Methods, devices, electronic equipment, and storage media for optimizing steel pipe sawing volume

Abstract

This invention relates to the field of rolled steel pipe material utilization optimization technology, and particularly to a method, apparatus, electronic device, and storage medium for optimizing steel pipe sawing amount. The method of optimizing steel pipe sawing amount first obtains a first parameter dataset; then, it uses a dimension transformation array to convert the first parameter dataset into a first vector, and selects the class closest to the first vector from multiple vector classes as the belonging class; next, it substitutes the first parameter dataset into the wall thickness prediction model corresponding to the belonging class to obtain a wall thickness distribution dataset; finally, it optimizes the steel pipe sawing amount based on the wall thickness distribution dataset. This invention completes a coarse classification of sawing amount based on the factors affecting steel pipe wall thickness, and then obtains the wall thickness distribution prediction result by inputting the factors affecting steel pipe wall thickness into the wall thickness prediction model of the vector class. The model prediction result is easily guaranteed, improving the steel pipe yield.

Description

Technical Field

[0001] This invention relates to the field of material utilization optimization technology for rolled steel pipes, and in particular to a method, apparatus, electronic device, and storage medium for optimizing steel pipe sawing quantity. Background Technology

[0002] After seamless steel pipes are rolled, the defective portions at both ends are cut off. The main problem with the cut-off portions is wall thickness deviation. Currently, there are two types of length-fixing methods in steel pipe production: absolute length-fixing and range-fixing.

[0003] For absolute length requirements, the length deviation of steel pipes must be between 50-100mm. During sawing, absolute precision in length must be ensured. Finished pipes of uniform length are more likely to meet the requirements, and generally, the middle value is chosen for sawing. For range-based length requirements, a larger deviation is permissible, sometimes exceeding 2000mm. In these cases, a larger length is typically chosen for sawing, reducing the amount of material cut and increasing the weight of the finished pipe to improve the yield. The latter situation accounts for 70% of the total steel pipe production.

[0004] Currently, the cutting amount for each group of steel pipes is determined on the production site based on manual sampling of the wall thickness on the cooling bed. This amount is then manually input into the HMI (Head and Shoulder Meter) setting for the pipe cutting amount on the pipe saw, ensuring that the steel pipes are within the length tolerance range specified in the order. On-site investigation revealed that manual sampling cannot accurately determine the wall thickness deviation range for each steel pipe, and the given cutting amount easily leads to overcutting at the beginning and end, reducing the product yield.

[0005] Therefore, it is necessary to develop and design a method for optimizing the amount of steel pipe sawing. Summary of the Invention

[0006] The present invention provides a method, apparatus, electronic device and storage medium for optimizing steel pipe sawing quantity, which solves the problem of low yield in the prior art when the steel pipe sawing quantity is determined by manual sampling.

[0007] In a first aspect, embodiments of the present invention provide a method for optimizing the amount of steel pipe sawing, comprising:

[0008] Obtain a first parameter dataset, wherein the first parameter dataset includes multiple material parameters that affect the wall thickness of the steel pipe and / or multiple processing parameters that affect the wall thickness of the steel pipe;

[0009] The first parameter dataset is converted into a first vector using a dimension transformation array, and the class that is closest to the first vector is selected from multiple vector classes as the belonging class. The vector class is obtained by clustering multiple historical first vectors, and the steel pipes corresponding to multiple vectors in the vector class have similar sawing lengths.

[0010] Substitute the first parameter dataset into the wall thickness prediction model corresponding to the class of belonging to obtain the wall thickness distribution dataset.

[0011] Optimize the steel pipe sawing amount based on the wall thickness distribution dataset.

[0012] In one possible implementation, the plurality of vector classes are obtained by clustering multiple sample parameter datasets, including:

[0013] Obtain a first transformation array and multiple sample parameter datasets, wherein each sample parameter dataset corresponds to a steel pipe sawing length, and the number of data in the first transformation array is the same as the number of data in the sample parameter datasets;

[0014] For each sample parameter dataset, the parameters in the dataset are multiplied with the corresponding data in the first transformation array, and the results are used to construct a process vector;

[0015] Clustering multiple process vectors yields multiple process classes;

[0016] Perform a consistency check on the steel pipe sawing length for each process class and obtain the check results;

[0017] If there is a process class whose inspection results exceed the consistency threshold, then the first transformation array is adjusted according to the inspection results, and the process jumps to the step of multiplying the parameters in the dataset with the corresponding data in the first transformation array for each sample parameter dataset, and constructing the result into a process vector.

[0018] Otherwise, the plurality of process classes are treated as the plurality of vector classes, and the first transformation array is treated as the dimension transformation array.

[0019] In one possible implementation, the step of performing a steel pipe sawing length consistency check on each process class and obtaining the check result includes:

[0020] For each procedure class, perform the following steps:

[0021] The statistical process class corresponds to the distribution range of the steel pipe sawing length;

[0022] Divide the number of process vectors in the process class by the width of the distribution interval, and add the resulting interval density to the density queue.

[0023] If the number of iterations has not been reached, the process vector corresponding to the steel pipe sawing length located at the edge of the interval is removed from the process class, and the process jumps to the step of the distribution interval of the steel pipe sawing length corresponding to the statistical process class.

[0024] Otherwise, a density difference queue is constructed according to the first formula and the density queue, wherein the first formula is:

[0025]

[0026] In the formula, For the density difference queue Each density difference value, For the density queue One density value;

[0027] The maximum value in the density difference queue is used as the length consistency check result.

[0028] In one possible implementation, adjusting the first transformation array based on the inspection result includes:

[0029] Iterate through multiple procedure classes to retrieve a procedure class, and perform the following steps after each retrieval:

[0030] Based on the inspection results, multiple target process vectors are extracted from the process class, wherein the process class after extracting the multiple target process vectors satisfies the length consistency check result.

[0031] Assign target process classes to the plurality of target process vectors, wherein the target process class corresponds to the distribution range of the steel pipe sawing length and is adapted to the plurality of target process vectors;

[0032] The class center of the target process class is used as the target vector;

[0033] A transformation vector is generated based on the second formula, the plurality of target process vectors, and the target vector, wherein the second formula is:

[0034]

[0035] In the formula, For the first The adjustment vector corresponding to the nth target process vector. One element, For the target vector One element, The first of the target process vectors One element, For the first The standardized adjustment vector corresponding to the nth target process vector. One element, This represents the total number of elements in the target process vector. For transformation vectors, This represents the total number of target process vectors.

[0036] The transformation vector is multiplied bitwise by the first transformation array, and the result is used as the adjusted first transformation array.

[0037] In one possible implementation, the process of constructing the wall thickness prediction model includes:

[0038] Multiple sample steel pipes are obtained, wherein each sample steel pipe corresponds to a sawing length, and the sawing lengths of the multiple sample steel pipes are within a predetermined range;

[0039] The multiple sample steel pipes are divided into a first group and a second group;

[0040] A first predetermined number of sample steel pipes are randomly selected from the first group to serve as multiple first sample steel pipes;

[0041] For each of the multiple first sample steel pipes, the first parameter dataset of the steel pipe is input into the initial model, and the output of the model is used as the process output. The output of the model includes multiple wall thickness pairs, each wall thickness pair including the maximum wall thickness and the minimum wall thickness, and each wall thickness pair corresponds to a point of the steel pipe.

[0042] Based on the outputs of multiple processes and the wall thickness distribution data of multiple first sample steel pipes, the process loss is determined;

[0043] The parameters of the initial model are adjusted based on the process loss.

[0044] If the number of iterations has not been reached, proceed to the step of randomly extracting a first predetermined number of sample steel pipes from the first group as multiple first sample steel pipes;

[0045] Otherwise, for each steel pipe in the second group, the first parameter dataset of the steel pipe is input into the initial model, and the output of the model is used as the validation output;

[0046] The verification loss is determined based on multiple verification outputs and the wall thickness distribution data of multiple steel pipes in the second group;

[0047] If the verification loss is greater than the loss threshold, the number of parameters in the initial model is adjusted according to the verification loss, and the process jumps to the step of randomly extracting a first predetermined number of sample steel pipes from the first group as multiple first sample steel pipes.

[0048] Otherwise, the initial model is used as the wall thickness prediction model.

[0049] In one possible implementation, the initial model is:

[0050]

[0051] In the formula, For the first Liede The output of the row element function As the first coefficient, It is a natural constant. For the first Liede The input of row element functions, For the first Liede The row function is used for the first row in the previous column. The input weights of row element functions, For the first The total number of metafunctions in the column. For the first Liede Bias of row element functions.

[0052] In one possible implementation, the initial model includes multiple matrix arrangements of metafunctions, and adjusting the number of parameters in the initial model according to the validation loss includes:

[0053] Calculate the difference between the verification loss and the process loss;

[0054] If the degree of difference is higher than the first threshold, the number of the plurality of metafunctions is reduced proportionally according to the degree of difference, and the initial model is reconstructed based on the adjusted plurality of metafunctions.

[0055] If the degree of difference is lower than the second threshold, the number of the plurality of metafunctions is increased proportionally according to the degree of difference, and the initial model is reconstructed based on the adjusted plurality of metafunctions, wherein the second threshold is lower than the first threshold;

[0056] Otherwise, increase the number of iterations.

[0057] In a second aspect, embodiments of the present invention provide a steel pipe sawing quantity optimization device for implementing the steel pipe sawing quantity optimization method as described in the first aspect or any possible implementation thereof, the steel pipe sawing quantity optimization device comprising:

[0058] The parameter acquisition module is used to acquire a first parameter dataset, wherein the first parameter dataset includes multiple material parameters that affect the wall thickness of the steel pipe and / or multiple processing parameters that affect the wall thickness of the steel pipe;

[0059] The coarse classification module is used to convert the first parameter dataset into a first vector using a dimension transformation array, and select the class that is closest to the first vector from multiple vector classes as the class to which it belongs. The vector class is obtained by clustering multiple historical first vectors, and the steel pipes corresponding to multiple vectors in the vector class have similar sawing lengths.

[0060] The wall thickness distribution prediction module is used to substitute the first parameter dataset into the wall thickness prediction model corresponding to the class to obtain the wall thickness distribution dataset.

[0061] as well as,

[0062] The sawing amount optimization module is used to optimize the steel pipe sawing amount based on the wall thickness distribution dataset.

[0063] Thirdly, embodiments of the present invention provide an electronic device, including a memory and a processor, wherein the memory stores a computer program executable on the processor, and the processor executes the computer program to implement the steps of the method as described in the first aspect or any possible implementation of the first aspect.

[0064] Fourthly, embodiments of the present invention provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the method as described in the first aspect or any possible implementation thereof.

[0065] The beneficial effects of the embodiments of the present invention compared with the prior art are as follows:

[0066] This invention discloses a method for optimizing steel pipe sawing quantity. First, a first parameter dataset is obtained, comprising multiple material parameters affecting steel pipe wall thickness and / or multiple processing parameters affecting steel pipe wall thickness. Then, a dimension transformation array is used to convert the first parameter dataset into a first vector. From multiple vector classes, the class closest to the first vector is selected as the belonging class. The vector class is obtained through clustering multiple historical first vectors, and the steel pipes corresponding to multiple vectors in the vector class have similar sawing lengths. Next, the first parameter dataset is substituted into the wall thickness prediction model corresponding to the belonging class to obtain a wall thickness distribution dataset. Finally, the steel pipe sawing quantity is optimized based on the wall thickness distribution dataset. This invention transforms the data of factors affecting steel pipe wall thickness using a transformation array, classifies the results into vector classes, thus completing a coarse classification of sawing quantity. Then, by inputting the data of factors affecting steel pipe wall thickness into the wall thickness prediction model of the vector class, the wall thickness distribution prediction result is obtained. Because the prediction model is built based on classes, the model is small, requires less sample data, and the model prediction result is easily guaranteed, thereby improving the steel pipe yield. Attached Figure Description

[0067] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0068] Figure 1 This is a flowchart of the steel pipe sawing amount optimization method provided by the embodiments of the present invention;

[0069] Figure 2 This is a schematic diagram of the vector class construction process provided by the embodiments of the present invention;

[0070] Figure 3 This is a functional block diagram of the steel pipe sawing amount optimization device provided in the embodiments of the present invention;

[0071] Figure 4 This is a functional block diagram of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0072] In the following description, specific details such as particular system structures and techniques are set forth for illustrative purposes and not for limitation, so as to provide a thorough understanding of embodiments of the invention. However, those skilled in the art will understand that the invention can be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, and methods are omitted so as not to obscure the description of the invention with unnecessary detail.

[0073] To make the objectives, technical solutions, and advantages of the present invention clearer, specific embodiments will be described below in conjunction with the accompanying drawings.

[0074] The embodiments of the present invention will be described in detail below. This example is implemented based on the technical solution of the present invention, and provides detailed implementation methods and specific operation processes. However, the protection scope of the present invention is not limited to the following embodiments.

[0075] Figure 1 A flowchart of a method for optimizing steel pipe sawing amount provided in an embodiment of the present invention.

[0076] like Figure 1 As shown, a flowchart illustrating the implementation of the steel pipe sawing quantity optimization method provided by the embodiments of the present invention is presented, and is described in detail below:

[0077] In step 101, a first parameter dataset is obtained, wherein the first parameter dataset includes multiple material parameters that affect the wall thickness of the steel pipe and / or multiple processing parameters that affect the wall thickness of the steel pipe.

[0078] In step 102, the first parameter dataset is converted into a first vector using a dimension transformation array, and the class closest to the first vector is selected from multiple vector classes as the belonging class. The vector class is obtained by clustering multiple historical first vectors, and the steel pipes corresponding to multiple vectors in the vector class have similar sawing lengths.

[0079] In some implementations, the plurality of vector classes are obtained by clustering multiple sample parameter datasets, including:

[0080] Obtain a first transformation array and multiple sample parameter datasets, wherein each sample parameter dataset corresponds to a steel pipe sawing length, and the number of data in the first transformation array is the same as the number of data in the sample parameter datasets;

[0081] For each sample parameter dataset, the parameters in the dataset are multiplied with the corresponding data in the first transformation array, and the results are used to construct a process vector;

[0082] Clustering multiple process vectors yields multiple process classes;

[0083] Perform a consistency check on the steel pipe sawing length for each process class and obtain the check results;

[0084] If there is a process class whose inspection results exceed the consistency threshold, then the first transformation array is adjusted according to the inspection results, and the process jumps to the step of multiplying the parameters in the dataset with the corresponding data in the first transformation array for each sample parameter dataset, and constructing the result into a process vector.

[0085] Otherwise, the plurality of process classes are treated as the plurality of vector classes, and the first transformation array is treated as the dimension transformation array.

[0086] In some implementations, the step of performing a steel pipe sawing length consistency check on each process class and obtaining the check result includes:

[0087] For each procedure class, perform the following steps:

[0088] The statistical process class corresponds to the distribution range of the steel pipe sawing length;

[0089] Divide the number of process vectors in the process class by the width of the distribution interval, and add the resulting interval density to the density queue.

[0090] If the number of iterations has not been reached, the process vector corresponding to the steel pipe sawing length located at the edge of the interval is removed from the process class, and the process jumps to the step of the distribution interval of the steel pipe sawing length corresponding to the statistical process class.

[0091] Otherwise, a density difference queue is constructed according to the first formula and the density queue, wherein the first formula is:

[0092]

[0093] In the formula, For the density difference queue Each density difference value, For the density queue One density value;

[0094] The maximum value in the density difference queue is used as the length consistency check result.

[0095] In some implementations, adjusting the first conversion array based on the inspection results includes:

[0096] Iterate through multiple procedure classes to retrieve a procedure class, and perform the following steps after each retrieval:

[0097] Based on the inspection results, multiple target process vectors are extracted from the process class, wherein the process class after extracting the multiple target process vectors satisfies the length consistency check result.

[0098] Assign target process classes to the plurality of target process vectors, wherein the target process class corresponds to the distribution range of the steel pipe sawing length and is adapted to the plurality of target process vectors;

[0099] The class center of the target process class is used as the target vector;

[0100] A transformation vector is generated based on the second formula, the plurality of target process vectors, and the target vector, wherein the second formula is:

[0101]

[0102] In the formula, For the first The adjustment vector corresponding to the nth target process vector. One element, For the target vector One element, The first of the target process vectors One element, For the first The standardized adjustment vector corresponding to the nth target process vector. One element, This represents the total number of elements in the target process vector. For transformation vectors, This represents the total number of target process vectors.

[0103] The transformation vector is multiplied bitwise by the first transformation array, and the result is used as the adjusted first transformation array.

[0104] For example, the present invention aims to predict the wall thickness distribution by using parameters of some processing steps and parameters of the rough tube (the unrolled steel tube). Some technologies attempt to solve the wall thickness prediction problem using artificial intelligence algorithms. However, from a production perspective, due to the large number of steel tube specifications and the high dimensionality of the predicted output distribution data, it is necessary to build a complex model and train it with a massive number of samples. This poses a huge challenge for the collection of preliminary data and the training of the model.

[0105] This invention attempts to predict the wall thickness distribution of steel pipes using a small, highly specific model. For example, the acquired production process parameters (typically including rolling speed, torque, and rolling motor current) and various parameters of the tube (typically tube length, wall thickness distribution, outer diameter, material type, and temperature before entering the mill) are coarsely classified to find a wall thickness prediction model suitable for these parameters. Then, the predicted wall thickness distribution is obtained using the wall thickness prediction model based on the production process parameters and the various parameters of the tube.

[0106] The advantages of this approach are that less resources and manpower are required for collecting preliminary data and building the model. Furthermore, the experience gained in the early stages of model building can be applied to the later stages, greatly reducing the difficulty of using and building the model.

[0107] To achieve the above objectives, in terms of coarse data classification, the present invention collects the obtained data into a first parameter dataset. This parameter dataset is processed by transforming the array into a first vector. The class to which the first vector should be classified is found from multiple vector classes. The wall thickness distribution dataset is obtained by inputting the first parameter dataset into the class prediction model.

[0108] Regarding the construction of vector classes, such as Figure 2As shown, this invention is obtained through clustering multiple sample parameter datasets 201. The sample parameter datasets 201 are datasets with the same parameters as the first parameter dataset, and each sample parameter dataset 201 corresponds to a steel pipe's cutting length 202. These sample parameter datasets 201 are multiplied bitwise using the first transformation array 203 to construct a process vector. These process vectors are then clustered to obtain multiple process classes 204. Each process class 204 then calculates the steel pipe cutting length interval 205 corresponding to the vectors in the class, and performs a consistency check on the steel pipe cutting lengths 202 in the class based on the interval. If all classes pass the consistency check, these process classes 204 are treated as vector classes, and the first transformation array 203 is used as the dimension transformation array. When one class fails the consistency check, the first transformation array 203 needs to be adjusted for these classes that fail the consistency check.

[0109] Regarding consistency checks, this invention employs a method of progressively reducing vectors within a class to obtain a density queue for the class intervals. Specifically, the number of vectors in the process class is divided by the width of the distribution interval, and the result is added to the density queue as the interval density. Then, vectors corresponding to the sawing lengths at both ends of the interval are removed. The steps of calculating the interval density and adding vectors to the density queue are repeated again. After repeating this process a preset number of times, a queue of interval densities corresponding to different numbers of process vectors is formed. Then, the density difference queue is calculated using the first formula:

[0110]

[0111] In the formula, For the density difference queue Each density difference value, For the density queue Density value.

[0112] The maximum value of the density difference queue is used as the result of the consistency check.

[0113] Regarding the adjustment of the first transformation array, the process vectors affecting the consistency check results are extracted and adjusted to the classes in the nearest interval. Generally, during adjustment, the center vector of this nearest class is used as the target vector, and the extracted process vector is used as the target process vector. The second formula, the target vector, and the target process vector are used to generate the transformation vector for adjusting the first transformation array. The second formula is as follows:

[0114]

[0115] In the formula, For the first The adjustment vector corresponding to the nth target process vector. One element, For the target vector One element, The first of the target process vectors One element, For the first The standardized adjustment vector corresponding to the nth target process vector. One element, This represents the total number of elements in the target process vector. For transformation vectors, This represents the total number of target process vectors.

[0116] The adjusted first transformation array is obtained by multiplying the transformation vector bitwise with the first transformation array.

[0117] In step 103, the first parameter dataset is substituted into the wall thickness prediction model corresponding to the class to obtain the wall thickness distribution dataset.

[0118] In some implementations, the process of constructing the wall thickness prediction model includes:

[0119] Multiple sample steel pipes are obtained, wherein each sample steel pipe corresponds to a sawing length, and the sawing lengths of the multiple sample steel pipes are within a predetermined range;

[0120] The multiple sample steel pipes are divided into a first group and a second group;

[0121] A first predetermined number of sample steel pipes are randomly selected from the first group to serve as multiple first sample steel pipes;

[0122] For each of the multiple first sample steel pipes, the first parameter dataset of the steel pipe is input into the initial model, and the output of the model is used as the process output. The output of the model includes multiple wall thickness pairs, each wall thickness pair including the maximum wall thickness and the minimum wall thickness, and each wall thickness pair corresponds to a point of the steel pipe.

[0123] Based on the outputs of multiple processes and the wall thickness distribution data of multiple first sample steel pipes, the process loss is determined;

[0124] The parameters of the initial model are adjusted based on the process loss.

[0125] If the number of iterations has not been reached, proceed to the step of randomly extracting a first predetermined number of sample steel pipes from the first group as multiple first sample steel pipes;

[0126] Otherwise, for each steel pipe in the second group, the first parameter dataset of the steel pipe is input into the initial model, and the output of the model is used as the validation output;

[0127] The verification loss is determined based on multiple verification outputs and the wall thickness distribution data of multiple steel pipes in the second group;

[0128] If the verification loss is greater than the loss threshold, the number of parameters in the initial model is adjusted according to the verification loss, and the process jumps to the step of randomly extracting a first predetermined number of sample steel pipes from the first group as multiple first sample steel pipes.

[0129] Otherwise, the initial model is used as the wall thickness prediction model.

[0130] In some implementations, the initial model is:

[0131]

[0132] In the formula, For the first Liede The output of the row element function As the first coefficient, It is a natural constant. For the first Liede The input of row element functions, For the first Liede The row function is used for the first row in the previous column. The input weights of row element functions, For the first The total number of metafunctions in the column. For the first Liede Bias of row element functions.

[0133] In some implementations, the initial model includes multiple matrix arrangements of metafunctions, and adjusting the number of parameters in the initial model according to the validation loss includes:

[0134] Calculate the difference between the verification loss and the process loss;

[0135] If the degree of difference is higher than the first threshold, the number of the plurality of metafunctions is reduced proportionally according to the degree of difference, and the initial model is reconstructed based on the adjusted plurality of metafunctions.

[0136] If the degree of difference is lower than the second threshold, the number of the plurality of metafunctions is increased proportionally according to the degree of difference, and the initial model is reconstructed based on the adjusted plurality of metafunctions, wherein the second threshold is lower than the first threshold;

[0137] Otherwise, increase the number of iterations.

[0138] In step 104, the steel pipe sawing amount is optimized based on the wall thickness distribution dataset.

[0139] For example, the present invention constructs a wall thickness prediction model using multiple sample steel pipes. For a certain steel pipe model, since it corresponds to a vector class, the sawing lengths of the multiple sample steel pipes used to construct the model also fall within the length range corresponding to that vector class. For a certain sample steel pipe, it corresponds to a first parameter dataset and a wall thickness distribution dataset.

[0140] Multiple sample steel pipes were divided into a first group and a second group. The first group was used to adjust the parameters of the initial model so that the adjusted initial model could fit the relationship between the first parameter dataset and the wall thickness distribution dataset. The second group was used to adjust the structure of the model to avoid overfitting or underfitting. The initial model used in this invention is as follows:

[0141]

[0142] In the formula, For the first Liede The output of the row element function As the first coefficient, It is a natural constant. For the first Liede The input of row element functions, For the first Liede The row function is used for the first row in the previous column. The input weights of row element functions, For the first The total number of metafunctions in the column. For the first Liede Bias of row element functions.

[0143] We can see that the model has multiple metafunctions, and each metafunction has multiple parameters. In terms of parameter adjustment, this invention randomly extracts a first predetermined number of sample steel pipes from the first group. These steel pipes are then fed into the initial model one by one with the first parameter dataset. The model output is used as the process output, and the process loss is calculated using the wall thickness distribution dataset of these steel pipes. The obtained process loss is then used to adjust the model parameters using the gradient descent method. After adjustment, the first predetermined number of sample steel pipes are randomly extracted from the first group again, and the above parameter adjustment steps are repeated. When the number of repetitions reaches a predetermined value, the current parameter adjustment process ends.

[0144] At this point, the first parameter dataset of each steel pipe in the second group is input into the initial model, and the model output is used as the validation output. The validation loss is calculated with the wall thickness distribution dataset of these steel pipes. If the validation loss value deviates significantly from the minimum value of the process loss, or if the validation loss itself has a large value, then it is necessary to adjust the number of parameters in the model according to the validation loss. One application scenario is to adjust the number of metafunctions in the initial model. After the adjustment is completed, the above steps of taking sample steel pipes from the first group and adjusting the model parameters are repeated until the validation loss meets the pre-set loss threshold.

[0145] Regarding adjusting the number of model parameters based on the validation loss, this invention first calculates the difference between the validation loss and the process loss, which is generally expressed as a percentage, i.e., the percentage value obtained by dividing the difference between the validation loss and the process loss by the process loss.

[0146] If the difference exceeds a certain first threshold, it means that the initial model is too complex relative to the sample steel pipe. The number of metafunctions in the initial model needs to be reduced proportionally according to the difference, and then the initial model needs to be reconstructed.

[0147] If the difference is lower than a certain set second threshold (the second threshold is less than the first threshold), it means that the complexity of the initial model is insufficient. The number of metafunctions in the initial model needs to be increased proportionally according to the difference, and then the initial model needs to be reconstructed.

[0148] If the difference is between two thresholds, it means that the initial model complexity meets the requirements of the sample steel pipe, and the number of iterations needs to be increased.

[0149] Substituting the first parameter dataset obtained in the aforementioned steps into the wall thickness prediction model corresponding to the class, a wall thickness distribution dataset can be obtained. Based on this predicted wall thickness distribution dataset, the amount of steel pipe to be cut can be determined.

[0150] The present invention provides an implementation method for optimizing steel pipe sawing quantity. First, a first parameter dataset is obtained, comprising multiple material parameters affecting steel pipe wall thickness and / or multiple processing parameters affecting steel pipe wall thickness. Then, a dimension transformation array is used to convert the first parameter dataset into a first vector. From multiple vector classes, the class closest to the first vector is selected as the belonging class. The vector class is obtained through clustering multiple historical first vectors, and the steel pipes corresponding to multiple vectors in the vector class have similar sawing lengths. Next, the first parameter dataset is substituted into the wall thickness prediction model corresponding to the belonging class to obtain a wall thickness distribution dataset. Finally, the steel pipe sawing quantity is optimized based on the wall thickness distribution dataset. This invention transforms the data of factors affecting steel pipe wall thickness using a transformation array, classifies the results into vector classes, thus completing a coarse classification of sawing quantity. Then, by inputting the data of factors affecting steel pipe wall thickness into the wall thickness prediction model of the vector class, the wall thickness distribution prediction result is obtained. Because the prediction model is built based on classes, the model is small, requires less sample data, and the model prediction result is easily guaranteed, thereby improving the steel pipe yield.

[0151] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0152] The following are embodiments of the apparatus of the present invention. For details not described in detail, please refer to the corresponding method embodiments described above.

[0153] Figure 3 This is a functional block diagram of the steel pipe sawing amount optimization device provided in the embodiments of the present invention, with reference to... Figure 3 The steel pipe sawing quantity optimization device includes: a parameter acquisition module 301, a coarse classification module 302, a wall thickness distribution prediction module 303, and a sawing quantity optimization module 304, wherein:

[0154] The parameter acquisition module 301 is used to acquire a first parameter dataset, wherein the first parameter dataset includes multiple material parameters that affect the wall thickness of the steel pipe and / or multiple processing parameters that affect the wall thickness of the steel pipe;

[0155] The coarse classification module 302 is used to convert the first parameter dataset into a first vector using a dimension transformation array, and select the class that is closest to the first vector from multiple vector classes as the class to which it belongs. The vector class is obtained by clustering multiple historical first vectors, and the steel pipes corresponding to multiple vectors in the vector class have similar sawing lengths.

[0156] The wall thickness distribution prediction module 303 is used to substitute the first parameter dataset into the wall thickness prediction model corresponding to the class to obtain the wall thickness distribution dataset.

[0157] The sawing amount optimization module 304 is used to optimize the steel pipe sawing amount based on the wall thickness distribution dataset.

[0158] Figure 4 This is a functional block diagram of the electronic device provided in an embodiment of the present invention. For example... Figure 4 As shown, the electronic device 4 of this embodiment includes a processor 400 and a memory 401, wherein the memory 401 stores a computer program 402 that can run on the processor 400. When the processor 400 executes the computer program 402, it implements the steps of the various steel pipe sawing quantity optimization methods and embodiments described above, for example... Figure 1 Steps 101 to 104 are shown.

[0159] For example, the computer program 402 may be divided into one or more modules / units, which are stored in the memory 401 and executed by the processor 400 to complete the present invention.

[0160] The electronic device 4 can be a desktop computer, laptop, handheld computer, cloud server, or other computing device. The electronic device 4 may include, but is not limited to, a processor 400 and a memory 401. Those skilled in the art will understand that... Figure 4 This is merely an example of electronic device 4 and does not constitute a limitation on electronic device 4. It may include more or fewer components than shown, or combine certain components, or different components. For example, electronic device 4 may also include input / output devices, network access devices, buses, etc.

[0161] The processor 400 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0162] The memory 401 can be an internal storage unit of the electronic device 4, such as a hard disk or memory. The memory 401 can also be an external storage device of the electronic device 4, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card. Furthermore, the memory 401 can include both internal and external storage units of the electronic device 4. The memory 401 is used to store the computer program 402 and other programs and data required by the electronic device 4. The memory 401 can also be used to temporarily store data that has been output or will be output.

[0163] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the aforementioned method embodiments, and will not be repeated here.

[0164] In the above embodiments, the descriptions of each embodiment have their own emphasis. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0165] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0166] In the embodiments provided by this invention, it should be understood that the disclosed devices / electronic devices and methods can be implemented in other ways. For example, the device / electronic device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0167] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0168] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0169] If the integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above-described embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various methods and apparatus embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0170] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for optimizing the amount of steel pipe cutting, characterized in that, include: Obtain a first parameter dataset, wherein the first parameter dataset includes multiple material parameters that affect the wall thickness of the steel pipe and / or multiple processing parameters that affect the wall thickness of the steel pipe; The first parameter dataset is converted into a first vector using a dimension transformation array, and the class that is closest to the first vector is selected from multiple vector classes as the belonging class. The vector class is obtained by clustering multiple historical first vectors, and the steel pipes corresponding to multiple vectors in the vector class have similar sawing lengths. Substitute the first parameter dataset into the wall thickness prediction model corresponding to the class of belonging to obtain the wall thickness distribution dataset. Optimize the steel pipe sawing amount based on the wall thickness distribution dataset; The plurality of vector classes are obtained by clustering multiple sample parameter datasets, including: Obtain a first transformation array and multiple sample parameter datasets, wherein each sample parameter dataset corresponds to a steel pipe sawing length, and the number of data in the first transformation array is the same as the number of data in the sample parameter datasets; For each sample parameter dataset, the parameters in the dataset are multiplied with the corresponding data in the first transformation array, and the results are used to construct a process vector; Clustering multiple process vectors yields multiple process classes; Perform a consistency check on the steel pipe sawing length for each process class and obtain the check results; If there is a process class whose inspection results exceed the consistency threshold, then the first transformation array is adjusted according to the inspection results, and the process jumps to the step of multiplying the parameters in the dataset with the corresponding data in the first transformation array for each sample parameter dataset, and constructing the result into a process vector. Otherwise, the plurality of process classes are treated as the plurality of vector classes, and the first transformation array is treated as the dimension transformation array.

2. The method for optimizing steel pipe sawing quantity according to claim 1, characterized in that, The process of performing a steel pipe sawing length consistency check on each process class and obtaining the check results includes: For each procedure class, perform the following steps: The statistical process class corresponds to the distribution range of the steel pipe sawing length; Divide the number of process vectors in the process class by the width of the distribution interval, and add the resulting interval density to the density queue. If the number of iterations has not been reached, the process vector corresponding to the steel pipe sawing length located at the edge of the interval is removed from the process class, and the process jumps to the step of the distribution interval of the steel pipe sawing length corresponding to the statistical process class. Otherwise, a density difference queue is constructed according to the first formula and the density queue, wherein the first formula is: In the formula, For the density difference queue Each density difference value, For the density queue One density value; The maximum value in the density difference queue is used as the length consistency check result.

3. The method for optimizing steel pipe sawing quantity according to claim 1, characterized in that, The step of adjusting the first conversion array based on the inspection results includes: Iterate through multiple procedure classes to retrieve a procedure class, and perform the following steps after each retrieval: Based on the inspection results, multiple target process vectors are extracted from the process class, wherein the process class after extracting the multiple target process vectors satisfies the length consistency check result. Assign target process classes to the plurality of target process vectors, wherein the target process class corresponds to the distribution range of the steel pipe sawing length and is adapted to the plurality of target process vectors; The class center of the target process class is used as the target vector; A transformation vector is generated based on the second formula, the plurality of target process vectors, and the target vector, wherein the second formula is: In the formula, For the first The adjustment vector corresponding to the nth target process vector. One element, For the target vector One element, The first of the target process vectors One element, For the first The standardized adjustment vector corresponding to the nth target process vector. One element, This represents the total number of elements in the target process vector. For transformation vectors, This represents the total number of target process vectors. The transformation vector is multiplied bitwise by the first transformation array, and the result is used as the adjusted first transformation array.

4. The method for optimizing steel pipe sawing quantity according to any one of claims 1-3, characterized in that, The construction process of the wall thickness prediction model includes: Multiple sample steel pipes are obtained, wherein each sample steel pipe corresponds to a sawing length, and the sawing lengths of the multiple sample steel pipes are within a predetermined range; The multiple sample steel pipes are divided into a first group and a second group; A first predetermined number of sample steel pipes are randomly selected from the first group to serve as multiple first sample steel pipes; For each of the multiple first sample steel pipes, the first parameter dataset of the steel pipe is input into the initial model, and the output of the model is used as the process output. The output of the model includes multiple wall thickness pairs, each wall thickness pair including the maximum wall thickness and the minimum wall thickness, and each wall thickness pair corresponds to a point of the steel pipe. Based on the outputs of multiple processes and the wall thickness distribution data of multiple first sample steel pipes, the process loss is determined; The parameters of the initial model are adjusted based on the process loss. If the number of iterations has not been reached, proceed to the step of randomly extracting a first predetermined number of sample steel pipes from the first group as multiple first sample steel pipes; Otherwise, for each steel pipe in the second group, the first parameter dataset of the steel pipe is input into the initial model, and the output of the model is used as the validation output; The verification loss is determined based on multiple verification outputs and the wall thickness distribution data of multiple steel pipes in the second group; If the verification loss is greater than the loss threshold, the number of parameters in the initial model is adjusted according to the verification loss, and the process jumps to the step of randomly extracting a first predetermined number of sample steel pipes from the first group as multiple first sample steel pipes. Otherwise, the initial model is used as the wall thickness prediction model.

5. The method for optimizing steel pipe sawing quantity according to claim 4, characterized in that, The initial model is: In the formula, For the first Liede The output of the row function, As the first coefficient, It is a natural constant. For the first Liede The input of row element functions, For the first Liede The row function is used for the first row in the previous column. The input weights of row element functions, For the first The total number of metafunctions in the column. For the first Liede Bias of row element functions.

6. The method for optimizing steel pipe sawing quantity according to claim 4, characterized in that, The initial model includes multiple matrix-arranged metafunctions, and adjusting the number of parameters in the initial model according to the validation loss includes: Calculate the difference between the verification loss and the process loss; If the degree of difference is higher than the first threshold, the number of the plurality of metafunctions is reduced proportionally according to the degree of difference, and the initial model is reconstructed based on the adjusted plurality of metafunctions. If the degree of difference is lower than the second threshold, the number of the plurality of metafunctions is increased proportionally according to the degree of difference, and the initial model is reconstructed based on the adjusted plurality of metafunctions, wherein the second threshold is lower than the first threshold; Otherwise, increase the number of iterations.

7. A device for optimizing steel pipe sawing volume, characterized in that, For implementing the steel pipe sawing quantity optimization method as described in any one of claims 1-6, the steel pipe sawing quantity optimization device comprises: The parameter acquisition module is used to acquire a first parameter dataset, wherein the first parameter dataset includes multiple material parameters that affect the wall thickness of the steel pipe and / or multiple processing parameters that affect the wall thickness of the steel pipe; The coarse classification module is used to convert the first parameter dataset into a first vector using a dimension transformation array, and select the class that is closest to the first vector from multiple vector classes as the class to which it belongs. The vector class is obtained by clustering multiple historical first vectors, and the steel pipes corresponding to multiple vectors in the vector class have similar sawing lengths. The wall thickness distribution prediction module is used to substitute the first parameter dataset into the wall thickness prediction model corresponding to the class to obtain the wall thickness distribution dataset. as well as, The sawing amount optimization module is used to optimize the steel pipe sawing amount based on the wall thickness distribution dataset.

8. An electronic device comprising a memory and a processor, wherein the memory stores a computer program executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 6 above.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 6 above.

Images