A quenching and cooling system and its control method for hydrogen-induced cracking resistant subsea pipeline steel

Abstract

This invention discloses a quenching and cooling system and its control method for hydrogen-induced cracking resistant subsea pipeline steel, belonging to the field of metallurgical automatic control. The system includes an ultra-fast cooling execution unit, a multi-source sensing detection unit, and an intelligent control unit containing an immune memory bank, clonal selection optimization, and negative selection monitoring modules. The immune memory bank stores cooling parameter memory antibodies; the clonal selection module uses the antibody with the highest affinity as the initial solution, and after clonal amplification and mutation, selects the optimal cooling parameters for execution; the negative selection monitoring module matches the operating conditions in real time and issues alarms. After cooling, HIC testing is performed according to the NACETM0284-2016 standard, and the results are fed back to update the memory bank. After cooling, the steel plate yield strength is ≥485MPa, tensile strength is ≥570MPa, yield ratio is ≤0.86, CLR is ≤5%, CTR is ≤1.5%, and CSR is ≤0.5%.

Description

Technical Field

[0001] This invention belongs to the field of metallurgical automatic control and intelligent optimization control technology, specifically relating to a quenching and cooling system and control method for hydrogen-induced cracking subsea pipeline steel. Background Technology

[0002] As offshore oil and gas resource development continues to extend into deeper and more open seas, the safety and reliability requirements for subsea pipelines, as key infrastructure for marine energy transportation, are increasingly stringent. Deep-sea oil and gas transportation media often contain... Submarine pipeline steel in acidic corrosive media, wet Submarine pipelines face the risk of damage from hydrogen-induced cracking (HIC) and sulfide stress cracking (SSCC) in the environment. To meet the requirements of deep-sea, long-distance, high-pressure, and high-flow oil and gas transportation, it is of great significance to develop submarine pipeline steel with high strength, high toughness, and excellent resistance to HIC.

[0003] The quenching and cooling process is a crucial step in determining the final microstructure and properties of subsea pipeline steel. Post-rolling accelerated cooling processes, exemplified by ultrafast cooling technology, rapidly cool the steel plate to a final cooling temperature range of 450–600°C at a cooling rate of 15–40°C / s. This yields a multiphase microstructure consisting of acicular ferrite, fine-grained ferrite, and dispersed martensite-austenite (MA) components, achieving a good balance between high strength and excellent resistance to high thermal osmosis (HIC). However, the ultrafast quenching process exhibits complex dynamic characteristics characterized by strong nonlinearity, large hysteresis, and multivariate coupling—including transient heat conduction along the thickness of the steel plate, drastic temperature variations in the heat transfer coefficient between the steel plate surface and the cooling medium, and the nonlinear influence of the cooling rate and final cooling temperature on the final microstructure. These factors make precise control of cooling parameters extremely difficult.

[0004] In the field of intelligent control, various methods have been applied to temperature control in heat treatment processes. For example, fuzzy inference is used to achieve online adjustment of the quenching medium temperature; offline genetic optimization is used to determine the heating curve; and particle swarm optimization is used to statically optimize the quenching cooling rate. However, these methods mostly focus on parameter optimization or static modeling of a single stage, and are all offline optimization or open-loop control, lacking a complete intelligent closed-loop control architecture that can integrate the reuse of historical production experience, real-time adaptive online optimization, and rapid identification of abnormal conditions. In particular, none of the existing methods have established a closed-loop feedback optimization mechanism between cooling parameters and final service performance for the hydrogen-induced cracking (HIC) resistance performance of subsea pipeline steel.

[0005] Furthermore, there are no reports in the existing technology of applying intelligent control methods based on immune memory and adaptive optimization to the real-time control of the quenching and cooling process of steel for subsea pipelines. In existing quenching and cooling control technologies, successful cooling experience accumulated in historical production processes mainly relies on manual recording and inheritance, lacking a systematic knowledge storage and automatic reuse mechanism; the determination of cooling parameters for new steel grades or new specifications of products still mainly relies on experience-based trial and error or offline testing, resulting in low optimization efficiency and long cycles; there is a lack of automatic identification and early warning capabilities for abnormal operating conditions during the cooling process (such as water volume fluctuations, abnormal water temperature, and steel plate temperature deviations), and the detection of anomalies depends on manual inspection and experience judgment by operators. Summary of the Invention

[0006] To address the technical problems in existing technologies, such as the reliance on manual recording and inheritance of cooling experience lacking a systematic storage and automatic reuse mechanism; the determination of cooling parameters for new steel grades and specifications mainly relying on experience-based trial and offline testing leading to low optimization efficiency and long cycles; the reliance on manual inspection and experience-based judgment for the discovery of abnormal operating conditions during quenching and cooling, lacking automatic identification and real-time early warning capabilities; and the lack of a closed-loop feedback optimization mechanism between cooling parameters and final service performance for hydrogen-induced cracking resistance performance indicators of subsea pipeline steel, this invention provides a quenching and cooling system and its control method for hydrogen-induced cracking resistant subsea pipeline steel.

[0007] The technical solution adopted by the present invention to solve the above problems is as follows: A quenching and cooling system for submarine pipeline steel resistant to hydrogen-induced cracking includes an ultra-fast cooling actuator, a multi-source sensing and detection unit, and an intelligent control unit.

[0008] The ultra-fast cooling unit includes an upper manifold assembly, a lower manifold assembly, and a side spray device. Each manifold is equipped with an independent flow regulating valve, with a cooling rate adjustment range of 15–40℃ / s and a final cooling temperature adjustment range of 450–600℃.

[0009] The multi-source sensing detection unit includes a steel plate temperature measuring instrument installed at the ultra-fast cooling inlet, a steel plate temperature measuring instrument installed at the ultra-fast cooling outlet, and flow sensors and water temperature sensors installed at each manifold.

[0010] The intelligent control unit is connected to the ultrafast cold execution unit and the multi-source sensing detection unit, and is equipped with an immune memory bank module, a clonal selection optimization module and a negative selection monitoring module.

[0011] The immune memory bank module stores at least one cooling parameter memory antibody corresponding to the steel composition identifier and finished product specification of the subsea pipeline steel resistant to hydrogen-induced cracking. Each cooling parameter memory antibody includes the steel composition identifier, finished product specification, cooling rate, final cooling temperature, and corresponding hydrogen-induced cracking resistance performance index.

[0012] The clonal selection optimization module selects the memory antibody with the highest affinity from the immune memory bank module as the initial solution based on the steel grade composition identifier and finished product specifications of the current steel plate. After clonal amplification and mutation operations on the initial solution, a candidate antibody population is generated. The optimal antibody is screened from the candidate antibody population using the steel plate thickness direction temperature uniformity index and the final cooling temperature deviation index as fitness indicators. The affinity is determined by a weighted sum of the steel grade composition matching item and the finished product specification matching item. When the steel grade composition identifier matches, the matching item receives a full score; otherwise, it receives zero. When the finished product specification matches completely, the matching item receives a full score; if the thickness deviation is within the set allowable range, it receives half a score; otherwise, it receives zero. Furthermore, the weight of the steel grade composition matching item is higher than that of the finished product specification matching item. When there is no record in the immune memory bank that matches both the steel grade composition identifier and the finished product specification, the clonal selection optimization module generates initial cooling parameters based on a preset expert experience rule base and increases the size of the initial antibody population from twenty to thirty. During clonal expansion, the clone size is positively correlated with the affinity value of the initial solution; the mutation operation is performed within a range of ±30% of the cooling rate corresponding to the initial solution and ±50 degrees Celsius of the final cooling temperature.

[0013] The negative selection monitoring module collects the detection signals from the multi-source sensing unit in real time and matches the detection signals with a preset normal operating condition pattern set. When the matching degree between the detection signal and the normal operating condition pattern set is lower than a preset threshold, an alarm signal is output. The normal operating condition pattern set is a set of patterns constructed based on historical normal operating data using a clustering algorithm. Each pattern in the pattern set contains feature vectors of inlet temperature, flow rate of each manifold, water temperature, and outlet temperature. The normal operating condition pattern set is constructed by clustering historical normal operating data using the K-means clustering algorithm, and the number of clusters k is determined using the elbow rule. Each pattern vector in the pattern set is a four-dimensional feature vector obtained by performing minimum-maximum normalization processing on the inlet temperature, flow rate of each manifold, water temperature, and outlet temperature.

[0014] The ultrafast cooling execution unit performs the cooling operation based on the cooling rate and final cooling temperature corresponding to the optimal antibody output by the clone selection optimization module.

[0015] This invention also provides a method for controlling the quenching and cooling of steel for hydrogen-induced cracking subsea pipelines, employing the quenching and cooling system described above, and including the following steps: Step 1: Obtain the steel grade composition identification and finished product specifications of the steel plate to be cooled.

[0016] Step two: Retrieve cooling parameter memory antibodies from the immune memory bank that match the current steel grade composition identifier and finished product specifications. Determine the affinity of each candidate antibody based on whether the current steel grade composition identifier matches the steel grade composition identifier in the memory antibody, whether the current finished product specifications match the finished product specifications in the memory antibody, and whether the thickness deviation is within a set range. The affinity is highest when the steel grade composition identifier and finished product specifications match perfectly, followed by the affinity when the steel grade composition identifiers do not match but the finished product specifications match perfectly. When the thickness deviation of the finished product specifications is within the set allowable range (i.e., the thickness deviation does not exceed 10%), the finished product specification matching item takes half the score; if it exceeds the set allowable range, it takes zero. Select the memory antibody with the highest affinity as the initial cooling parameter.

[0017] Step 3: Using the initial cooling parameters as the center, generate the initial antibody population within a cooling rate range of ±30% and a final cooling temperature range of ±50 degrees Celsius.

[0018] Step four involves cloning and amplifying each antibody in the initial antibody population. The number of clones is positively correlated with the antibody's affinity value; antibodies with higher affinity values ​​yield more clones. The cloned antibodies are then subjected to mutation operations within a cooling rate range of ±30% and a final cooling temperature range of ±50 degrees Celsius to generate candidate antibody populations.

[0019] Step 5: Using the thickness-direction temperature uniformity index and the final cooling temperature deviation index as fitness indices, calculate the fitness value of each antibody in the candidate antibody population. The thickness-direction temperature uniformity index is the standard deviation of the temperature at each node along the thickness direction of the steel plate, calculated by solving the temperature field distribution along the thickness direction of the steel plate using the finite difference method based on the one-dimensional transient heat conduction equation. The final cooling temperature deviation index is the absolute value of the difference between the measured temperature at the steel plate outlet and the target final cooling temperature. The fitness value is negatively correlated with the thickness-direction temperature uniformity index and also negatively correlated with the final cooling temperature deviation index.

[0020] Step 6: Select the optimal antibody from the candidate antibody population based on the fitness value.

[0021] Step 7: Using the cooling rate and final cooling temperature corresponding to the optimal antibody as control targets, adjust the flow rate of each manifold of the ultra-fast cooling execution unit to cool the steel plate.

[0022] Step 8: During the cooling process, inlet temperature, flow rate of each manifold, water temperature, and outlet temperature are collected in real time. The collected data is normalized to form a real-time feature vector. The distance between this feature vector and the center vectors of each category in the normal operating condition pattern set is calculated as the matching degree. Cooling continues when the matching degree is less than or equal to a preset threshold; an alarm signal is output when the matching degree is greater than the preset threshold. The matching degree uses Euclidean distance as a metric, and the preset threshold is the maximum Euclidean distance from the pattern vector within each category to the center vector of that category. The method for constructing the normal operating condition pattern set includes: collecting multiple sets of inlet temperature, flow rate of each manifold, water temperature, and outlet temperature data under historical normal operating conditions; normalizing each set of data to form a pattern vector; using a clustering algorithm to divide the pattern vectors into several categories; and the center vectors of each category constitute the normal operating condition pattern set.

[0023] Step nine: Test the hydrogen-induced cracking resistance of the cooled steel plate and obtain the crack length ratio, crack width ratio and crack sensitivity ratio according to the American Society of Corrosion Engineers standard TM0284-2016.

[0024] Step 10: Update the immune memory library with the optimal antibody obtained from this cooling process, along with the crack length rate, crack width rate, and crack sensitivity rate, as the antibody's hydrogen-induced cracking resistance performance indicators. During the update, if a memory antibody with the same steel composition identifier and finished product specification already exists in the immune memory library, compare the hydrogen-induced cracking resistance performance indicators. If the crack length rate, crack width rate, and crack sensitivity rate of the current data are all no higher than the existing records, then replace the existing records with the current data; otherwise, retain the existing records.

[0025] Compared with the prior art, the present invention has the following advantages: Compared with existing technologies, this invention constructs a dedicated intelligent quenching cooling control architecture for HIC resistance of subsea pipeline steel. It stores successful cooling parameters and their corresponding NACETM0284-2016 standard HIC resistance performance indicators (crack length ratio (CLR), crack width ratio (CTR), and crack sensitivity ratio (CSR)) under different steel compositions and finished product specifications in the form of structured memory antibodies through an immune memory library. This achieves systematic accumulation and automatic reuse of cooling experience, avoiding the inefficient traditional method of relying on manual recording and transmission. The clonal selection optimization module uses high-affinity antibodies from the immune memory library as initial solutions and performs clonal amplification and mutation operations within a defined range of ±30% cooling rate and ±50℃ final cooling temperature. It can search for the optimal combination of cooling parameters suitable for the current working conditions within milliseconds, meeting the stringent real-time requirements of ultrafast quenching processes. Compared with traditional optimization methods relying on experience-based trial and error or offline experiments, this significantly shortens the time required for optimization. The invention establishes a cycle for determining cooling parameters for new steel grades and specifications. The negative selection monitoring module constructs a normal operating condition mode set based on historical normal operation data. During the cooling process, it compares the Euclidean distance matching degree between multi-dimensional sensor features such as inlet temperature, flow rate of each manifold, water temperature, and outlet temperature and the normal mode in real time. This enables rapid identification and output of alarm signals when cooling anomalies occur, such as water flow fluctuations, abnormal water temperature, and steel plate temperature deviations, overcoming the lag problem of traditional anomaly detection relying on manual inspection. In particular, after cooling, the invention performs hydrogen-induced cracking resistance performance testing on the steel plate according to the NACETM0284-2016 standard and feeds back the three key indicators CLR, CTR, and CSR to the immune memory bank, establishing a complete closed loop of "cooling execution → HIC testing → memory update → parameter optimization." This ensures that the antibodies in the immune memory bank are continuously iterated and optimized based on the final anti-HIC service performance, guaranteeing that the system obtains real performance data and continuously accumulates knowledge with each production run. This invention addresses the stringent requirements for subsea pipeline steel, which must simultaneously meet the following specifications: yield strength ≥485MPa, tensile strength ≥570MPa, yield-to-tensile ratio ≤0.86, and after immersion in NACETM0284-2016 standard A solution for 96 hours, CLR ≤5%, CTR ≤1.5%, and CSR ≤0.5%. Through the aforementioned intelligent quenching and cooling control, an ideal multiphase microstructure of acicular ferrite + fine-grained ferrite + dispersed MA components can be stably obtained, with a grain size ≥11 and MA component size ≤2μm. Compared to traditional open-loop empirical control methods, this invention can adaptively optimize cooling parameters even under operating conditions such as cooling water temperature fluctuations, and can stably control the hydrogen-induced cracking resistance performance within the design requirements. The complete closed-loop feedback mechanism of this invention, "cooling execution → HIC verification → memory update → parameter optimization," ensures that the antibodies in the immune memory bank are continuously iterated and optimized based on the final anti-HIC service performance, achieving automatic knowledge accumulation and continuous improvement in cooling accuracy. Attached Figure Description

[0026] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0027] In the attached diagram: Figure 1 This is a schematic diagram of the overall architecture of an embodiment of the present invention; Figure 2 This is a schematic diagram of the overall process of an embodiment of the present invention; Figure 3 This is a schematic diagram of the workflow of the immune memory bank module in an embodiment of the present invention; Figure 4 This is a schematic diagram of the workflow of the clone selection optimization module in an embodiment of the present invention; Figure 5 This is a schematic diagram of the workflow of the negative selection monitoring module in an embodiment of the present invention; Figure 6 This is a schematic diagram of the immune memory bank update and maintenance process in an embodiment of the present invention. Detailed Implementation

[0028] The preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the scope of the invention.

[0029] Example 1

[0030] This embodiment provides an overall technical description of the intelligent quenching and cooling system and its control method for hydrogen-induced cracking-resistant subsea pipeline steel according to the present invention.

[0031] like Figure 1 As shown, the intelligent quenching and cooling system for hydrogen-induced cracking-resistant subsea pipeline steel of the present invention includes an intelligent control unit, an ultra-fast cooling execution unit, and a multi-source sensing and detection unit. The intelligent control unit is connected to both the ultra-fast cooling execution unit and the multi-source sensing and detection unit.

[0032] The ultra-fast cooling actuator includes an upper manifold assembly, a lower manifold assembly, and a side spray device. Each manifold is equipped with an independent flow regulating valve, with a cooling rate adjustment range of 15–40℃ / s and a final cooling temperature adjustment range of 450–600℃. The multi-source sensing unit includes a steel plate temperature measuring instrument installed at the ultra-fast cooling inlet and outlet, as well as flow sensors and water temperature sensors installed at each manifold.

[0033] like Figure 1As shown, the intelligent control unit is equipped with an immune memory bank module, a clone selection optimization module, and a negative selection monitoring module. The three modules are arranged horizontally side by side, with the immune memory bank module on the left, the clone selection optimization module in the middle, and the negative selection monitoring module on the right.

[0034] The immune memory bank module stores at least one memory antibody containing the steel grade composition identifier and the corresponding cooling parameter memory antibody for the finished product specifications of at least one type of steel resistant to hydrogen-induced cracking subsea pipelines. For example... Figure 3 As shown, each cooling parameter memory antibody includes the steel grade composition identifier, finished product specification, cooling rate, final cooling temperature, and corresponding hydrogen-induced cracking resistance performance indicators (CLR, CTR, and CSR). The immune memory bank module supports three operations: write, retrieval, and update. The write operation creates a new antibody record when a new steel grade is produced for the first time. The retrieval operation finds a matching memory antibody based on the input steel grade composition identifier and finished product specification. The update operation compares the hydrogen-induced cracking resistance performance indicators when a memory antibody with the same steel grade composition identifier and finished product specification already exists in the immune memory bank, retaining the one with the better CLR, CTR, and CSR.

[0035] During the retrieval process, the immune memory bank module calculates the affinity of each candidate antibody using an affinity function. The affinity function is:

[0036] Where A represents affinity. For steel grade composition identification matching, when the input steel grade composition identification is completely consistent with the steel grade composition identification in the memory antibody, =1, when inconsistent =0; For finished product specification matching, when the input finished product specification is completely consistent with the finished product specification in the memory antibody, =1, when the thickness deviation does not exceed 10%. =0.5, when the thickness deviation exceeds 10% =0; and For the weighting coefficients, satisfying Preferred =0.6, =0.4. The affinity is highest when both the steel grade composition and the finished product specifications match, and second highest when only the finished product specifications match.

[0037] When no memory antibody matches the current steel grade composition identifier and finished product specification in the immune memory bank (i.e., no matching record exists for either the steel grade composition identifier or the finished product specification), the clone selection optimization module generates initial cooling parameters based on a preset expert experience rule base. This expert experience rule base includes recommended cooling rates and final cooling temperature ranges corresponding to different steel grades (X65MS, X70MS, X80MS, etc.) and different finished product specification thickness ranges (≤20mm, >20mm and ≤35mm, >35mm). For cases where there is no matching steel grade composition identifier but a matching finished product specification, existing memory antibodies with similar alloy content under the same finished product specification are used as a reference, and the cooling rate is linearly corrected according to the difference in alloy element content of the steel grade, with a correction coefficient of 0.95~1.05. For cases where there is no matching steel grade composition identifier and finished product specification, the corresponding recommended cooling rate and final cooling temperature values ​​are selected from the expert experience rule base based on the finished product specification thickness as the initial cooling parameters. Simultaneously, the size of the initial antibody group is increased from the conventional 20 to 30 to enhance the global search capability.

[0038] like Figure 4 As shown, the cloning selection optimization module first selects the memory antibody with the highest affinity from the immune memory bank module as the initial solution based on the steel grade composition identifier and finished product specifications of the current steel plate. Then, centered on this initial solution, an initial antibody population is generated within a cooling rate range of ±30% and a final cooling temperature range of ±50℃. Subsequently, each antibody in the initial antibody population undergoes clonal amplification, with the cloning scale function being:

[0039] in, This refers to the number of clones of a single antibody. Based on the scale of the clone, Let be the affinity value of the i-th antibody. It is the sum of the affinity values ​​of all antibodies in the initial antibody population. This is a rounding function. Clonal amplification allows for a greater number of clones to be obtained from antibodies with higher affinity.

[0040] After cloning, the cloned antibody is mutated. The mutation operator is:

[0041] in, V represents the modified cooling parameters (cooling rate or final cooling temperature), and V represents the current cooling parameters. The coefficient of variation ranges from 0.1 to 0.3. The parameters are defined as follows: the cooling rate varies by ±30% of the current value, and the final cooling temperature varies by ±50℃ of the current value. These are standard normally distributed random numbers. A candidate antibody population is generated through mutation operations.

[0042] After generating the candidate antibody population, the fitness value of each antibody in the candidate antibody population is calculated using the temperature uniformity index U in the thickness direction of the steel plate and the final cooling temperature deviation index E as fitness indices.

[0043] The thickness-direction temperature uniformity index U is the standard deviation of the temperature at each node along the thickness direction of the steel plate, and its calculation is based on the one-dimensional transient heat conduction equation. The one-dimensional transient heat conduction equation is:

[0044] in, For the density of the steel plate, For specific heat capacity, Thermal conductivity, Let be the temperature value at position x in the thickness direction at time t. The boundary conditions are:

[0045] in, The convective heat transfer coefficient between the steel plate surface and the cooling water is determined empirically based on the cooling water flow rate and water temperature, within the process range of 15–40°C / s and 450–600°C described in this invention. The value range is 1500 to 4500. , The surface temperature of the steel plate. Let L be the cooling water temperature and L be half the thickness of the steel plate. Using the steel plate's initial water inlet temperature and the aforementioned heat transfer boundary conditions, the one-dimensional transient heat conduction equation is discretized and solved using the finite difference method to obtain the temperature values ​​of each discrete node along the steel plate thickness at different times. The finite difference method employs an implicit difference scheme for discretization, with a time step of 0.01 s and a spatial step of 1 / 20 of the steel plate thickness. For steel plates thicker than 40 mm, the spatial step is refined to 1 / 30 of the steel plate thickness to ensure computational accuracy.

[0046] The formula for calculating the thickness-direction temperature uniformity index U is:

[0047] Where n is the total number of discrete nodes in the thickness direction. Let i be the temperature value of the i-th node. This represents the average temperature at each node. A smaller U-value indicates a more uniform temperature distribution along the thickness direction.

[0048] The formula for calculating the final cooling temperature deviation index E is:

[0049] in, The measured temperature at the steel plate outlet. The target final cooling temperature is E. The smaller the E value, the smaller the deviation between the actual final cooling temperature and the target final cooling temperature.

[0050] The fitness function is:

[0051] Where F is the fitness value, U is the temperature uniformity index in the thickness direction, and E is the final cooling temperature deviation index. and For the weighting coefficients, satisfying =1, preferred =0.5, =0.5. and These represent the reciprocals of U and E, respectively. The smaller U or E is, the larger its reciprocal, and the higher the fitness value. The clone selection optimization module selects the antibody with the highest fitness from the candidate antibody population based on the fitness value as the optimal antibody, and outputs the cooling rate and final cooling temperature corresponding to the optimal antibody as control targets to the ultrafast cooling execution unit.

[0052] like Figure 5 As shown, the negative selection monitoring module consists of two stages: offline construction and online monitoring. In the offline construction stage, multiple sets of inlet temperature, flow rate of each manifold, water temperature, and outlet temperature data under historical normal operating conditions are collected. After normalization, each set of data is used as a pattern vector. The K-means clustering algorithm is used to divide the pattern vector into several categories. The number of clusters, k, is determined using the elbow method, which calculates the sum of squared errors (SSE) within clusters for different k values ​​(k ranges from 3 to 8), and selects the k value corresponding to the inflection point where the SSE decreases significantly as the optimal number of categories. The center vectors of each category constitute the normal operating condition pattern set. The normalization process uses the min-max normalization method to map each set of data to the [0,1] interval. In the online monitoring stage, inlet temperature, flow rate of each manifold, water temperature, and outlet temperature data are collected in real time. After normalization, the collected data is used as a real-time feature vector. The Euclidean distance between this real-time feature vector and the center vectors of each category in the normal operating condition pattern set is calculated, and the matching degree function is:

[0053] Where D is the Euclidean distance between the collected data vector and the center vector of the normal operating condition mode category. To collect the k-th dimension feature after data normalization, Let k be the k-th dimension of the class center vector. Let D be the standard deviation of the k-th feature, and p be the feature dimension (in this system, p=4, corresponding to the fused features of inlet temperature, flow rate of each manifold, water temperature, and outlet temperature). The calculated minimum Euclidean distance is compared with a preset threshold, which is the maximum Euclidean distance from the pattern vector to the category center vector within each category. When D ≤ the preset threshold, the operating condition is considered normal, and cooling operation continues; when D > the preset threshold, the operating condition is considered abnormal, and an alarm signal is output to notify the operator to intervene.

[0054] After receiving the cooling rate and final cooling temperature corresponding to the optimal antibody output by the clone selection optimization module, the ultra-fast cooling execution unit controls the water spray flow of the upper manifold group, lower manifold group and side spray device by adjusting the opening of the independent flow regulating valve of each manifold, so that the steel plate is cooled to the target final cooling temperature according to the target cooling rate.

[0055] like Figure 2 As shown, the complete process of the intelligent quenching and cooling control method for hydrogen-induced cracking subsea pipeline steel of the present invention includes the following steps.

[0056] Step S1: Obtain the steel grade composition identification and finished product specifications of the steel plate to be cooled.

[0057] Step S2: Retrieve cooling parameter memory antibodies from the immune memory bank that match the current steel grade composition identifier and finished product specifications, and then use an affinity function. Calculate the affinity of each candidate antibody and select the memory antibody with the highest affinity as the initial cooling parameter.

[0058] Step S3: Generate the initial antibody population within the range of ±30% cooling rate and ±50℃ final cooling temperature, centered on the initial cooling parameters.

[0059] Step S4: Each antibody in the initial antibody population is cloned and amplified. The number of clones is positively correlated with the antibody affinity value, and the clone size function is: The cloned antibody is then subjected to mutation operations, with the mutation operator being... , generating candidate antibody populations.

[0060] Step S5: Based on the one-dimensional transient heat conduction equation Calculate the temperature field distribution along the thickness of the steel plate and obtain the standard deviation of the temperature at each node along the thickness. As an index of temperature uniformity in the thickness direction, calculation As an indicator of final cooling temperature deviation, The fitness value of each antibody in the candidate antibody population is calculated for the fitness function.

[0061] Step S6: Select the antibody with the highest fitness from the candidate antibody group based on the fitness value as the optimal antibody.

[0062] Step S7: Using the cooling rate and final cooling temperature corresponding to the optimal antibody as control targets, adjust the flow rate of each manifold of the ultra-fast cooling execution unit to cool the steel plate.

[0063] Step S8: During the cooling process, inlet temperature, flow rate of each manifold, water temperature and outlet temperature are collected in real time. After normalizing the collected data, a real-time feature vector is constructed. The Euclidean distance between this feature vector and the center vectors of each category in the normal operating condition mode set is calculated. When D ≤ preset threshold, cooling continues; when D > preset threshold, an alarm signal is output.

[0064] Step S9: Test the hydrogen-induced cracking resistance of the cooled steel plate by immersing it in solution A of NACETM0284-2016 standard for 96 hours and obtaining the crack length ratio (CLR), crack width ratio (CTR), and crack sensitivity ratio (CSR).

[0065] Step S10: Update the immune memory bank with the optimal antibody obtained from this cooling process, along with CLR, CTR, and CSR, as indicators of the antibody's anti-hydrogen-induced cracking performance. For example... Figure 6 As shown, during the update, the system first checks whether a memory antibody with the same composition identifier and finished product specification as the current steel grade already exists in the immune memory bank. If it does not exist, a new antibody record is created and the current best antibody, along with its CLR, CTR, and CSR, is stored in the memory bank. If it already exists, the system compares the current data with the existing record's CLR, CTR, and CSR. If the current data's CLR, CTR, and CSR are not higher than the existing record (i.e., its resistance to hydrogen-induced cracking is better or equal), the current data replaces the existing record. Otherwise, the current data is discarded and the existing record is retained.

[0066] Example 2

[0067] This embodiment uses a 25mm diameter hydrogen-resistant subsea pipeline steel (steel composition identifier: X65MS) as an example to illustrate the actual operation process of the system of the present invention when the same steel composition identifier and finished product specification record already exist in the immune memory bank. In this embodiment, the chemical composition of the hydrogen-resistant subsea pipeline steel used is as follows by weight percentage: C: 0.045%, Si: 0.22%, Mn: 1.05%, P: 0.006%, S: 0.0012%, Al: 0.028%, Nb: 0.055%, Ti: 0.016%, V: 0.032%, Cu: 0.20%, Ni: 0.28%, Cr: 0.18%, Mo: 0.20%, Ca: 0.0015%, Mg: 0.0018%, N: 0.004%, O: 0.002%, with the balance Fe being an unavoidable impurity; Ca / S = 1.25, Mg / Al = 0.064.

[0068] In this embodiment, the reference water temperature for normal operation of the cooling water system is set to 28°C.

[0069] Prior to this cooling operation, there was a memory antibody record in the immune memory bank with steel grade identification as X65MS and finished product specification of 25mm. The cooling parameters in the record were cooling rate of 26℃ / s and final cooling temperature of 515℃, and the corresponding hydrogen-induced cracking resistance performance indexes were CLR=3.2%, CTR=0.8%, and CSR=0.3%.

[0070] The system executes step S1 to obtain the steel grade composition identifier of the current steel plate as X65MS and the finished product specification as 25mm. Step S2 retrieves a matching memory antibody from the immune memory bank. Since the steel grade composition identifier and finished product specification are a perfect match, the affinity function... middle The affinity A = 0.6 × 1 + 0.4 × 1 = 1.0, reaching the highest value. This memory antibody was selected as the initial cooling parameter, that is, the initial cooling rate is 26℃ / s and the initial and final cooling temperatures are 515℃.

[0071] Step S3 generates an initial antibody population centered on the initial cooling parameters. The cooling rate ranges from 26×(1-30%) to 26×(1+30%), i.e., 18.2 to 33.8 °C / s, and the final cooling temperature ranges from 515-50 to 515+50, i.e., 465 to 565 °C. Within this range, 20 initial antibodies are randomly generated. Step S4 is then performed... Based on a cloning scale of 10, each antibody was cloned and amplified. The antibody with the highest affinity yielded the largest number of clones, and the antibody with the lowest affinity yielded the smallest number of clones. Then, mutation operations were applied to all cloned antibodies, and the coefficient of variation was... Using 0.2, a candidate antibody population containing 150 candidate antibodies is generated.

[0072] Step S5 is executed to calculate the fitness of each candidate antibody. This is done using a one-dimensional transient heat conduction equation. Based on this, the finite difference method was used to discretize the steel plate thickness direction into 20 nodes. Using a water immersion temperature of 780℃ as the initial condition and the heat transfer coefficient as the boundary condition, the temperature field distribution along the steel plate thickness direction under the corresponding cooling parameters for each candidate antibody was calculated. After obtaining the temperature at each node, calculations were performed... Simultaneously calculate ,Pick ,according to Calculate the fitness value of each antibody. Perform step S6 to screen the antibody with the highest fitness as the optimal antibody. In this embodiment, the optimal antibody obtained by screening has the following cooling parameters: cooling rate 25℃ / s and final cooling temperature 520℃.

[0073] Step S7 sets the cooling rate of 25℃ / s and the final cooling temperature of 520℃ as control targets, adjusting the flow control valves of the upper manifold, lower manifold, and side spray device of the ultra-fast cooling actuator to cool the steel plate. During cooling, step S8 is executed. The inlet temperature measured by the inlet thermometer is 782℃, the deviation between the actual flow rate and the set value reported by the flow sensors in each manifold is no more than ±2%, the water temperature sensor measures 28℃, and the outlet temperature meter measures 518℃. The real-time collected data is normalized to form a feature vector. The Euclidean distance D = 0.52 between this vector and the center vectors of each category in the normal operating condition mode set is calculated. The preset threshold is 0.85. If D ≤ the preset threshold, the operating condition is normal, and cooling continues.

[0074] After cooling, step S9 was performed. Samples of the cooled steel plate were taken and immersed in solution A according to NACETM0284-2016 standard for 96 hours to test its resistance to hydrogen-induced cracking. The measured CLR was 2.8%, CTR was 0.6%, and CSR was 0.2%. Step S10 was performed to update the optimal antibody (cooling rate 25℃ / s, final cooling temperature 520℃) and the measured CLR = 2.8%, CTR = 0.6%, and CSR = 0.2% to the immune memory bank. Since the immune memory bank already contained records for steel grade X65MS and specification 25mm, comparing the hydrogen-induced cracking resistance indicators, the current data CLR = 2.8% was superior to the existing recorded CLR = 3.2%, CTR = 0.6% was superior to the existing recorded CTR = 0.8%, and CSR = 0.2% was superior to the existing recorded CSR = 0.3%. Therefore, the current data replaced the existing records. Upon inspection, the metallographic structure of the steel plate prepared in this embodiment is acicular ferrite + fine-grained ferrite + dispersed MA component, with a grain size of grade 12, an average MA component size of 1.2 μm, and a maximum size of 1.8 μm; yield strength of 492 MPa, tensile strength of 582 MPa, yield ratio of 0.85; CLR=2.8%, CTR=0.6%, CSR=0.2%, and all indicators meet the design requirements.

[0075] Example 3

[0076] This embodiment uses a 20mm diameter hydrogen-resistant subsea pipeline steel (steel composition identifier X70MS) as an example to illustrate the actual operation process of the system of the present invention when there are finished products of the same specifications but different steel composition identifiers in the immune memory bank. In this embodiment, the chemical composition of the hydrogen-resistant subsea pipeline steel used is as follows by weight percentage: C: 0.040%, Si: 0.18%, Mn: 0.95%, P: 0.004%, S: 0.0008%, Al: 0.030%, Nb: 0.050%, Ti: 0.014%, V: 0.028%, Cu: 0.18%, Ni: 0.25%, Cr: 0.16%, Mo: 0.18%, Ca: 0.0012%, Mg: 0.0015%, N: 0.003%, O: 0.002%, with the balance Fe being an unavoidable impurity; Ca / S = 1.50, Mg / Al = 0.050.

[0077] Prior to this cooling operation, the immune memory bank already contained a record of a memory antibody with a steel composition of X65MS and a finished product size of 20mm. The cooling parameters in this record were a cooling rate of 28℃ / s and a final cooling temperature of 530℃, with corresponding hydrogen-induced cracking resistance indices of CLR=3.5%, CTR=0.9%, and CSR=0.4%. The immune memory bank did not contain a record of a steel composition of X70MS with a finished product size of 20mm.

[0078] The system executes step S1 to obtain the steel grade composition identifier of the current steel plate as X70MS and the finished product specification as 20mm. Step S2 retrieves a matching memory antibody from the immune memory bank. The currently input does not match the steel grade composition identifier recorded as X65MS-20mm. However, the finished product specifications match. Affinity A = 0.6 × 0 + 0.4 × 1 = 0.4. Since there are no other more matching records in the immune memory bank, this memory antibody is selected as the initial cooling parameter, that is, the initial cooling rate is 28℃ / s and the initial and final cooling temperatures are 530℃.

[0079] Step S3 generates an initial antibody population centered on the initial cooling parameters. The cooling rate ranges from 28×(1-30%) to 28×(1+30%), i.e., 19.6–36.4 °C / s, and the final cooling temperature ranges from 530-50°C to 530+50°C, i.e., 480–580 °C. Within this range, 20 initial antibodies are randomly generated. Step S4 then... =10 as the basic clone size, each antibody was cloned and amplified. Then, mutation operations were applied to all cloned antibodies, and the coefficient of variation was... Take 0.25 to generate a candidate antibody population.

[0080] Step S5 is executed to calculate the fitness of each candidate antibody. Based on the one-dimensional transient heat conduction equation, the finite difference method is used to calculate the thickness-direction temperature field distribution under the corresponding cooling parameters for each candidate antibody. After obtaining the U and E values, the following steps are performed... Calculate the fitness value of each antibody. Perform step S6 to screen the antibody with the highest fitness as the optimal antibody. In this embodiment, the optimal antibody obtained by screening has the following cooling parameters: cooling rate 30℃ / s and final cooling temperature 510℃.

[0081] Step S7 sets the cooling rate of 30℃ / s and the final cooling temperature of 510℃ as control targets, and adjusts the flow rate of each manifold of the ultra-fast cooling unit to cool the steel plate. During the cooling process, step S8 is executed. The Euclidean distance D between the real-time sensor data and the normal operating condition mode set is 0.48, and the preset threshold is 0.85, indicating normal operating conditions. After cooling is completed, step S9 is executed, and the hydrogen-induced cracking resistance performance is tested by immersing the steel plate in solution A according to NACETM0284-2016 standard for 96 hours. The measured CLR is 3.8%, CTR is 1.0%, and CSR is 0.4%. Step S10 updates the optimal antibody (cooling rate 30℃ / s, final cooling temperature 510℃) and the measured CLR = 3.8%, CTR = 1.0%, and CSR = 0.4% to the immune memory bank. Since there is no record for steel grade X70MS and specification 20mm in the immune memory bank, a new antibody record is created and stored in the memory bank. Upon inspection, the metallographic structure of the steel plate prepared in this embodiment is acicular ferrite + fine-grained ferrite + dispersed MA component, with a grain size of 11.5 grade; yield strength of 496 MPa, tensile strength of 588 MPa, yield ratio of 0.84; CLR=3.8%, CTR=1.0%, CSR=0.4%, and all indicators meet the design requirements.

[0082] Example 4

[0083] This embodiment uses a 45mm diameter hydrogen-resistant subsea pipeline steel (steel composition identifier: X80MS) as an example to illustrate the actual operation process of the system of the present invention when there are no matching records in the immune memory bank. In this embodiment, the chemical composition of the hydrogen-resistant subsea pipeline steel used, by weight percentage, is: C: 0.050%, Si: 0.20%, Mn: 1.15%, P: 0.005%, S: 0.0010%, Al: 0.035%, Nb: 0.065%, Ti: 0.020%, V: 0.040%, Cu: 0.25%, Ni: 0.35%, Cr: 0.25%, Mo: 0.25%, Ca: 0.0020%, Mg: 0.0025%, N: 0.004%, O: 0.002%, with the balance Fe being an unavoidable impurity. The Ca / S ratio is 2.00, and the Mg / Al ratio is 0.071.

[0084] Prior to this cooling operation, there were no records in the immune memory bank for steel grade composition identified as X80MS, nor for finished product specifications of 45mm.

[0085] The system executes step S1 to obtain the steel grade composition identifier of the current steel plate as X80MS and the finished product specification as 45mm. Step S2 retrieves a matching memory antibody from the immune memory bank. Since there are no matching records for either the steel grade composition identifier or the finished product specification, the antibody with the highest affinity has an extremely low affinity value. In this embodiment, the system uses this low-affinity antibody as a basis, combined with empirical default values, to generate initial cooling parameters, namely, an initial cooling rate of 22℃ / s and an initial and final cooling temperature of 480℃.

[0086] Step S3 generates an initial antibody population centered on the initial cooling parameters. The cooling rate ranges from 22×(1-30%) to 22×(1+30%), i.e., 15.4 to 28.6℃ / s, and the final cooling temperature ranges from 480-50℃ to 480+50℃, i.e., 430 to 530℃. Within this range, 30 initial antibodies are randomly generated (the initial antibody population size is increased to enhance global search capabilities due to the lack of matching records). Step S4 is then executed... =12 is the basic clone size for clonal amplification and mutation operations, with a coefficient of variation of 12. Take 0.3 to generate a candidate antibody population.

[0087] Step S5 is executed to calculate the fitness of each candidate antibody. Based on the one-dimensional transient heat conduction equation, the finite difference method is used to calculate the thickness-direction temperature field distribution under the corresponding cooling parameters for each candidate antibody. For a 45mm thick plate, the thickness direction is discretized into 30 nodes to obtain a finer temperature distribution. After obtaining the U and E values, the following steps are performed... Calculate the fitness value of each antibody. Perform step S6 to screen the antibody with the highest fitness as the optimal antibody. In this embodiment, the optimal antibody obtained by screening has the following cooling parameters: cooling rate 20℃ / s and final cooling temperature 470℃.

[0088] Step S7 sets the cooling rate of 20℃ / s and the final cooling temperature of 470℃ as control targets, and adjusts the flow rates of each manifold of the ultra-fast cooling actuator to cool the steel plate. During the cooling process, step S8 is executed. The inlet temperature measured by the inlet thermometer is 765℃, the flow sensors of each manifold report stable actual flow rates, and the outlet temperature measured by the outlet thermometer is 472℃. The Euclidean distance D between the real-time sensor data and the normal operating condition mode set is 0.55, the preset threshold is 0.85, and the operating condition is normal.

[0089] After cooling, step S9 was performed, and the hydrogen-induced cracking resistance was tested by immersing the steel in solution A according to NACETM0284-2016 standard for 96 hours. The measured CLR was 4.2%, CTR was 1.3%, and CSR was 0.5%. Step S10 was performed to update the optimal antibody (cooling rate 20℃ / s, final cooling temperature 470℃) and the measured CLR = 4.2%, CTR = 1.3%, and CSR = 0.5% to the immune memory bank. Since there was no record for steel grade X80MS and specification 45mm in the immune memory bank, a new antibody record was created and stored in the memory bank. The metallographic structure of the steel plate prepared in this embodiment was found to be acicular ferrite + fine-grained ferrite + dispersed MA component, with a grain size of grade 11; yield strength 502MPa, tensile strength 595MPa, yield ratio 0.84; CLR = 4.2%, CTR = 1.3%, and CSR = 0.5%, all of which met the design requirements.

[0090] Example 5

[0091] This embodiment is a comparative embodiment used to illustrate the superiority of the system of the present invention compared with traditional open-loop control. The same steel composition (X65MS, 25mm) as in Embodiment 2 is used for cooling operation, but the traditional open-loop control method is used, that is, without cloning selection optimization, cooling is performed directly with empirically set fixed cooling parameters (cooling rate 25℃ / s, final cooling temperature 520℃), and no negative selection monitoring module is provided.

[0092] During the cooling process, the cooling water temperature increased by 5°C compared to the preset conditions (from 28°C to 33°C). Traditional open-loop control could not detect this change and make adjustments, resulting in an actual cooling rate decrease to approximately 22°C / s and an actual final cooling temperature increase to approximately 538°C. After cooling, the hydrogen-induced cracking resistance was tested by immersing the device in solution A according to NACETM0284-2016 standard for 96 hours. The measured CLR was 6.8%, CTR was 2.1%, and CSR was 0.8%, with the CLR of 6.8% exceeding the design requirement of ≤5%.

[0093] Comparing Example 5 with Example 2, it can be seen that under the same steel grade and specifications, the system of the present invention, through the cloning selection optimization module, adaptively adjusts the cooling parameters according to real-time operating conditions, and through the negative selection monitoring module, monitors abnormal operating conditions in real time. It can automatically match the optimal cooling parameters under disturbances such as changes in cooling water temperature, ensuring that the HIC performance indicators stably meet design requirements. In contrast, traditional open-loop control methods lack adaptive optimization and online monitoring capabilities, making it difficult to guarantee consistent cooling effects when operating conditions fluctuate, and the HIC performance indicators are at risk of exceeding limits. The system of the present invention, through a complete closed loop of "cooling execution → HIC verification → memory update → parameter optimization," achieves continuous optimization of cooling parameters and automatic accumulation of knowledge, continuously improving the cooling control accuracy and HIC performance stability of subsequent products of the same specifications.

[0094] Example 6

[0095] This embodiment uses 20mm diameter hydrogen-induced cracking resistant subsea pipeline steel (steel composition identification: X65MS) as an example to verify the control effect of the system of the present invention at the lower limit of cooling rate (15℃ / s). The chemical composition of the steel used in this embodiment is the same as that in Example 2. The immune memory bank already contains historical records of this steel grade and specification. The optimal antibody output by the clone selection optimization module corresponds to a cooling rate of 15℃ / s and a final cooling temperature of 580℃. After the cooling operation, the HIC test was performed by immersion in NACETM0284-2016 standard A solution for 96 hours. The measured CLR=4.5%, CTR=1.2%, and CSR=0.4%, and all indicators meet the design requirements, indicating that the present invention can still effectively control the hydrogen-induced cracking resistance under the lower limit of cooling rate.

[0096] Example 7

[0097] This embodiment uses 16mm thick hydrogen-induced cracking resistant subsea pipeline steel (steel composition identified as X70MS) as an example to verify the control effect of the system of the present invention at the upper limit of the cooling rate (40℃ / s). The immune memory bank already contains historical records of this steel grade and specification. The optimal antibody output by the clone selection optimization module corresponds to a cooling rate of 40℃ / s and a final cooling temperature of 460℃. After cooling, the sample was immersed in NACETM0284-2016 standard A solution for 96 hours for HIC testing. The measured CLR=2.5%, CTR=0.5%, and CSR=0.2%, all meeting the design requirements, indicating that the present invention can still achieve excellent hydrogen-induced cracking resistance even under the upper limit of the cooling rate.

[0098] Example 8

[0099] This embodiment uses 35mm thick hydrogen-induced cracking resistant subsea pipeline steel (steel composition identified as X80MS) as an example to verify the control effect of the system of the present invention at the upper limit of the final cooling temperature (600℃). The optimal antibody output by the clone selection optimization module corresponds to a cooling rate of 18℃ / s and a final cooling temperature of 600℃. After cooling, HIC testing was performed according to the NACETM0284-2016 standard. The measured CLR=4.8%, CTR=1.4%, and CSR=0.5%, all of which meet the design requirements, indicating that the present invention can still guarantee qualified HIC performance under the upper limit of the final cooling temperature.

[0100] Example 9

[0101] This embodiment uses 12mm thick hydrogen-induced cracking resistant subsea pipeline steel (steel composition identified as X65MS) as an example to verify the control effect of the system of the present invention at the lower limit of the final cooling temperature (450℃). The optimal antibody output by the cloning selection optimization module corresponds to a cooling rate of 35℃ / s and a final cooling temperature of 450℃. After cooling, HIC testing was performed according to the NACETM0284-2016 standard. The measured CLR=2.2%, CTR=0.4%, and CSR=0.1%, all of which meet the design requirements, indicating that the present invention can achieve good hydrogen-induced cracking resistance under the lower limit of the final cooling temperature.

[0102] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the claims of the present invention.

Claims

1. A quenching and cooling system for subsea pipeline steel resistant to hydrogen-induced cracking, characterized in that, The system includes an ultra-fast cooling actuator, a multi-source sensor detection unit, and an intelligent control unit. The ultra-fast cooling actuator includes an upper manifold assembly, a lower manifold assembly, and a side spray device, with each manifold equipped with an independent flow regulating valve. The multi-source sensor detection unit includes a steel plate temperature sensor located at the ultra-fast cooling inlet and outlet, as well as flow sensors and water temperature sensors located at each manifold. The intelligent control unit is connected to the ultra-fast cooling actuator and the multi-source sensor detection unit. Specifically: The intelligent control unit is equipped with an immune memory bank module, a clone selection optimization module, and a negative selection monitoring module; The immune memory bank module stores at least one cooling parameter memory antibody corresponding to the steel composition identifier and finished product specification of the subsea pipeline steel resistant to hydrogen-induced cracking. Each cooling parameter memory antibody includes the steel composition identifier, finished product specification, cooling rate, final cooling temperature and corresponding hydrogen-induced cracking resistance performance index. The clonal selection optimization module is used to select an initial solution based on the steel grade composition identification and finished product specifications of the current steel plate. After clonal amplification and mutation of the initial solution, a candidate antibody group is generated. The optimal antibody is screened from the candidate antibody group using the steel plate thickness direction temperature uniformity index and the final cooling temperature deviation index as fitness indicators. The negative selection monitoring module is used to collect the detection signals of the multi-source sensing detection unit in real time, match the detection signals with a preset normal operating condition mode set, and output an alarm signal when the matching degree is lower than a preset threshold. The ultrafast cooling execution unit receives control commands sent by the intelligent control unit and performs cooling operations according to the cooling rate and final cooling temperature corresponding to the optimal antibody output by the clone selection optimization module.

2. The quenching and cooling system for hydrogen-induced cracking resistant subsea pipeline steel according to claim 1, characterized in that, When the clone selection optimization module selects an initial solution, it selects the memory antibody with the highest affinity from the immune memory bank module as the initial solution. The affinity is determined by a weighted sum of the steel composition matching item and the finished product specification matching item. When the steel composition identification matches, the matching item takes full marks; when it does not match, it takes zero marks. When the finished product specification matches completely, the matching item takes full marks; when the thickness deviation is within the set allowable range, it takes half marks; when it exceeds the set allowable range, it takes zero marks. Moreover, the weight value of the steel composition matching item is higher than the weight value of the finished product specification matching item.

3. The quenching and cooling system for hydrogen-induced cracking resistant subsea pipeline steel according to claim 2, characterized in that, When there is no record in the immune memory bank that matches the steel composition identifier and the finished product specification, the clone selection optimization module generates initial cooling parameters based on a preset expert experience rule base and increases the size of the initial antibody population from twenty to thirty.

4. The quenching and cooling system for hydrogen-induced cracking resistant subsea pipeline steel according to claim 1, characterized in that, The cooling rate adjustment range of the ultra-fast cooling execution unit is 15–40℃ / s, and the final cooling temperature adjustment range is 450–600℃.

5. The quenching and cooling system for hydrogen-induced cracking resistant subsea pipeline steel according to claim 1, characterized in that, The normal operating condition pattern set is constructed by clustering historical normal operating data using the K-means clustering algorithm, and the number of clusters k is determined by the elbow rule; each pattern vector in the pattern set is a four-dimensional feature vector obtained by performing minimum-maximum normalization on inlet temperature, flow rate of each manifold, water temperature and outlet temperature.

6. A method for controlling the quenching and cooling of steel for hydrogen-induced cracking subsea pipelines, employing the quenching and cooling system as described in claim 1, characterized in that, Includes the following steps: (1) Obtain the steel grade composition identification and finished product specifications of the steel plate to be cooled; (2) Retrieve cooling parameter memory antibodies that match the current steel grade composition identifier and finished product specification from the immune memory bank, and determine the affinity of each candidate antibody in the following manner: determined by the weighted sum of the steel grade composition matching item and the finished product specification matching item. When the steel grade composition identifier matches, the matching item takes full score; when it does not match, it takes zero value. When the finished product specification matches completely, the matching item takes full score; when the thickness deviation is within the set allowable range, it takes half score; when it exceeds the set allowable range, it takes zero value. The weight value of the steel grade composition matching item is higher than the weight value of the finished product specification matching item. The memory antibody with the highest affinity is used as the initial cooling parameter. (3) Based on the initial cooling parameters, generate an initial antibody population within a cooling rate range of ±30% and a final cooling temperature range of ±50 degrees Celsius. (4) Each antibody in the initial antibody group is cloned and amplified. The number of clones is positively correlated with the affinity value of the antibody. The cloned antibodies are then mutated to generate a candidate antibody group. (5) Using the temperature uniformity index in the thickness direction of the steel plate and the final cooling temperature deviation index as fitness indices, calculate the fitness value of each antibody in the candidate antibody group; (6) Select the optimal antibody from the candidate antibody population based on the fitness value; (7) Using the cooling rate and final cooling temperature corresponding to the optimal antibody as control targets, adjust the flow rate of each manifold of the ultra-fast cooling execution unit to cool the steel plate; (8) During the cooling process, the inlet temperature, flow rate of each manifold, water temperature and outlet temperature are collected in real time. The collected data are normalized to form a real-time feature vector. The matching degree of the feature vector with the center vector of each category of the preset normal working condition mode set is calculated. Cooling continues when the matching degree is less than or equal to the preset threshold. An alarm signal is output when the matching degree is greater than the preset threshold. (9) Test the hydrogen-induced cracking resistance of the cooled steel plate and obtain the crack length ratio, crack width ratio and crack sensitivity ratio according to the NACETM0284-2016 standard. (10) The optimal antibody obtained from this cooling, along with the crack length rate, crack width rate, and crack sensitivity rate, are used as indicators of the antibody's resistance to hydrogen-induced cracking and are updated in the immune memory bank.

7. The quenching and cooling control method for hydrogen-induced cracking resistant subsea pipeline steel according to claim 6, characterized in that, The number of clones amplified in step (4) is positively correlated with the affinity value of the antibody; the higher the affinity value, the more clones are obtained. The mutation operation is carried out within a cooling rate range of ±30% and a final cooling temperature range of ±50 degrees Celsius.

8. The quenching and cooling control method for hydrogen-induced cracking resistant subsea pipeline steel according to claim 6, characterized in that, The thickness direction temperature uniformity index mentioned in step (5) is the standard deviation of the temperature at each node in the thickness direction of the steel plate, which is calculated by solving the temperature field distribution in the thickness direction of the steel plate using the finite difference method based on the one-dimensional transient heat conduction equation; the final cooling temperature deviation index is the absolute value of the difference between the measured temperature at the steel plate outlet and the target final cooling temperature; the fitness value is negatively correlated with the thickness direction temperature uniformity index and negatively correlated with the final cooling temperature deviation index.

9. The quenching and cooling control method for hydrogen-induced cracking resistant subsea pipeline steel according to claim 6, characterized in that, The matching degree in step (8) is measured by Euclidean distance, and the preset threshold is the maximum Euclidean distance from the pattern vector to the center vector of each category.

10. The quenching and cooling control method for hydrogen-induced cracking resistant subsea pipeline steel according to claim 6, characterized in that, When updating to the immune memory bank in step (10), if there is a memory antibody in the immune memory bank that is the same as the current steel composition identifier and finished product specification, then compare the hydrogen-induced cracking performance index. If the crack length rate, crack width rate and crack sensitivity rate of the current data are not higher than the existing records, then replace the existing records with the current data. Otherwise, retain the existing record.

Images