

[0090] The output layer nodes use the sigmoid function to map the final output of the model to a probability range of 0 to 1, directly representing the corrosion risk probability of 304 austenitic stainless steel.

[0091] The improved BP neural network model is trained using a custom loss function, the specific expression of which is as follows:

[0092]

[0093] in, This is the loss function value, used to measure the deviation between the model's predicted values ​​and the actual values; To determine the number of training samples, the training samples were obtained by collecting the temperature field characteristics, electrochemical noise characteristics, and corresponding actual corrosion risk labels of 304 austenitic stainless steel under different heat treatment conditions. For the first The true corrosion risk label for each sample is determined based on the actual corrosion test results. For the first The predicted corrosion risk probability for each sample is obtained from the model output; This is the L2 regularization parameter, used to suppress model overfitting. It controls the magnitude of model weights to avoid excessive dependence of the model on training samples. For the first Each model weight value covers all connection weights between the input layer and the hidden layer, and between the hidden layer and the output layer; To establish a correlation regularization parameter, which is used to strengthen the correlation between temperature field features and electrochemical noise features, ensuring that the model fully considers the synergistic effect of the two types of features when calculating the corrosion risk probability; For the first A temperature field characteristic value; For the first One electrochemical noise characteristic value; The number of temperature field characteristics; This represents the number of electrochemical noise characteristics.

[0094] During model training, the loss function value is continuously minimized using gradient descent, and the model's weight and activation function parameters are adjusted until the model's prediction error reaches the preset requirements. After training is complete, the intelligent discrimination module will also periodically update the model parameters based on newly collected historical data, enabling the model to adapt to changes in heat treatment processes and ensuring discrimination accuracy during long-term monitoring.

[0095] The early warning module includes a threshold comparison unit, an alarm generation unit, a communication interface, and a log recording unit.

[0096] The threshold comparison unit compares the corrosion risk probability output by the intelligent discrimination module with a preset threshold in real time. The preset threshold is set according to the material characteristics and heat treatment process requirements of 304 austenitic stainless steel, and its value ranges from 0.5 to 0.9. It can be dynamically adjusted according to the safety requirements of the actual application scenario.

[0097] When the probability of corrosion risk exceeds a preset threshold, the alarm generation unit immediately generates a corrosion risk warning signal and triggers the audible and visual alarm device. The audible and visual alarm device includes a buzzer, LED indicator lights, and a display screen. The buzzer emits a continuous alarm sound, the LED indicator lights flash continuously, and the display screen shows the location of the corrosion risk and the corresponding corrosion risk probability value in real time, so that operators can quickly locate the risk area.

[0098] The communication interface transmits warning signals to external devices, including the heat treatment control system, central monitoring room, or mobile terminals, via wired or wireless means. Warning signals sent to the heat treatment control system can trigger automatic adjustments to process parameters, while warning signals sent to the central monitoring room and mobile terminals can promptly remind operators to take corrective measures to prevent further corrosion risks.

[0099] The log recording unit meticulously records the occurrence time of each warning event, the risk location of the heat-treated workpiece, the corrosion risk probability value, and the corresponding temperature field characteristics and electrochemical noise characteristics. Based on these records, detailed warning reports are generated for historical data retrieval and corrosion risk trend analysis, providing data support for subsequent process optimization.

[0100] The early warning module also integrates a feedback mechanism, which automatically adjusts the heat treatment process parameters, including temperature setpoint and holding time, based on the early warning results and historical monitoring data. By optimizing the process parameters, the corrosion risk of 304 austenitic stainless steel is reduced, and a closed-loop linkage between monitoring and control is achieved.

[0101] The data storage unit archives and stores all relevant data, including temperature field data, electrochemical noise data, synchronization data, extracted feature data, model parameters, and early warning records. The stored data is used to support the retraining and parameter updates of the intelligent discrimination model, and to provide a data foundation for the trend analysis of corrosion risks, ensuring the traceability and continuous optimization capabilities of the system.

[0102] In summary, this embodiment details the implementation of a corrosion resistance monitoring system for 304 austenitic stainless steel correlated with the heat treatment temperature field. It achieves real-time and accurate monitoring of the workpiece temperature field distribution through a temperature field monitoring network, captures changes in electrochemical signals on the workpiece surface through an electrochemical noise monitoring module, and then completes data synchronization, feature extraction, risk assessment, and early warning output through a data fusion and intelligent analysis center. The system employs an improved BP neural network model, using custom activation and loss functions to enhance the model's ability to capture the correlation between temperature field features and electrochemical noise features, thereby improving the accuracy of corrosion risk probability calculation.

[0103] Example 2

[0104] like Figure 3 As shown in this embodiment, the specific workflow of the 304 austenitic stainless steel corrosion resistance monitoring system correlated with the heat treatment temperature field during actual operation is illustrated. This system achieves real-time assessment and early warning of material corrosion risk during heat treatment by coordinating temperature field monitoring and electrochemical noise monitoring. The specific steps are as follows:

[0105] 1. System Startup and Initialization Configuration: First, start the temperature field monitoring network and the electrochemical noise monitoring module. The system executes a self-test procedure to test the communication connection of all K-type thermocouples in the network and to check the status of the working electrode and reference electrode of the three-electrode system. The calibration module calls a standard temperature source to perform zero-point and range calibration on the thermocouples to ensure temperature measurement accuracy. The electrode maintenance unit checks the electrolyte connectivity and electrode surface condition to confirm that the monitoring conditions are ready.

[0106] 2. Temperature Field Data Acquisition and Processing: Multiple K-type thermocouples arranged on the surface of the heat-treated workpiece begin synchronously acquiring temperature signals, with a sampling rate of no less than 1Hz. The temperature data acquisition unit collects the readings of all thermocouples and transmits them in real time to the temperature field reconstruction module via a communication interface. Based on the received discrete temperature point data, this module uses a spatial interpolation algorithm to calculate and generate a two-dimensional temperature field distribution map covering the entire workpiece surface. It further analyzes the local temperature value, the magnitude and direction of the temperature gradient vector, and the rate of temperature change over time at each measurement point.

[0107] 3. Electrochemical Noise Signal Acquisition and Preprocessing: The electrochemical noise acquisition instrument is started synchronously, and at a sampling rate of not less than 1000Hz, it simultaneously measures the current noise signal between the two working electrodes and the potential noise signal between the working electrode and the reference electrode. The acquired raw signals are sent to the signal preprocessing unit, which first performs digital filtering to suppress high-frequency random noise and 50Hz power frequency interference. Then, it extracts time-domain and frequency-domain features from the purified signals, including the standard deviation of current noise, the standard deviation of potential noise, noise resistance, energy entropy based on wavelet transform, and marginal spectral entropy based on Hilbert-Huang transform.

[0108] 4. Multi-source Data Synchronization and Spatial Registration: The data synchronization module receives temperature field data and electrochemical noise data from the preceding steps. The time synchronization unit uses a high-precision clock source to stamp all data packets with a uniform timestamp with an accuracy of no less than 1 microsecond. The spatial registration unit, based on the pre-input workpiece geometric model and sensor layout coordinates, precisely locates each electrochemical noise measurement point on the workpiece surface and associates it with the nearest neighbor temperature field measurement point. For non-coincident points, the corresponding temperature field characteristic values ​​are estimated using an interpolation algorithm, ultimately forming a time-aligned and spatially matched synchronized dataset.

[0109] 5. Feature Extraction and Normalization: The feature extraction module parses the synchronous dataset. The temperature field feature extraction unit obtains the local temperature, temperature gradient magnitude, and temperature change rate of each associated point. The electrochemical noise feature extraction unit reads the preprocessed current noise standard deviation, noise resistance, wavelet energy entropy, and marginal spectral entropy. Subsequently, this module normalizes all extracted feature data to eliminate numerical differences caused by different physical dimensions of the features, preparing for subsequent model analysis.

[0110] 6. Intelligent Corrosion Risk Probability Judgment: The intelligent judgment module combines the normalized temperature field features and electrochemical noise features into a feature vector, which is then input into a pre-trained improved BP neural network model. This model performs nonlinear mapping and calculation on the input features, ultimately generating a corrosion risk probability value between 0 and 1 at the output layer. This value quantitatively represents the likelihood of corrosion occurring at the current monitoring point.

[0111] 7. Warning Threshold Comparison and Decision-Making: The threshold comparison unit in the warning module receives the corrosion risk probability value output by the intelligent discrimination module in real time and compares it with the preset risk probability threshold. This threshold is set in the range of 0.5 to 0.9 based on the material characteristics of 304 austenitic stainless steel and specific heat treatment process requirements, and can be dynamically adjusted according to actual working conditions.

[0112] 8. Warning Signal Generation and Output: When the corrosion risk probability value at any monitoring point exceeds the preset threshold, the alarm generation unit is immediately triggered. This unit generates a warning signal containing the risk location and probability value, and drives the audible and visual alarm device to issue an alarm. Simultaneously, the communication interface sends detailed warning information via wired or wireless network to the heat treatment process control system, the central monitoring room's operating interface, or the mobile terminals of relevant personnel.

[0113] 9. Data Archiving and Early Warning Log Recording: The log recording unit archives key information about this early warning event, including the timestamp of the event, the spatial location of the risk point, the corrosion risk probability value, and the temperature field and electrochemical noise characteristics corresponding to the triggering of the early warning. This data is stored in the data storage unit to generate historical early warning reports, supporting subsequent data traceability and corrosion trend analysis.

[0114] 10. System Maintenance and Adaptive Model Updates: During continuous system operation, the calibration module and electrode maintenance unit perform scheduled maintenance tasks according to a predetermined plan or triggering conditions to ensure the long-term stability and accuracy of sensor measurements. Simultaneously, the intelligent discrimination module utilizes historical operating data accumulated in the data storage unit to periodically retrain and update the parameters of the improved BP neural network model, enabling the model to adapt to the slow changes in heat treatment processes and maintain the accuracy of corrosion risk discrimination.

[0115] 11. Process Feedback Control: The feedback mechanism integrated into the early warning module can automatically send process parameter adjustment suggestions or instructions to the heat treatment control system after generating an early warning, based on the preset control strategy. For example, it can fine-tune the furnace temperature setting or modify the holding time, in order to proactively reduce corrosion risks from the process level and form a closed-loop management of monitoring-early warning-control.

[0116] Through the aforementioned sequential steps, the system enables real-time, online monitoring and intelligent early warning of the correlation between temperature field and corrosion state during the heat treatment of 304 austenitic stainless steel.

[0117] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A system for monitoring the corrosion resistance of 304 austenitic stainless steel in relation to a heat treatment temperature field, characterized in that, The temperature field monitoring network, the electrochemical noise monitoring module and the data fusion and intelligent analysis center are included. The temperature field monitoring network is used for monitoring the temperature field distribution of the heat treatment workpiece in real time and generating temperature field data. The electrochemical noise monitoring module is used for monitoring the electrochemical noise signal of the heat treatment workpiece in real time and generating electrochemical noise data. The data fusion and intelligent analysis center is connected with the temperature field monitoring network and the electrochemical noise monitoring module, used for receiving the temperature field data and the electrochemical noise data, performing time synchronization and space registration on the temperature field data and the electrochemical noise data to form synchronous data, extracting temperature field features and electrochemical noise features from the synchronous data, calculating a corrosion risk probability based on the temperature field features and the electrochemical noise features through an intelligent discrimination model, and generating a corrosion risk warning signal when the corrosion risk probability exceeds a preset threshold.

2. The system for monitoring the corrosion resistance of 304 austenitic stainless steel in relation to a thermal treatment temperature field according to claim 1, characterized in that, The temperature field monitoring network includes a plurality of thermocouples, a temperature data acquisition unit and a temperature field reconstruction module. The plurality of thermocouples are all K-type thermocouples, which directly contact the surface of the heat treatment workpiece and are arranged on the heat treatment workpiece at a non-uniform density, wherein the K-type thermocouples are arranged at a high density in the weld area, the heat affected zone and the area with large temperature gradient of the heat treatment workpiece. The temperature data acquisition unit is configured to synchronously acquire temperature readings of all the K-type thermocouples and transmit the temperature readings to the temperature field reconstruction module through a communication interface. The temperature field reconstruction module is configured to calculate the temperature value of each point on the surface of the heat treatment workpiece based on the temperature readings by using an interpolation algorithm to generate a two-dimensional temperature field distribution map and calculate the local temperature value, the temperature gradient vector and the temperature change rate scalar in the temperature field distribution map. The temperature gradient vector is represented by the gradient size and direction, and the temperature change rate scalar is represented by the change rate of the temperature with time, and the temperature field reconstruction module is further configured to update the temperature field distribution map in real time and send the temperature field data to the data fusion and intelligent analysis center. The temperature field monitoring network further includes a calibration module for periodically calibrating the K-type thermocouples, and the calibration module adjusts the offset error by comparing the standard temperature source with the K-type thermocouple readings.

3. The system for monitoring corrosion resistance of 304 austenitic stainless steel in relation to a thermal treatment temperature field according to claim 1, characterized in that, The electrochemical noise monitoring module includes a three-electrode system, an electrochemical noise acquisition instrument and a signal preprocessing unit. The three-electrode system includes two 304 austenitic stainless steel working electrodes of the same material and one reference electrode, the spacing between the two working electrodes is set to be in the range of 1mm to 5mm, and the three-electrode system forms an electrochemical connection with the surface of the heat treatment workpiece through an electrolyte. The electrochemical noise acquisition instrument is configured to synchronously measure the current noise between the two working electrodes and the potential noise between the working electrode and the reference electrode. The signal preprocessing unit is configured to perform digital filtering processing on the original electrochemical noise signal to remove high-frequency noise and 50Hz power frequency interference, and extract electrochemical noise features, the electrochemical noise features including current noise standard deviation, potential noise standard deviation, noise resistance, wavelet energy entropy and marginal spectrum entropy of Hilbert-Huang transform. The wavelet energy entropy is a Shannon entropy of energy values of frequency bands obtained by decomposing a signal into the frequency bands through wavelet transform, and the marginal spectrum entropy is a Shannon entropy of spectrum values obtained by calculating a marginal spectrum of the signal through Hilbert-Huang transform. The electrochemical noise monitoring module further comprises an electrode maintenance unit configured to monitor the state of the electrode and automatically remind to replace or clean the electrode.

4. The system for monitoring the corrosion resistance of 304 austenitic stainless steel in relation to a thermal treatment temperature field according to claim 1, characterized in that, The data fusion and intelligent analysis center comprises a data synchronization module, a feature extraction module, an intelligent discrimination module and an early warning module. The data synchronization module is configured to mark the temperature field data and the electrochemical noise data with a unified timestamp and perform spatial registration, so as to associate each electrochemical noise measurement point with the nearest temperature field measurement point and form synchronized data. The feature extraction module is configured to extract temperature field features and electrochemical noise features from the synchronized data, wherein the temperature field features comprise local temperature values, temperature gradient sizes and temperature change rates, and the electrochemical noise features comprise current noise standard deviation, noise resistance, wavelet energy entropy and marginal spectrum entropy. The intelligent discrimination module is configured to calculate a corrosion risk probability based on the temperature field features and the electrochemical noise features using an improved BP neural network model, wherein the improved BP neural network model comprises an input layer, at least one hidden layer and an output layer, the number of nodes of the input layer corresponds to the number of features, and the number of nodes of the output layer is 1, indicating the corrosion risk probability. The early warning module is configured to generate an early warning signal and trigger an audible and light alarm when the corrosion risk probability exceeds a preset threshold, and send early warning information to an external display or control system. The data fusion and intelligent analysis center further integrates a data storage unit configured to archive historical data to support model retraining and trend analysis.

5. The system for monitoring the corrosion resistance of 304 austenitic stainless steel in relation to the heat treatment temperature field according to claim 4, characterized in that, The data synchronization module comprises a time synchronization unit and a spatial registration unit. The time synchronization unit is configured to mark the temperature field data and the electrochemical noise data with a timestamp using a high-precision clock source. The spatial registration unit is configured to map the position of each electrochemical noise measurement point to the surface coordinates of the heat-treated workpiece based on a heat-treated workpiece geometric model and sensor position information, find the nearest temperature field measurement point to the point, and establish a spatial association between the electrochemical noise measurement point and the temperature field measurement point. The spatial registration unit is further configured to calculate the temperature field value at the electrochemical noise measurement point through an interpolation algorithm when the temperature field measurement point and the electrochemical noise measurement point do not coincide. The data synchronization module further comprises a data verification subunit configured to detect data loss or abnormality and trigger reacquisition when detecting data loss or abnormality.

6. The system for monitoring the corrosion resistance of 304 austenitic stainless steel in relation to a thermal treatment temperature field according to claim 4, characterized in that, The feature extraction module comprises a temperature field feature extraction unit and an electrochemical noise feature extraction unit. The temperature field feature extraction unit is configured to extract local temperature values, temperature gradient vectors and temperature change rate scalars from the synchronized data, wherein the local temperature values are temperature values at the electrochemical noise measurement points, the temperature gradient vectors are obtained by calculating the temperature difference between adjacent points in the temperature field distribution, and the temperature change rate scalars are obtained by calculating the temperature change rate at the same position over time. The electrochemical noise feature extraction unit is configured to extract current noise standard deviation, potential noise standard deviation, noise resistance, wavelet energy entropy and marginal spectrum entropy from the filtered electrochemical noise signal, wherein the current noise standard deviation and the potential noise standard deviation are standard deviations of the signal in the time domain, the noise resistance is a ratio of the potential noise standard deviation to the current noise standard deviation, the wavelet energy entropy is a Shannon entropy of energy values of sub-bands obtained by decomposing the electrochemical noise signal into the sub-bands of multiple scales through wavelet transform, and the marginal spectrum entropy is a Shannon entropy of a marginal spectrum obtained by performing Hilbert-Huang transform on the electrochemical noise signal to obtain a Hilbert spectrum of the signal. The feature extraction module is further configured to perform normalization processing on the extracted features to eliminate dimensional influence.

7. The system for monitoring the corrosion resistance of 304 austenitic stainless steel in relation to a thermal treatment temperature field according to claim 4, characterized in that, The intelligent discrimination module uses an improved BP neural network model to calculate the corrosion risk probability, the improved BP neural network model includes an input layer, a hidden layer and an output layer, wherein the input layer nodes receive the temperature field features and the electrochemical noise features, the hidden layer nodes use the following self-defined activation function: wherein is an input value of a hidden layer node, is a scaling factor of an activation function, is a shape parameter of an activation function, is a slope parameter of an activation function, which are obtained through training; The output layer nodes of the improved BP neural network model use a sigmoid function to output the corrosion risk probability. The training of the improved BP neural network model uses the following self-defined loss function: in It is the value of the loss function. It is the number of training samples. It is the first The true corrosion risk label for each sample It is the first The predicted corrosion risk probability for each sample. It is the L2 regularization parameter. It is the first Each model weight value, It is the association regularization parameter. It is the first A temperature field characteristic value, It is the first One electrochemical noise characteristic value, It is the quantity of temperature field characteristics. It is the number of electrochemical noise features; The intelligent discrimination module is further configured to periodically update the model parameters to adapt to process changes.

8. The system for monitoring the corrosion resistance of 304 austenitic stainless steel in relation to a thermal treatment temperature field according to claim 4, characterized in that, The early warning module includes a threshold comparison unit, an alarm generation unit, a communication interface and a log recording unit. The threshold comparison unit is configured to compare the corrosion risk probability calculated by the intelligent discrimination module with a preset threshold, the preset threshold is set to be in the range of 0.5 to 0.9 according to the material and heat treatment process requirements of 304 austenitic stainless steel, and can be dynamically adjusted. The alarm generation unit is configured to generate a corrosion risk early warning signal when the corrosion risk probability exceeds the preset threshold, and trigger an audible and light alarm device, the audible and light alarm device includes a buzzer, an LED indicator and a display screen, wherein the display screen displays the risk position and the probability value. The communication interface is configured to send the early warning signal to external devices including a heat treatment control system, a central monitoring room or a mobile terminal through wired or wireless means, so that the operator can take timely corrective measures; The log recording unit is configured to record the occurrence time of each early warning event, the heat treatment workpiece position, the corrosion risk probability value, and the corresponding temperature field and electrochemical noise feature values, and generate a detailed early warning report for historical data query and trend analysis.

9. The system for monitoring corrosion resistance of 304 austenitic stainless steel in relation to a heat treatment temperature field according to claim 8, characterized in that, The early warning module further integrates a feedback mechanism for automatically adjusting the heat treatment process parameters according to the early warning results.

Images