Pipeline welding path planning method and system

The dynamic energy consumption model is constructed through the LSTM model and the welding parameters are adjusted in real time, which solves the problem that traditional welding systems are difficult to cope with complex working conditions, and realizes the optimization of welding energy consumption and the improvement of weld quality.

CN118635732BActive Publication Date: 2024-11-19ENDU HEAVY IND (NANTONG) CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411114965.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-14
Publication Date
2024-11-19
Estimated Expiration
2044-08-14

AI Technical Summary

Technical Problem

Traditional automated welding systems lack the ability to flexibly respond to complex working conditions and are difficult to dynamically adjust welding parameters based on the immediate quality feedback formed by welds, resulting in energy waste and weld quality problems.

Method used

The dynamic energy consumption model is constructed using the long and short-term memory network (LSTM) model, collect welding current and voltage data in real time, generate feedback adjustment coefficients, and optimize welding parameters by selecting and adjusting models to minimize welding energy consumption and maximize weld quality. .

Benefits of technology

It realizes more precise energy control and weld quality optimization during the welding process, reduces energy waste and improves welding efficiency and quality.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118635732B_ABST
    Figure CN118635732B_ABST
Patent Text Reader

Abstract

The present invention provides a pipeline welding path planning method and system, which relate to the technical field of intelligent welding. The present invention collects welding current and voltage data in real time, and uses a dynamic energy consumption model constructed by LSTM to accurately predict the upcoming energy demand, thereby achieving more accurate energy control in the welding process. By analyzing weld appearance quality indicators and internal defect indicators and comparing them with set evaluation thresholds, specific feedback adjustment coefficients can be generated. These adjustment coefficients not only guide welding equipment to adjust welding parameters in real time to adapt to different welding conditions, but also can be optimized according to energy consumption and weld quality data in the welding process, ultimately achieving minimization of welding energy consumption and maximization of weld quality.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of intelligent welding technology, and in particular to a pipeline welding path planning method and system. Background Art

[0002] Pipeline welding technology plays a vital role in the petroleum, chemical, shipbuilding and other industries. The traditional pipeline welding method relies on experienced technicians to manually adjust the welding parameters. This method that relies on manual experience has many shortcomings such as complex operation, large fluctuations in welding quality and low energy utilization. Welding path planning involves determining the welding path, including welding sequence, direction, speed, etc., and adjusting the welding parameters on the path to achieve the desired welding quality and efficiency. In the welding process, welding path planning usually includes the following aspects:

[0003] Path design: Determine the welding path, including the sequence and location of welding points, which involves considering factors such as the geometry, structure, and material properties of the welded parts to determine the best welding path;

[0004] Speed ​​and direction: Determine the movement speed and direction during welding, including the moving speed, turning, and stopping of the welding head to ensure the stability and uniformity of the welding process;

[0005] Welding parameters: including the settings of welding current, voltage, wire feeding speed and other parameters. These parameters directly affect the formation of the molten pool, temperature distribution, weld quality, etc. during the welding process, so they need to be adjusted according to the specific situation;

[0006] Cao Shouqi et al.'s "Butt Weld Path Planning Method Based on Welding Path Node Extraction" describes a weld path planning method; in the prior art, the publication number CN114273754B discloses an automatic welding control method, which identifies the pipe docking position by moving the welding unit, so that the welding unit is located directly above the pipe docking position. Through this automatic control method, the labor cost is reduced, the welding unit is initialized, the initial position of the welding unit is obtained, the pipe docking position is scanned, the scanning data is obtained, the welding gun position is obtained through the scanning data, the welding parameters and the welding path are planned, and the welding unit is controlled to perform welding operations; the welding gun position is obtained by sensor calculation, and the welding parameters and welding path are automatically planned through a deep learning network, so as to realize the automation of the entire welding process, greatly reduce the labor cost, and eliminate the precision error caused by manual operation, thereby improving the welding precision;

[0007] Although the development of automated welding technology has provided certain solutions for improving welding quality and efficiency, traditional automated systems often still lack the ability to flexibly cope with complex working conditions. Common automated welding systems in the prior art usually preset welding parameters, which are difficult to dynamically adjust according to the immediate quality feedback of the weld formation during the actual welding process, resulting in serious energy waste and possible weld quality problems.

[0008] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0009] The object of the present invention is to provide a pipeline welding path planning method and system to solve the problems raised in the above background technology.

[0010] To achieve the above object, the present invention provides the following technical solutions:

[0011] A pipeline welding path planning method, the specific steps include:

[0012] Step S1, obtaining a planned path for pipeline welding, and determining welding parameters of the current welding equipment at each node of the pipeline according to the planned path, and then welding the pipeline joints by the welding equipment, while collecting time series data of current I(t) and voltage V(t) on the welding equipment, and collecting weld quality data formed in real time at the pipeline joints;

[0013] Step S2, obtaining the collected time series data of current I(t) and voltage V(t), and using them as input of a long short-term memory network (LSTM) model to construct a dynamic energy consumption model to predict energy consumption;

[0014] Step S3, obtaining weld quality data, the weld quality data respectively including weld appearance quality index and weld internal defect index, and determining the evaluation threshold A of the weld appearance quality index, comparing and analyzing the weld appearance quality index with the evaluation threshold A, and generating a first feedback adjustment coefficient. In pipeline welding path planning, the coefficient is used to provide a first adjustment strategy for welding parameters of welding equipment;

[0015] Step S4: Based on step S3, determine the evaluation threshold B of the weld internal defect index, compare and analyze the weld internal defect index with the evaluation threshold B, and generate a second feedback adjustment coefficient. In pipeline welding path planning, the coefficient is used to provide a second adjustment strategy for the welding parameters of the welding equipment;

[0016] Step S5, obtain the first feedback adjustment coefficient and the second feedback adjustment coefficient, and analyze and process them in combination with the dynamic energy consumption model to construct a selection adjustment model, which is used to provide a screening strategy for the first adjustment strategy and the second adjustment strategy, and calibrate the selected welding parameter adjustment value to minimize the welding energy consumption while ensuring the quality of the weld.

[0017] Furthermore, the first feedback adjustment coefficient and the second feedback adjustment coefficient are obtained, and are analyzed and processed in combination with the dynamic energy consumption model to construct a selection adjustment model, which specifically includes the following contents:

[0018] The selection adjustment model is defined as E(t), and the calculation formula is as follows:

[0019] E(t)=c1·K1(t)+c2·K2(t)

[0020] K1(t) and K2(t) are the first feedback adjustment coefficient and the second feedback adjustment coefficient based on the long short-term memory network LSTM model, respectively. t represents the time series label, which is a continuous time series.

[0021] c1 and c2 are non-negative weight parameters, and c1+c2=1, which are adjusted according to the sensitivity of the actual adjustment strategy; the value range of E(t) is set to [0,1];

[0022] Divide the value range of E(t) [0,1] into the following intervals:

[0023] In the interval A[0,0.25), the E(t) value in this interval indicates that the welding quality and energy consumption optimization have reached the ideal state, and the adjustment strategy is relatively slight or maintains the status quo;

[0024] The screening basis is when the weight of c1 is small, at this time c1<c2, and the slight adjustment strategy of K2 is inclined;

[0025] The adjustment measures are to maintain the current welding parameters, and the parameters are adjusted to 95% to 105% of the original parameters;

[0026] In interval B [0.25, 0.50), the E(t) value in this interval indicates good welding quality, but there is room for improvement in energy consumption and operating efficiency;

[0027] The screening basis is that the ratio of c1 and c2 is close, at this time |c1-c2|<0.1, combining the medium adjustment strategy of K1 and K2;

[0028] Adjustment measures include adjusting the current or voltage by up to 5% to ensure the optimization of welding parameters to reduce energy waste;

[0029] In the interval C[0.50,0.75), the E(t) value in this interval indicates that there are certain welding quality problems and the energy consumption is high;

[0030] When the screening basis is that the weight of c1 is large, at this time, c1>c2, the moderate adjustment strategy of K1 is adopted;

[0031] The adjustment measures are to significantly adjust the welding parameters, and adjust the welding speed and power by at most 10%;

[0032] In the interval D[0.75,1], the E(t) value in this interval indicates that the welding quality is poor, the energy consumption is serious, and urgent optimization is needed;

[0033] When the screening basis is that the weight of c2 is large, at this time, c1<c2, the strict adjustment strategy of K2 is adopted;

[0034] The adjustment measures include significantly adjusting the welding parameters or considering re-welding, and adjusting the welding speed and power by up to 20% increase or decrease.

[0035] A pipeline welding path planning system, the system is used to execute the pipeline welding path planning method, comprising:

[0036] Data acquisition module: used to obtain the planned path of pipeline welding, and determine the welding parameters of the current welding equipment at each node of the pipeline according to the planned path, and then weld the pipeline joints through the welding equipment, and collect the time series data of the current I(t) and voltage V(t) on the welding equipment, as well as the real-time weld quality data formed at the pipeline joints;

[0037] LSTM building module: used to obtain the collected time series data of current I(t) and voltage V(t), and use it as the input of the long short-term memory network LSTM model to build a dynamic energy consumption model to predict energy consumption;

[0038] The first coefficient generation module is used to obtain weld quality data, which includes weld appearance quality index and weld internal defect index, and determine the evaluation threshold A of the weld appearance quality index, compare and analyze the weld appearance quality index with the evaluation threshold A, and generate a first feedback adjustment coefficient. In pipeline welding path planning, the coefficient is used to provide a first adjustment strategy for welding parameters of welding equipment;

[0039] The second coefficient generation module is used to determine the evaluation threshold B of the internal defect index of the weld, and compare and analyze the internal defect index of the weld with the evaluation threshold B to generate a second feedback adjustment coefficient. In pipeline welding path planning, this coefficient is used to provide a second adjustment strategy for the welding parameters of the welding equipment;

[0040] Model building module: used to obtain the first feedback adjustment coefficient and the second feedback adjustment coefficient, and analyze and process them in combination with the dynamic energy consumption model to build a selection adjustment model. The selection adjustment model is used to provide a screening strategy for the first adjustment strategy and the second adjustment strategy, and calibrate the selected welding parameter adjustment value to minimize the welding energy consumption while ensuring the quality of the weld.

[0041] Compared with the prior art, the beneficial effects of the present invention are as follows: by real-time collection of welding current and voltage data, the dynamic energy consumption model constructed using LSTM can accurately predict the upcoming energy demand, thereby achieving more precise energy control during the welding process; by analyzing the weld appearance quality index and the internal defect index, and comparing them with the set evaluation threshold, specific feedback adjustment coefficients can be generated. These adjustment coefficients not only guide the welding equipment to adjust the welding parameters in real time to adapt to different welding conditions, but also can be optimized according to the energy consumption and weld quality data during the welding process, ultimately achieving the minimization of welding energy consumption and the maximization of weld quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a schematic diagram of the overall method flow of the present invention;

[0043] Figure 2 It is a system module block diagram of the present invention. DETAILED DESCRIPTION

[0044] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0045] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0046] Embodiment 1:

[0047] See also Figure 1 , the present invention provides a technical solution:

[0048] A pipeline welding path planning method, the specific steps include:

[0049] Step S1, obtaining a planned path for pipeline welding, and determining welding parameters of the current welding equipment at each node of the pipeline according to the planned path, and then welding the pipeline joints by the welding equipment, while collecting time series data of current I(t) and voltage V(t) on the welding equipment, and collecting weld quality data formed in real time at the pipeline joints;

[0050] Step S2, obtaining the collected time series data of current I(t) and voltage V(t), and using them as input of a long short-term memory network (LSTM) model to construct a dynamic energy consumption model to predict energy consumption;

[0051] Step S3, obtaining weld quality data, the weld quality data respectively including weld appearance quality index and weld internal defect index, and determining an evaluation threshold A of the weld appearance quality index, comparing and analyzing the weld appearance quality index with the evaluation threshold A, and generating a first feedback adjustment coefficient, which is used to provide a first adjustment strategy for the welding parameters of the welding equipment;

[0052] Step S4: on the basis of step S3, determining the evaluation threshold B of the weld internal defect index, and comparing and analyzing the weld internal defect index with the evaluation threshold B to generate a second feedback adjustment coefficient, which is used to provide a second adjustment strategy for the welding parameters of the welding equipment;

[0053] Step S5, obtain the first feedback adjustment coefficient and the second feedback adjustment coefficient, and analyze and process them in combination with the dynamic energy consumption model to construct a selection adjustment model, which is used to provide a screening strategy for the first adjustment strategy and the second adjustment strategy, and calibrate the selected welding parameter adjustment value to minimize the welding energy consumption while ensuring the quality of the weld.

[0054] Embodiment 2:

[0055] Further explanation is given on the basis of the first embodiment, the time series data of the current I(t) and the voltage V(t) collected are obtained and used as the input of the long short-term memory network LSTM model to construct a dynamic energy consumption model to predict the energy consumption; specifically, the following contents are included:

[0056] The construction of the dynamic energy consumption model includes data preprocessing, establishment of LSTM model, loss function and optimization, model training and verification, and implementation and feedback loop;

[0057] The data preprocessing includes obtaining the current and voltage time series data of the pipeline welding equipment, normalizing, removing noise and outliers, and improving the model training effect and prediction accuracy;

[0058] Specifically, first, the time series data of current I(t) and voltage V(t) on the pipeline welding equipment are obtained, and these data are preprocessed, including normalization and cleaning, and noise and outliers are removed. The purpose of data preprocessing is to improve the efficiency of model training and the accuracy of prediction. The following formula is used:

[0059]

[0060] Where I'(t) and V'(t) represent the normalized current and voltage data;

[0061] The establishment of the LSTM model includes using the LSTM model to process time series data, inputting preprocessed current and voltage data, and predicting optimal current and voltage to reduce energy consumption;

[0062] Specifically, the LSTM model is established using the long short-term memory network LSTM model to process time series data. LSTM can capture the time dependency in the data and is suitable for processing and predicting time-related sequence data. An LSTM network is constructed, in which the input layer receives the processed current and voltage data [I'(t), V'(t)], and the output layer predicts the optimal current and voltage at a certain point in the future to minimize energy consumption. The structure of the LSTM model is set as:

[0063] Input layer: accepts two-dimensional input [I'(t), V'(t)];

[0064] Hidden layer: contains several LSTM units, the specific number is adjusted according to the complexity of the data and experimental results;

[0065] Output layer: Output the predicted optimal current I opt (t+1) and voltage V opt (t+1);

[0066] The loss function and optimization include adjusting the model parameters using a mean square error loss function and an Adam optimization algorithm by minimizing the error between the predicted value and the actual energy consumption;

[0067] Specifically, the training of the loss function and the optimization model is achieved by minimizing the error between the predicted value and the actual energy consumption. The loss function is defined as the mean square error MSE, and the formula is:

[0068]

[0069] Among them, I pred (t) and Vpred (t) are the current and voltage predicted by the model, and I actual (t) and V actual (t) is the actual measured value; the Adam optimization algorithm is used to optimize the model parameters. The Adam optimization algorithm has high efficiency and adaptive characteristics when processing large-scale data sets;

[0070] The model training and verification includes dividing the data set into a training set and a test set, using the training set to train the LSTM model, and verifying the prediction effect of the model through the test set;

[0071] Specifically, the model training and verification uses the preprocessed data set to divide into a training set and a test set. The training set data is used to train the LSTM model, and the test set is used to verify the prediction performance of the model. The model parameters are continuously optimized iteratively until the prediction error on the test set is steadily reduced to the preset requirements. The preset requirements are determined by using data analysis software and based on data analysis;

[0072] The implementation and feedback loop includes deploying the trained model to actual welding operations, inputting real-time current and voltage data, outputting predicted optimal current and voltage settings, and optimizing the model through monitoring and feedback data to achieve a better pipeline welding process;

[0073] Specifically, the implementation and feedback loop deploys the trained model to the actual welding operation, inputs the current and voltage data of the welding process in real time, outputs the predicted optimal current and voltage settings, monitors the comparison between the model prediction results and the actual energy consumption, and collects feedback for continuous optimization and adjustment of the model;

[0074] The weld appearance quality index includes weld geometry parameters, surface quality characteristics and weld brightness parameters, and the weld geometry parameters are marked as JHc, the surface quality characteristics are marked as BMt, and the weld brightness parameters are marked as HFLd;

[0075] The weld geometry parameter JHc uses laser scanner related tools to measure the width, height and depth data of the weld;

[0076] Assume the width, height and depth of the weld are w, h and d respectively, and define the fractional function to represent the weld geometric parameter JHc as follows:

[0077] JHc=(wp 1 ×hp 2 ×dp 3 ) / (wq 1 +hq 2 +dq 3 )

[0078] Among them, p1, p2, p3, q1, q2 and q3 are coefficients for selecting corresponding parameters, corresponding to the weights of width, height and depth in the corresponding formulas. These coefficients can be selected according to the importance of welding quality and actual conditions;

[0079] Set the width, height and depth to [u1,u2], and the coefficients to [u3,u4];

[0080] According to the definition of the formula, the value range of the weld geometry parameter JHc is [u1,u2], where the values ​​of u1,u2c,u3,u4 are set according to the actual situation;

[0081] Different range values ​​will lead to changes in the results of the weld geometry parameter JHc. Through normalized dimension operation, the range of the weld geometry parameter JHc is limited to [0,1]. When the value of JHc is close to 0, it means that the width, height and depth of the weld are small, which will cause the quality of the weld to be poor. When the value of JHc is close to 1, it means that the width, height and depth of the weld are large. Except for special cases, it means that the quality of the weld is good.

[0082] The evaluation threshold A of the weld geometry parameter JHc is set to R1 and R2, and R1 and R2 are within the range of [0,1], R1<R2, and R1 and R2 are located on both sides of the median of the range [0,1], [0, R1] is close to the interval of 0, and [R2, 1] is close to the interval of 1;

[0083] Determination coefficients p1, p2, p3, q1, q2, and q3:

[0084] The selection of coefficients is based on the importance of the weld geometry parameters. For example, the weld width is relatively important for quality, so a larger weight p1 is selected, while the depth is less important for quality, so a smaller weight p3 is selected.

[0085] Determine the value of the coefficient by experiment or data analysis, and find the most suitable coefficient by measuring a batch of welds and comparing them with the corresponding weld quality;

[0086] Determine the value range u1,u2c,u3,u4:

[0087] The selection of the value range should take into account the actual welding conditions and the requirements for weld quality. According to the specifications of the welding equipment and the size of the pipeline, the reasonable value range of the weld width, height and depth should be determined;

[0088] Likewise, the range of values ​​is determined by experiment or data analysis to ensure that the values ​​within the range can cover typical weld quality conditions;

[0089] Regarding the sentence "except for special circumstances, the quality of the weld is good", it is necessary to explain the special circumstances. The "good quality of the weld" here is a conclusion based on experience and observations under typical circumstances, but in actual applications, there are the following special circumstances:

[0090] Special cases: welding requirements or specific application scenarios require specific weld quality requirements. Therefore, for these cases, a larger weld geometry parameter JHc value does not necessarily mean a better quality weld. In this case, the standard for evaluating weld quality needs to be determined according to specific needs.

[0091] The surface quality feature BMt uses visual inspection or machine vision technology to evaluate the weld surface and determine the smoothness, unevenness, and pits;

[0092] Smoothness: The smoothness of the weld surface is measured using a smoothness evaluator or measuring tool. The result is a numerical value that characterizes the level of smoothness and is quantitatively expressed in a range of 1 to 10.

[0093] The unevenness is measured by measuring the unevenness of the weld surface, using a non-contact profiler or surface scanning technology to obtain quantitative data on the unevenness, and the quantitative data is within a range of 1 to 10;

[0094] The potholes are identified on the weld surface by image processing technology or computer vision algorithm, image analysis and edge detection, and the number of potholes is quantified; the surface quality feature BMt is calculated as follows:

[0095] BMt=w1×BM1+w2×BM2+w3×BM3

[0096] Among them, BM1, BM2 and BM3 represent the quantitative values ​​of smoothness, bumpiness and pothole parameters respectively, and w1, w2 and w3 are the corresponding non-negative weight coefficients, and w1+w2+w3=1; the weight coefficients w1, w2 and w3 are set according to actual needs and priorities, and are determined by the expert group through experiments or data analysis;

[0097] Assuming that the levels of smoothness, roughness, and pothole parameters are between 1 and 10, the value range of BMt is set between [0,1] through normalized dimension operation. The closer the value is to 0, the worse the surface quality characteristics are, while the closer the value is to 1, the better the surface quality characteristics are.

[0098] The evaluation threshold A of the surface quality feature BMt is set to R3 and R4, and R3 and R4 are within the range of [0,1], R3<R4, and R3 and R4 are located on both sides of the median of the range [0,1], [0, R3] is close to the interval of 0, and [R4, 1] is close to the interval of 1;

[0099] The weld brightness HFLd is visually evaluated by comparison with the standard, and is obtained as follows;

[0100] Contrast: By comparing the difference between the weld area and the surrounding ambient light, the brightness of the weld is quantified, and the contrast value between the average brightness of the weld area and the background brightness is calculated through image processing algorithms;

[0101] Standard comparison: Compare the weld with a known standard or reference sample to evaluate the brightness level of the weld. The brightness of the weld can be quantitatively described using a comparison standard. The brightness value is compared with the actual prepared standard brightness. The HFLd calculation formula is as follows:

[0102] HFLd=w4×HF1+w5×HF2

[0103] Among them, HF1 represents the weld brightness value obtained according to the contrast method, HF2 represents the weld brightness value obtained according to the standard comparison method, and w4 and w5 are the corresponding weight coefficients;

[0104] In the contrast method, the contrast of the weld brightness is calculated using the following function:

[0105] HF1=(Iw-Ib) / Ib

[0106] Where Iw represents the average brightness of the weld area, and Ib represents the background brightness;

[0107] In the standard comparison method, the following function is used to evaluate the brightness level of the weld:

[0108] HF2=Hs / Hr

[0109] Among them, Hs represents the brightness value of the weld, and Hr represents the standard brightness actually prepared;

[0110] The weight coefficients w4 and w5 are set according to actual needs and priorities to ensure that the importance of each parameter is reasonably considered;

[0111] The range of the weld brightness HFLd is set to [0, +∞), and the dimension operation is normalized so that the range of the weld brightness HFLd is within [0, 1). The value of the range represents the relative size of the brightness. A smaller value represents a lower brightness, while a larger value represents a higher brightness. The threshold or range for darker conditions and brighter brightness is set. A lower value is regarded as a darker weld brightness, while a higher value represents a brighter weld brightness. Different range values ​​will lead to different technical effects. The welding parameters are automatically adjusted based on the range of HFLd to achieve better weld quality.

[0112] The evaluation threshold A of the weld brightness HFLd is set to R5 and R6, and R5 and R6 are within the value range [0,1), R5<R6, and R5 and R6 are located on both sides of the median of the value range [0,1), [0, R5] is a smaller interval close to 0, and [R6, 1) is a larger interval close to 1.

[0113] Embodiment three:

[0114] Further explanation based on the second embodiment, the weld appearance quality index is compared and analyzed with the evaluation threshold A to generate a first feedback adjustment coefficient, which is used to provide a first adjustment strategy for the welding parameters of the welding equipment; specifically includes the following contents:

[0115] The first feedback adjustment coefficient is defined as K1, and the calculation formula is as follows:

[0116]

[0117] Among them, JHc is the weld geometry parameter, BMt is the surface quality feature, HFLd is the weld brightness parameter, R1, R2 are the evaluation thresholds of JHc, which are the lower and higher standards respectively, set in the range of [0,1]; R3, R4 are the evaluation thresholds of BMt, also set in the range of [0,1]; R5, R6 are the evaluation thresholds of HFLd, set in the range of [0,1);

[0118] The G(x,y,z) function is a normalized Gaussian function used to measure the index x and its threshold center distance and adjust its sensitivity by the standard deviation σ(x,y,z);

[0119] K1 is the first feedback adjustment coefficient, and its value range is between [0,1]. The closer the value is to 1, the closer the welding quality is to the target threshold. The closer the value is to 0, the worse the quality is, and a larger adjustment is needed.

[0120] The σ(x,y,z) function is used to calculate the standard deviation based on the evaluation threshold range to ensure that the G(x,y,z) function has appropriate sensitivity and response range. By setting σ to one-sixth of the threshold range, it is ensured that G(x,y,z) changes faster near the threshold boundary, thereby enhancing the responsiveness of the model.

[0121] Divide the value range of K1 [0,1] into specific intervals, namely major adjustment interval [0,0.2), medium adjustment interval [0.2,0.5), minor adjustment interval [0.5,0.8) and optimization interval [0.8,1.0], and set clear action strategies and performance levels for each interval;

[0122] The first adjustment strategy includes the following contents:

[0123] For the interval [0,0.2), it is a major adjustment area, which means that the welding quality is very poor and far away from the expected threshold; the welding quality is less than 20%, and a major adjustment is required; specifically, the current is increased by 30% to enhance the welding depth, and the voltage is increased by 25% to improve the stability of the molten pool;

[0124] Check equipment accuracy and operating parameters, reset welding path; JHc, BMt, HFLd values ​​deviate significantly from expectations, requiring systematic investigation and 10% incremental adjustment;

[0125] For the interval [0.2, 0.5), it is a medium adjustment area, which indicates that the welding quality needs to be improved, and the welding quality is between 20% and 50%; adjust 10% of the current welding speed and 13% of the temperature control parameters; increase the values ​​of JHc, BMt, and HFLd by 12%, and then approach the corresponding positive correlation evaluation threshold A;

[0126] For the interval [0.5, 0.8), it is a slight adjustment area. This interval indicates that the welding quality is good, but there is still room for improvement. The welding quality is between 50% and 80%. Fine-tune the welding parameters. Specifically, increase the current by 10% to fine-tune the welding quality and increase the voltage by 5% to optimize the molten pool dynamics to further improve the quality. Adjust the values ​​of 5% JHc, BMt, and HFLd to make the quality close to perfect.

[0127] For the interval [0.8,1.0], it is the optimization zone, the welding quality is very close to or reaches the target, the welding quality exceeds 80%, close to the ideal state, keep the current parameters, conduct routine monitoring, the values ​​of JHc, BMt, and HFLd are in the ideal state, and the status quo can be maintained;

[0128] When any of the parameters JHc, BMt, or HFLd increases, it means that a certain quality indicator of the weld is improved, thereby increasing K1, indicating that the overall welding quality is developing in a good direction. Conversely, a decrease in any parameter will usually lead to a decrease in K1, indicating that more attention and adjustment are needed; if JHc is reduced due to insufficient width, height or depth, then K1 will decrease accordingly, reflecting a decrease in welding performance.

[0129] Embodiment 4:

[0130] Further explanation based on the third embodiment, the internal defect index of the weld includes the porosity, slag inclusion and crack defect inside the weld, and the porosity is marked as Qk, the slag inclusion is marked as JZd, and the crack defect is marked as LWq;

[0131] For the porosity Qk, pores are the pores inside the weld, which are formed by gas trapped during the welding process. The porosity refers to the number or density of pores inside the weld, reflecting the degree of porosity inside the weld. The proportion of pores in the cross-sectional area of ​​the weld is used to express the porosity, expressed as a percentage or the number per unit area. Ultrasonic testing is a non-destructive testing method that can detect the porosity inside the weld through the propagation of sound waves. By analyzing the ultrasonic echo signal, the location, size and number of pores can be determined. The calculation formula for the porosity Qk is as follows:

[0132]

[0133] Among them, Agas is the area of ​​pores in the cross-sectional area of ​​the weld; Atotal is the total area of ​​the cross-sectional area of ​​the weld;

[0134] The evaluation threshold B for the porosity Qk is E1. E1 needs to take into account the requirements and standards of the welding application. A higher porosity will affect the performance of the weld. Therefore, the evaluation threshold is usually set to a relatively small value. For general welding applications, the evaluation threshold E1 is set to 5%.

[0135] Quantitative explanation: The evaluation threshold E1 is a percentage, which indicates the maximum allowable proportion of pores in the cross-sectional area of ​​the weld. If the total area of ​​the weld is 100 square millimeters, then the evaluation threshold E1 is 5%, which means that the allowable pore area in the weld cannot exceed 5 square millimeters.

[0136] If the welded joint is used for critical components or high-pressure applications, the evaluation threshold of the porosity should be set relatively low to ensure the tightness and strength of the welded joint. In this case, the evaluation threshold E1 is set to a smaller percentage, such as 1% to 3%;

[0137] On the contrary, if the welded joint is used for general structural parts or non-critical applications, a higher amount of porosity can be tolerated. In this case, the evaluation threshold E1 will be relatively high, such as 5% to 10%;

[0138] In practical applications, relevant welding standards and specifications should also be considered. These standards will provide recommendations or requirements for specific porosity evaluation thresholds for different application scenarios. Therefore, when determining the evaluation threshold E1, it is necessary to comprehensively consider the specific requirements of the welding application and the applicable standards and specifications;

[0139] For the slag inclusion degree JZd, slag inclusion refers to the unmelted flux or other external impurities in the weld during the welding process. The slag inclusion degree is a parameter that indicates the degree of slag inclusion in the weld. It is expressed by the proportion of slag inclusion in the cross-sectional area of ​​the weld, expressed as a percentage or the number per unit area. X-ray detection is used to detect the slag inclusion in the weld. By analyzing the X-ray photos, the location, size and distribution of the slag inclusion can be determined. The calculation formula of the slag inclusion degree JZd is as follows:

[0140]

[0141] Among them, Bslag is the area of ​​slag inclusion in the weld cross-sectional area, and Btotal is the total area of ​​the weld cross-sectional area;

[0142] The evaluation threshold B for the slag inclusion degree JZd is E2, which needs to be determined based on welding materials, application environment and relevant standards; welding standards will specify the maximum allowable proportion of slag inclusion;

[0143] When selecting the evaluation threshold E2, the welding materials used should be considered first. Different types of welding materials have different tolerances to slag inclusions. Some welding materials are more likely to produce slag inclusions, so a lower evaluation threshold needs to be set; while other welding materials are relatively less likely to produce slag inclusions, so a higher evaluation threshold needs to be set.

[0144] Secondly, the application environment is also an important factor in determining the evaluation threshold. If the welded joint will work in a harsh environment, such as high temperature, high pressure or corrosive environment, then the evaluation threshold E2 should be set lower accordingly to ensure the durability and reliability of the welded joint.

[0145] Finally, it is necessary to consider relevant welding standards and specifications, which will provide specific requirements or recommendations on the degree of slag inclusion to ensure that the welded joint meets the standards of a specific industry or product; therefore, when determining the evaluation threshold E2, the applicable standards and specifications should be referred to and adjusted according to actual conditions;

[0146] Quantitative explanation: The evaluation threshold E2 is a percentage, which indicates the maximum allowable proportion of slag inclusion in the cross-sectional area of ​​the weld. If the total area of ​​the weld is 100 square millimeters, then the evaluation threshold E2 is 3%, which means that the allowable slag inclusion area in the weld cannot exceed 3 square millimeters.

[0147] For the crack defect LWq, crack refers to the fracture or crack inside the weld, which is caused by factors such as stress concentration during welding, excessive cooling rate or material properties. The crack defect is a parameter indicating the degree of cracks in the weld, which is expressed by the proportion of cracks in the cross-sectional area of ​​the weld. In quantitative description, the crack defect can be defined as the total volume of cracks in the cross-sectional area of ​​the weld, expressed as a percentage or the number per unit area. Ultrasonic testing is also used to detect cracks inside the weld. By analyzing the ultrasonic echo signal, the location, size and distribution of the crack are determined; the calculation formula of the crack defect LWq is as follows:

[0148]

[0149] Where Vcrack is the total volume of cracks in the weld cross-sectional area; Vtotal is the total volume of the weld cross-sectional area;

[0150] The evaluation threshold B for crack defects LWq is E3, which needs to take into account the stress, use environment and relevant standards of the welded parts; smaller cracks are easily tolerated, but large or dense cracks need to be repaired or re-welded;

[0151] First, the stress to which the welded parts are subjected needs to be considered. If the welded joint will be subjected to large forces or vibrations, the evaluation threshold E3 should be set lower to ensure that the welded joint can withstand these stresses without cracking.

[0152] Secondly, the application environment will also affect the setting of the evaluation threshold. If the welded joint will work in a high temperature, low temperature or corrosive environment, the evaluation threshold E3 should be adjusted according to the actual situation to ensure the durability and reliability of the welded joint.

[0153] Finally, relevant welding standards and specifications are also important references for determining the evaluation threshold. These standards usually provide specific requirements or recommendations for crack defects to ensure that the welded joints meet the standards of specific industries or products. Therefore, when determining the evaluation threshold E3, the applicable standards and specifications should be referred to and adjusted according to the actual situation.

[0154] Quantitative explanation: The evaluation threshold E3 is a percentage, which indicates the maximum allowable proportion of cracks in the cross-sectional area of ​​the weld. If the total volume of the weld is 100 cubic millimeters, then the evaluation threshold E3 of 1% means that the total volume of cracks allowed in the weld cannot exceed 1 cubic millimeter.

[0155] The method compares and analyzes the internal defect index of the weld with the evaluation threshold B to generate a second feedback adjustment coefficient, which is used to provide a second adjustment strategy for the welding parameters of the welding equipment; specifically includes the following contents:

[0156] The second feedback adjustment coefficient is defined as K2, and the calculation formula is as follows:

[0157]

[0158] Among them, H(Qk,E1), H(JZd,E2), H(LWq,E3) are all normalized functions, and K2 is the second feedback adjustment coefficient;

[0159] Qk is the porosity; JZd is the slag inclusion degree, LWq is the crack defect, E1, E2, and E3 are the evaluation thresholds B of the porosity, slag inclusion degree, and crack defect, respectively;

[0160] H(x,Ei) is a normalization function used to convert the relationship between the welding defect parameter x and the evaluation threshold Ei into a value between 0 and 1, indicating the relative severity of the defect degree;

[0161] Set the value range of formula K2 to [0,1-e -3 ], take four decimal places, specifically [0,0.9502];

[0162] When Qk, JZd, and LWq are close to 0, the weld is close to perfect and defect-free. At this time, K2 is close to 0, indicating that no adjustment is required.

[0163] When Qk, JZd, and LWq are close to their evaluation thresholds E1, E2, and E3, K2 is close to its maximum value, indicating that a larger adjustment is needed to improve the welding quality;

[0164] The value range of K2 is divided into several specific intervals, namely interval A[0,0.25), interval B[0.25,0.50), interval C[0.50,0.75), interval D[0.75,1-e -3 ]; The second adjustment strategy includes the following contents:

[0165] The interval division and description of the second feedback adjustment coefficient K2 are as follows:

[0166] For the interval A[0,0.25), K2 in this interval indicates that the welding quality is close to perfect and the defect level is extremely low; maintain the current welding parameters without adjustment; when any increase in Qk, JZd, or LWq is within 5% of the current value, K2 will increase slightly, but still remain in this interval, indicating that the welding quality is still within an acceptable range;

[0167] For interval B [0.25, 0.50), K2 indicates that the welding quality is good but there are slight defects in this interval; slightly adjust the welding parameters to reduce the defects, and specifically adjust the welding speed, power or material flow parameters to increase or decrease by at most 5%;

[0168] If the value of Qk, JZd, or LWq is close to its evaluation threshold, E1, E2, E3, and K2 are increased and close to interval C, optimizing the weld quality, reducing the formation of pores and slag inclusions, and making fine adjustments to combat minor defects;

[0169] For interval C [0.50, 0.75), K2 in this interval indicates that the welding quality is moderate and the defects that need attention begin to become significant; moderate adjustments are made to the welding parameters to significantly reduce the defects, specifically adjusting the welding speed, power or material flow rate to increase or decrease by at most 10% to 15%; if the values ​​of Qk, JZd, or LWq increase further, K2 will enter interval D, requiring more stringent adjustments. Through moderate adjustments, the weld quality is significantly improved, the significant pores, slag inclusions and cracks are reduced, and the overall performance of the weld is improved;

[0170] For the interval D[0.75,1-e -3], K2 in this range indicates that the welding quality is poor and there are serious defects. It is necessary to make major adjustments to the welding parameters or consider re-welding. The specific adjustment of welding speed, power or material flow rate should be increased or decreased by 20% to 25% at most; K2 is close to the maximum value, indicating that the severity of the defect is extremely high, and the welding parameters should be adjusted urgently;

[0171] The calculation of K2 depends on the relationship between Qk, JZd, LWq and their respective thresholds E1, E2, E3; an increase in each defect index will lead to an increase in K2, thereby pushing the welding parameters toward a more stringent adjustment direction. Conversely, reducing these index values ​​will reduce the K2 value, causing the welding parameters to be adjusted back to a more relaxed state.

[0172] Embodiment five:

[0173] Further explanation is given on the basis of the fourth embodiment, obtaining the first feedback adjustment coefficient and the second feedback adjustment coefficient, and analyzing and processing them in combination with the dynamic energy consumption model to construct a selection adjustment model, specifically including the following contents:

[0174] The selection adjustment model is defined as E(t), and the calculation formula is as follows:

[0175] E(t)=c1·K1(t)+c2·K2(t)

[0176] K1(t) and K2(t) are the first feedback adjustment coefficient and the second feedback adjustment coefficient based on the long short-term memory network LSTM model, respectively. t represents the time series label, which is a continuous time series. K1(t) and K2(t) will change with time, reflecting the change of the second feedback adjustment coefficient at different time points.

[0177] c1 and c2 are non-negative weight parameters, and c1+c2=1, which are adjusted according to the sensitivity of the actual adjustment strategy;

[0178] The value range of E(t) is set to [0,1]; different intervals of E(t) values ​​will correspond to different adjustment strategies;

[0179] Divide the value range of E(t) [0,1] into the following intervals:

[0180] In the interval A[0,0.25), the E(t) value in this interval indicates that the welding quality and energy consumption optimization have reached the ideal state, and the adjustment strategy is relatively slight or maintains the status quo;

[0181] The screening basis is that when the weight of c1 is small, the slight adjustment strategy of K2 is preferred;

[0182] The adjustment measures are to maintain the current welding parameters, and the parameters are adjusted to 95% to 105% of the original parameters;

[0183] In interval B [0.25, 0.50), the E(t) value in this interval indicates good welding quality, but there is room for improvement in energy consumption and operating efficiency;

[0184] The screening basis is that the ratio of c1 and c2 is close, combined with the medium adjustment strategy of K1 and K2;

[0185] Adjustment measures include adjusting the current or voltage by up to 5% to ensure the optimization of welding parameters to reduce energy waste;

[0186] In the interval C[0.50,0.75), the E(t) value in this interval indicates that there are certain welding quality problems and the energy consumption is high;

[0187] When the screening basis is that the weight of c1 is large, the moderate adjustment strategy of K1 is adopted;

[0188] The adjustment measures are to significantly adjust the welding parameters, such as adjusting the welding speed, welding direction angle, welding point spacing, and power by up to 10%;

[0189] In the interval D[0.75,1], the E(t) value in this interval indicates that the welding quality is poor, the energy consumption is serious, and urgent optimization is needed;

[0190] When the screening basis is that the weight of c2 is large, the strict adjustment strategy of K2 is adopted;

[0191] The adjustment measures include significantly adjusting the welding parameters or considering re-welding, and adjusting the welding speed and power by up to 20%;

[0192] The selection and calibration of the adjustment strategy is based on the weight ratio of c1 and c2, quantified by the mathematical expression:

[0193] The formula for selecting the adjustment strategy is as follows:

[0194]

[0195] Among them, SXBl is the screening ratio;

[0196] The adjustment range is calculated as follows:

[0197] TZc=YCs×(1±SXBl×TZBl)

[0198] Among them, TZc is the adjustment parameter, YCs is the original parameter, SXBl is the screening ratio, and TZBl is the adjustment ratio.

[0199] Embodiment six:

[0200] Further illustrating on the basis of Example 5, the current I(t) and voltage V(t) data under different welding conditions are collected, and the long short-term memory network LSTM model is applied to predict the first feedback adjustment coefficient K1(t) and the second feedback adjustment coefficient K2(t), and then the welding parameters are adjusted by selecting the adjustment model E(t) according to these predicted values;

[0201] Equipment and materials: Use standard industrial welding robots equipped with sensors to collect current and voltage data in real time. The welding material is common low-carbon steel, and the welding environment is in a standardized industrial workshop.

[0202] The time series data of current I(t) and voltage V(t) are continuously recorded by the sensors built into the welding equipment;

[0203] LSTM model training: The LSTM model is trained using historical welding data to predict K1(t) and K2(t). The model training takes into account the historical data of welding speed, current, voltage, and weld quality.

[0204] According to K1(t) and K2(t) output by the LSTM model, the selection adjustment model E(t) = c1·K1(t) + c2·K2(t) is used for real-time parameter adjustment. The weights c1 and c2 are set according to the previous simulation study to optimize the welding quality and energy consumption.

[0205] Before the experiment begins, preheat the welding equipment to ensure that all equipment is working in a stable state, and record the welding parameters before adjustment as a baseline;

[0206] Collect data in real time during the welding process and calculate K1 and K2 through the trained LSTM model;

[0207] According to the value of E(t), the welding parameters are adjusted dynamically. If the E(t) value falls below 0.2, the current and voltage are greatly increased; if the E(t) value is between 0.2 and 0.5, the welding speed and temperature control parameters are appropriately adjusted;

[0208] After each adjustment, evaluate the weld quality, including determining the structural integrity and strength of the weld through X-ray inspection and mechanical testing;

[0209] Record the experimental data and adjustment results for further optimization of the LSTM model and selection of adjustment model parameters;

[0210] The following is a simulated data table, which records the relevant parameters and welding quality evaluation results of the experiment under different E(t) value ranges;

[0211] Table 1

[0212]

[0213] The table data analysis is as follows:

[0214] Regarding the relationship between energy consumption and welding quality:

[0215] As can be seen from the table, with the increase of E(t) value, the actual energy consumption gradually decreases from 15kW in welding 1 to 6kW in welding 5. At the same time, the welding quality score also improves significantly from 18% to 98%. This inverse relationship clearly shows that the invention achieves the dual goals of energy saving and quality improvement in effectively regulating the welding process.

[0216] Optimization strategy for welding parameters:

[0217] By comparing the current, voltage, welding speed and temperature adjustments in the welding sequence, we observed the differences in the effects of different adjustment strategies. Welding 1 and Welding 2 took larger current and voltage adjustments, while Welding 4 and Welding 5 had almost no adjustments, indicating that when the E(t) value was high, the welding parameters were close to the ideal state and did not need too much adjustment. This shows that the optimal welding parameters can be automatically recommended according to the actual energy efficiency, reducing human intervention and improving production efficiency and quality.

[0218] Intuitive description of welding quality:

[0219] The model can automatically analyze the current welding status and energy consumption and recommend the optimal adjustment strategy accordingly. This model not only optimizes the welding process, but also ensures the welding quality while minimizing energy consumption through real-time calibration parameters.

[0220] Embodiment seven:

[0221] See also Figure 2 , a pipeline welding path planning system, the system is used to execute the pipeline welding path planning method, comprising:

[0222] Data acquisition module: used to obtain the planned path of pipeline welding, and determine the welding parameters of the current welding equipment at each node of the pipeline according to the planned path, and then weld the pipeline joints through the welding equipment, and collect the time series data of the current I(t) and voltage V(t) on the welding equipment, as well as the real-time weld quality data formed at the pipeline joints;

[0223] LSTM building module: used to obtain the collected time series data of current I(t) and voltage V(t), and use it as the input of the long short-term memory network LSTM model to build a dynamic energy consumption model to predict energy consumption;

[0224] The first coefficient generation module is used to obtain weld quality data, which includes weld appearance quality index and weld internal defect index, and determine the evaluation threshold A of the weld appearance quality index, compare and analyze the weld appearance quality index with the evaluation threshold A, and generate a first feedback adjustment coefficient. In pipeline welding path planning, the coefficient is used to provide a first adjustment strategy for welding parameters of welding equipment;

[0225] The second coefficient generation module is used to determine the evaluation threshold B of the internal defect index of the weld, and compare and analyze the internal defect index of the weld with the evaluation threshold B to generate a second feedback adjustment coefficient. In pipeline welding path planning, this coefficient is used to provide a second adjustment strategy for the welding parameters of the welding equipment;

[0226] Model building module: used to obtain the first feedback adjustment coefficient and the second feedback adjustment coefficient, and analyze and process them in combination with the dynamic energy consumption model to build a selection adjustment model. The selection adjustment model is used to provide a screening strategy for the first adjustment strategy and the second adjustment strategy, and calibrate the selected welding parameter adjustment value to minimize the welding energy consumption while ensuring the quality of the weld.

[0227] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0228] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0229] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0230] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A pipeline welding path planning method, characterized in that: The specific steps include: Step S1, obtaining a planned path for pipeline welding, and determining welding parameters of the current welding equipment at each node of the pipeline according to the planned path, and then welding the pipeline joints by the welding equipment, while collecting time series data of current I(t) and voltage V(t) on the welding equipment, and collecting weld quality data formed in real time at the pipeline joints; Step S2, obtaining the collected time series data of current I(t) and voltage V(t), and using them as input of a long short-term memory network (LSTM) model to construct a dynamic energy consumption model to predict energy consumption; The construction of the dynamic energy consumption model includes data preprocessing, establishment of LSTM model, loss function and optimization, model training and verification, and implementation and feedback loop; The data preprocessing includes obtaining the current and voltage time series data of the pipeline welding equipment, normalizing, removing noise and outliers, and improving the model training effect and prediction accuracy; The establishment of the LSTM model includes using the LSTM model to process time series data, inputting preprocessed current and voltage data, and predicting optimal current and voltage to reduce energy consumption; The loss function and optimization include adjusting the model parameters using a mean square error loss function and an Adam optimization algorithm by minimizing the error between the predicted value and the actual energy consumption; The model training and verification includes dividing the data set into a training set and a test set, using the training set to train the LSTM model, and verifying the prediction effect of the model through the test set; The implementation and feedback loop includes deploying the trained model to actual welding operations, inputting real-time current and voltage data, outputting predicted optimal current and voltage settings, and optimizing the model through monitoring and feedback data to achieve a better pipeline welding process; Step S3, obtaining weld quality data, the weld quality data respectively including weld appearance quality index and weld internal defect index, and determining the evaluation threshold A of the weld appearance quality index, comparing and analyzing the weld appearance quality index with the evaluation threshold A, and generating a first feedback adjustment coefficient. In pipeline welding path planning, the coefficient is used to provide a first adjustment strategy for welding parameters of welding equipment; The weld appearance quality index includes weld geometry parameters, surface quality characteristics and weld brightness parameters, and the weld geometry parameters are marked as JHc, the surface quality characteristics are marked as BMt, and the weld brightness parameters are marked as HFLd; The first feedback adjustment coefficient is defined as K1, and the calculation formula is as follows: Among them, JHc is the weld geometry parameter, BMt is the surface quality feature, HFLd is the weld brightness parameter, R1, R2 are the evaluation thresholds of JHc, which are the lower and higher standards respectively, set in the range of [0,1]; R3, R4 are the evaluation thresholds of BMt, also set in the range of [0,1]; R5, R6 are the evaluation thresholds of HFLd, set in the range of [0,1); The G(x,y,z) function is a normalized Gaussian function used to measure the index x and its threshold center distance and adjust its sensitivity by the standard deviation σ(x,y,z); K1 is the first feedback adjustment coefficient, and its value range is between [0,1]. The closer the value is to 1, the closer the welding quality is to the target threshold. The closer the value is to 0, the worse the quality is, and a larger adjustment is needed. Divide the value range of K1 [0,1] into specific intervals, namely major adjustment interval [0,0.2), medium adjustment interval [0.2,0.5), minor adjustment interval [0.5,0.8) and optimization interval [0.8,1.0], and set clear action strategies and performance levels for each interval; For the interval [0,0.2), it is a major adjustment area, and this interval indicates that the welding quality is very poor; For the interval [0.2,0.5), it is the medium adjustment area, and this interval indicates that the welding quality needs to be improved; For the interval [0.5, 0.8), it is a slight adjustment area, and this interval indicates good welding quality; For the interval [0.8, 1.0], it is the optimization zone, and the welding quality is very close to or reaches the target; Step S4: Based on step S3, determine the evaluation threshold B of the weld internal defect index, compare and analyze the weld internal defect index with the evaluation threshold B, and generate a second feedback adjustment coefficient. In pipeline welding path planning, the coefficient is used to provide a second adjustment strategy for the welding parameters of the welding equipment; The weld internal defect index includes the amount of porosity, slag inclusion, and crack defects inside the weld, and the amount of porosity is marked as Qk, the slag inclusion is marked as JZd, and the crack defect is marked as LWq; The calculation formula of pore volume Qk is as follows: Among them, Agas is the area of ​​pores in the cross-sectional area of ​​the weld; Atotal is the total area of ​​the cross-sectional area of ​​the weld; the evaluation threshold B for the amount of pores Qk is E1; The calculation formula of slag inclusion degree JZd is as follows: Among them, Bslag is the area of ​​slag inclusion in the cross-sectional area of ​​the weld, and Btotal is the total area of ​​the cross-sectional area of ​​the weld; the evaluation threshold B for the slag inclusion degree JZd is E2; The calculation formula of the crack defect LWq is as follows: Wherein, Vcrack is the total volume of cracks in the cross-sectional area of ​​the weld; Vtotal is the total volume of the cross-sectional area of ​​the weld; the evaluation threshold B for the crack defect LWq is E3; The second feedback adjustment coefficient is defined as K2, and the calculation formula is as follows: Among them, H(Qk,E1), H(JZd,E2), H(LWq,E3) are all normalized functions, and K2 is the second feedback adjustment coefficient; Qk is the porosity; JZd is the slag inclusion degree, LWq is the crack defect, E1, E2, and E3 are the evaluation thresholds B of the porosity, slag inclusion degree, and crack defect, respectively; Set the value range of formula K2 to [0,1-e -3 ], when Qk, JZd, and LWq are close to 0, it is close to a perfect and defect-free weld, and K2 is close to 0, indicating that no adjustment is required; When Qk, JZd, and LWq are close to their evaluation thresholds E1, E2, and E3, K2 is close to its maximum value, indicating that a larger adjustment is needed to improve the welding quality; The value range of K2 is divided into several specific intervals, namely interval A[0,0.25), interval B[0.25,0.50), interval C[0.50,0.75), interval D[0.75,1-e -3 ]; Step S5, obtaining the first feedback adjustment coefficient and the second feedback adjustment coefficient, and analyzing and processing them in combination with the dynamic energy consumption model to construct a selection adjustment model, the selection adjustment model is used to provide a screening strategy for the first adjustment strategy and the second adjustment strategy, and calibrate the selected welding parameter adjustment value to minimize the welding energy consumption and ensure the quality of the weld; The first feedback adjustment coefficient and the second feedback adjustment coefficient are obtained, and are analyzed and processed in combination with the dynamic energy consumption model to construct a selection adjustment model, which specifically includes the following contents: The selection adjustment model is defined as E(t), and the calculation formula is as follows: E(t)=c1·K1(t)+c2·K2(t) K1(t) and K2(t) are the first feedback adjustment coefficient and the second feedback adjustment coefficient based on the long short-term memory network LSTM model, respectively. t represents the time series label, which is a continuous time series. c1 and c2 are non-negative weight parameters, and c1+c2=1, which are adjusted according to the sensitivity of the actual adjustment strategy; the value range of E(t) is set to [0,1]; Divide the value range of E(t) [0,1] into the following intervals: In the interval A[0,0.25), the E(t) value in this interval indicates that the welding quality and energy consumption optimization have reached the ideal state, and the adjustment strategy is relatively slight or maintains the status quo; The screening basis is when the weight of c1 is small, at this time c1<c2, and the slight adjustment strategy of K2 is inclined; The adjustment measures are to maintain the current welding parameters, and the parameters are adjusted to 95% to 105% of the original parameters; In interval B [0.25, 0.50), the E(t) value in this interval indicates good welding quality, but there is room for improvement in energy consumption and operating efficiency; The screening basis is that the ratio of c1 and c2 is close, at this time |c1-c2|<0.1, combining the medium adjustment strategy of K1 and K2; Adjustment measures include adjusting the current or voltage by up to 5% to ensure the optimization of welding parameters to reduce energy waste; In the interval C[0.50,0.75), the E(t) value in this interval indicates that there are certain welding quality problems and the energy consumption is high; When the screening basis is that the weight of c1 is large, at this time, c1>c2, the moderate adjustment strategy of K1 is adopted; The adjustment measures are to significantly adjust the welding parameters, and adjust the welding speed and power by at most 10%; In the interval D[0.75,1], the E(t) value in this interval indicates that the welding quality is poor, the energy consumption is serious, and urgent optimization is needed; When the screening basis is that the weight of c2 is large, at this time, c1<c2, the strict adjustment strategy of K2 is adopted; The adjustment measures include significantly adjusting the welding parameters or considering re-welding, and adjusting the welding speed and power by up to 20% increase or decrease.

2. A pipeline welding path planning method according to claim 1, characterized in that: The value range of the weld geometry parameter JHc is limited to [0,1]; when the value of JHc is close to 0, it means that the width, height and depth of the weld are small, which will cause the quality of the weld to be poor; when the value of JHc is close to 1, it means that the width, height and depth of the weld are large, which means the quality of the weld is good; The evaluation threshold A of the weld geometric parameter JHc is set to R1 and R2, and R1 and R2 are within the value range [0,1], R1<R2, and R1 and R2 are located on both sides of the median of the value range [0,1], [0, R1] is close to the 0 interval, and [R2, 1] is close to the 1 interval.

3. A pipeline welding path planning method according to claim 2, characterized in that: The value range of BMt is set to be between [0, 1], the closer the value is to 0, the worse the surface quality characteristics are, and the closer the value is to 1, the better the surface quality characteristics are; The evaluation threshold A of the surface quality feature BMt is set to R3 and R4, and R3 and R4 are within the range of [0,1], R3<R4, and R3 and R4 are located on both sides of the median of the range [0,1], [0, R3] is close to the interval of 0, and [R4, 1] is close to the interval of 1; The range of the weld brightness HFLd is set to [0, +∞), and a dimension normalization operation is performed on it so that the range of the weld brightness HFLd is within [0, 1). A smaller value indicates a lower brightness, while a larger value indicates a higher brightness. The evaluation threshold A of the weld brightness HFLd is set to R5 and R6, and R5 and R6 are within the value range [0,1), R5<R6, and R5 and R6 are located on both sides of the median of the value range [0,1), [0, R5] is a smaller interval close to 0, and [R6, 1) is a larger interval close to 1.

4. A pipeline welding path planning system, characterized in that: The system is used to execute the pipeline welding path planning method according to any one of claims 1 to 3, comprising: Data acquisition module: used to obtain the planned path of pipeline welding, and determine the welding parameters of the current welding equipment at each node of the pipeline according to the planned path, and then weld the pipeline joints through the welding equipment, and collect the time series data of the current I(t) and voltage V(t) on the welding equipment, as well as the real-time weld quality data formed at the pipeline joints; LSTM building module: used to obtain the collected time series data of current I(t) and voltage V(t), and use it as the input of the long short-term memory network LSTM model to build a dynamic energy consumption model to predict energy consumption; The first coefficient generation module is used to obtain weld quality data, which includes weld appearance quality index and weld internal defect index, and determine the evaluation threshold A of the weld appearance quality index, compare and analyze the weld appearance quality index with the evaluation threshold A, and generate a first feedback adjustment coefficient. In pipeline welding path planning, the coefficient is used to provide a first adjustment strategy for welding parameters of welding equipment; The second coefficient generation module is used to determine the evaluation threshold B of the internal defect index of the weld, and compare and analyze the internal defect index of the weld with the evaluation threshold B to generate a second feedback adjustment coefficient. In pipeline welding path planning, this coefficient is used to provide a second adjustment strategy for the welding parameters of the welding equipment; Model building module: used to obtain the first feedback adjustment coefficient and the second feedback adjustment coefficient, and analyze and process them in combination with the dynamic energy consumption model to build a selection adjustment model. The selection adjustment model is used to provide a screening strategy for the first adjustment strategy and the second adjustment strategy, and calibrate the selected welding parameter adjustment value to minimize the welding energy consumption while ensuring the quality of the weld.