A cable production intelligent management and control method and system based on feature data analysis

Abstract

The application discloses a kind of cable production intelligent management and control method and system based on feature data analysis, it is related to cable production management and control technical field, comprising: first, obtain cable structure information to determine type, synchronous acquisition corresponding type's historical production whole-process data, including equipment operation, process parameters and multiple data such as;Then, extract key features from historical data, construct cable production quality index prediction model and production equipment energy consumption prediction model;Again, obtain real-time production data and extract key features, input model to obtain quality and energy consumption index;Finally, based on two kinds of index dynamic adjustment production plan, cover set threshold, multi-scenario early warning and plan maintenance or adjustment strategy.The application has the advantages that, through data-driven, realize production whole-process intelligent management and control, can improve quality prediction accuracy, optimize energy consumption management, enhance production plan flexibility, provide technical support for efficient and accurate management of cable production.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of cable production management and control, in particular to a cable production intelligent management and control method and system based on feature data analysis. BACKGROUND

[0002] In the field of cable production, traditional management and control methods face problems such as low data utilization rate, difficulty in optimizing quality and energy consumption, etc. With the development of industrial intelligence, the characteristics hidden in massive production data need to be mined.

[0003] The existing cable production intelligent management and control method and system based on feature data analysis lack systematic analysis of cable production full-process data, making it difficult to accurately extract key features and build prediction models, resulting in lagging production plan adjustment and inefficient resource allocation. Therefore, a cable production intelligent management and control method and system based on feature data analysis are needed to solve the above problems. SUMMARY

[0004] To solve the above technical problems, a cable production intelligent management and control method and system based on feature data analysis are provided, which solves the problem of the existing cable production intelligent management and control method and system based on feature data analysis lacking systematic analysis of cable production full-process data, making it difficult to accurately extract key features and build prediction models, resulting in lagging production plan adjustment and inefficient resource allocation.

[0005] To achieve the above purposes, the technical scheme adopted by the present application is as follows:

[0006] A cable production intelligent management and control method based on feature data analysis, comprising:

[0007] Obtain cable structure information, determine cable type information, and synchronously collect historical cable production full-process data related to the cable type, including historical equipment operation data, historical process parameter data, historical quality detection data, historical material data, and historical environmental data;

[0008] Based on the historical cable production full-process data, extract historical key feature data, and build cable production quality index prediction model and production equipment energy consumption prediction model;

[0009] Obtain real-time equipment operation data, real-time process parameter data, real-time quality detection data, real-time material data, and real-time environmental data, then extract real-time key feature data and input it into the cable production quality index prediction model and production equipment energy consumption prediction model to obtain cable production quality index and production equipment energy consumption index;

[0010] Based on the cable production quality index and the production equipment energy consumption index, the cable production plan is dynamically adjusted.

[0011] In an optional embodiment, the cable structure information is acquired, and the cable type information is determined, specifically including:

[0012] The cable structure information is acquired through the way of drawing analysis;

[0013] The cable structure information is decomposed into quantified feature dimensions to form a feature vector set F = {f1, f2, …, fn}, each feature fn in the feature vector set corresponding to the physical properties and set parameters of the cable;

[0014] A cable type classification rule library is constructed, and the feature vector set is matched with the classification rule set in the cable type classification rule library one by one to determine the cable type information;

[0015] The construction step of the cable type classification rule library is:

[0016] Cable production information is acquired to construct a mapping relationship from structure features to cable types;

[0017] Based on the mapping relationship from structure features to cable types, a classification rule set R = {R1, R2, …, Rk} is defined, wherein each rule Rk corresponds to a set of feature constraint conditions;

[0018] The classification rule set is integrated to obtain the cable type classification rule library.

[0019] In an optional embodiment, the historical key feature data is extracted based on the historical cable production full-process data, specifically including:

[0020] Based on the historical cable production full-process data, time span information and collection source information corresponding to the historical cable production full-process data are acquired;

[0021] The historical cable production full-process data is divided according to production factors to establish a data dictionary;

[0022] The data in the data dictionary is data cleaned and preprocessed to obtain an initial feature matrix;

[0023] Based on the normal distribution assumption, the 3σ principle is used to identify outliers in the initial feature matrix, if |X i，j -μ j |>3σ j , it is determined that the feature vector X i，j of the feature j of the initial sample i is an outlier, and the outlier is removed;

[0024] The features in the initial feature matrix after removing the abnormal values are standardized by using the Z-score standardization method, a feature matrix is reconstructed, and the reconstructed feature matrix is taken as a basic feature matrix;

[0025] Based on the basic feature matrix, a historical equipment operation stability index and a historical process efficiency index are determined;

[0026] A feature vector X in the basic feature matrix is obtained i，j A Pearson correlation coefficient FR of the historical equipment operation stability index j A feature vector X in the basic feature matrix is obtained i，j A Pearson correlation coefficient TR of the historical process efficiency index j ;

[0027] A threshold τ is set, and then the feature vectors with FR j > τ or TR j > τ in the basic feature matrix are retained, and the remaining feature vectors are removed, to obtain a historical key feature reference matrix;

[0028] The retained feature vectors in the historical key feature reference matrix are extracted, and the feature values contained in the retained feature vectors in the historical key feature reference matrix are sequentially obtained as historical key feature values;

[0029] Based on a principal component analysis method, a cumulative variance contribution rate CumVar(m) corresponding to the historical key feature values is obtained, the first m principal components with CumVar(m)≥θ are retained, and a historical key feature standard vector is obtained;

[0030] The historical key feature standard vector is composed into a new feature matrix as historical key feature data.

[0031] In an optional embodiment, the step of constructing the cable production quality index prediction model specifically includes:

[0032] The data in the historical key feature data is filtered by feature importance ranking of a random forest, to obtain an optimal feature subset;

[0033] A collection time span T1 corresponding to the optimal feature subset is obtained, and the feature values in the optimal feature subset are sorted according to the timestamp information in the collection time span T1, to obtain X s1,t1 , t1∈T1, wherein X s1,t1 is the s1th feature value in the optimal feature subset at the t1th moment;

[0034] The mean, variance and extreme value of the s1th feature value within the collection time span T1 are obtained, or the frequency domain features of the s1th feature value within the collection time span T1 are extracted by Fourier transform.

[0035] The mean, variance and extreme value of the s1th characteristic value within the collection time span T1, and the frequency domain feature of the s1th characteristic value within the collection time span T1 are taken as model input parameters;

[0036] Based on the neural network, the historical key feature data is divided into a training set and a test set in a ratio of 7 to 3, the neural network model is trained synchronously, and after cross-validation and hyperparameter optimization, a cable production quality index prediction model is obtained.

[0037] In an optional embodiment, the step of constructing the production equipment energy consumption prediction model specifically comprises:

[0038] New features related to energy consumption are extracted from the historical key feature data, and the production equipment numbers corresponding to the new features are synchronously obtained;

[0039] Based on the production equipment numbers, the characteristic values in the historical key feature data corresponding to each production equipment number are determined, and the characteristic values in the historical key feature data obtained by the same production equipment are stored in the same data set, thereby obtaining a production equipment energy consumption prediction subset;

[0040] The collection time span T2 corresponding to the production equipment energy consumption prediction subset is obtained, and the characteristic values in the production equipment energy consumption prediction subset are sorted according to the timestamp information in the collection time span T2, thereby obtaining X s2,t2 , t2∈T2, wherein X s2,t2 is the s2th characteristic value in the production equipment energy consumption prediction subset at the t2th moment;

[0041] The mean, variance and extreme value of the s2th characteristic value within the collection time span T2, or the frequency domain feature of the s2th characteristic value within the collection time span T2 is extracted by Fourier transform;

[0042] The mean, variance and extreme value of the s2th characteristic value within the collection time span T2, and the frequency domain feature of the s2th characteristic value within the collection time span T2 are taken as model input parameters;

[0043] The historical key feature data of the same production equipment number is divided into a data set, thereby obtaining a production equipment energy consumption prediction first data set;

[0044] Based on linear regression, each production equipment energy consumption prediction first data set corresponding to each production equipment number is divided into a training set and a test set in a ratio of 8 to 2, a linear regression network model is trained synchronously, and after cross-validation and hyperparameter optimization, a production equipment energy consumption prediction model is obtained.

[0045] In an optional embodiment, the real-time key feature data is extracted and input into the cable production quality index prediction model and the production equipment energy consumption prediction model to obtain the cable production quality index and the production equipment energy consumption index, specifically comprising:

[0046] The production equipment number corresponding to the real-time key feature data is obtained, and then the real-time key feature data is decomposed into feature, and the data required by the cable production quality index prediction model is decomposed into first real-time key feature data, and the data required by the production equipment energy consumption prediction model is decomposed into second real-time key feature data;

[0047] The first real-time key feature data and the second real-time key feature data are input into the cable production quality index prediction model and the production equipment energy consumption prediction model respectively, and the cable production quality index and the production equipment energy consumption index output by the cable production quality index prediction model and the production equipment energy consumption prediction model are obtained.

[0048] In an optional embodiment, the cable production plan is dynamically adjusted based on the cable production quality index and the production equipment energy consumption index, specifically comprising:

[0049] The cable production quality index threshold and the production equipment energy consumption index threshold are set, and the prediction time period corresponding to the cable production quality index and the production equipment energy consumption index is obtained;

[0050] If the cable production quality index is greater than the cable production quality index threshold and the production equipment energy consumption index is greater than the production equipment energy consumption index threshold in the prediction time period, an emergency early warning mechanism is triggered immediately, and an alarm information is sent to the production management personnel;

[0051] If the cable production quality index is greater than the cable production quality index threshold and the production equipment energy consumption index is less than or equal to the production equipment energy consumption index threshold in the prediction time period, a quality early warning program is started, and a quality abnormality prompt is sent to the quality control department and the related production personnel;

[0052] If the cable production quality index is less than or equal to the cable production quality index threshold and the production equipment energy consumption index is greater than the production equipment energy consumption index threshold in the prediction time period, an energy consumption early warning notice is issued;

[0053] If the cable production quality index is less than or equal to the cable production quality index threshold and the production equipment energy consumption index is less than or equal to the production equipment energy consumption index threshold in the prediction time period, the current production plan and production process are maintained unchanged, and the production is continued according to the established production arrangement.

[0054] Further, an intelligent cable production management and control system based on feature data analysis is proposed, which is used to realize the management and control method of any one of the above, comprising:

[0055] The acquisition module is used for acquiring cable structure information, used for acquiring historical cable production whole-process data related to the cable type, used for acquiring real-time equipment operation data, real-time process parameter data, real-time quality detection data, real-time material data and real-time environmental data;

[0056] The data processing module is used for determining cable type information, used for extracting historical key feature data based on the historical cable production whole-process data, constructing a cable production quality index prediction model and a production equipment energy consumption prediction model, used for extracting real-time key feature data and inputting the real-time key feature data into the cable production quality index prediction model and the production equipment energy consumption prediction model to acquire cable production quality indexes and production equipment energy consumption indexes;

[0057] The management and control module is used for dynamically adjusting the cable production plan based on the cable production quality indexes and the production equipment energy consumption indexes.

[0058] The display module is used for presenting the process and result of the cable production intelligent management and control to a user.

[0059] In an optional embodiment, the acquisition module comprises:

[0060] The first acquisition unit is used for acquiring cable structure information;

[0061] The second acquisition unit is used for acquiring historical cable production whole-process data related to the cable type;

[0062] The third acquisition unit is used for acquiring real-time equipment operation data, real-time process parameter data, real-time quality detection data, real-time material data and real-time environmental data.

[0063] In an optional embodiment, the data processing module comprises:

[0064] The first processing unit is used for determining cable type information;

[0065] The second processing unit is used for extracting historical key feature data based on the historical cable production whole-process data, and constructing a cable production quality index prediction model and a production equipment energy consumption prediction model;

[0066] The third processing unit is used for extracting real-time key feature data and inputting the real-time key feature data into the cable production quality index prediction model and the production equipment energy consumption prediction model to acquire cable production quality indexes and production equipment energy consumption indexes.

[0067] Compared with the prior art, the present application has the following beneficial effects:

[0068] The cable production intelligent management and control method based on feature data analysis provided by the scheme realizes accurate identification of the cable type by obtaining cable structure information through drawing analysis and quantifying it into a feature vector set, and matching the cable type classification rule library to determine the cable type, lays a foundation for subsequent data acquisition and model construction; through time span and acquisition source analysis, data cleaning pretreatment, outlier elimination, data standardization and other operations on historical cable production whole-process data, key feature data in history is extracted, accurate screening of key features from massive data is realized, and the data quality and effectiveness of model input are improved; the random forest is used to filter the optimal feature subset, and a cable production quality index prediction model is constructed by combining a neural network, the training set and the test set are divided according to the 7:3 ratio, and the model is optimized, accurate prediction of the cable production quality index is realized, and the accuracy of quality prediction is improved; new features related to energy consumption are extracted from the historical key feature data, the data set is divided according to the production equipment number, and a production equipment energy consumption prediction model is constructed by using linear regression, effective prediction of the production equipment energy consumption is realized, and a scientific basis is provided for energy consumption management; real-time key feature data is obtained and decomposed to input the corresponding model to obtain quality and energy consumption indexes, real-time monitoring of the production process is realized, and the production state is ensured to be grasped in time; the quality and energy consumption index thresholds are set, the production plan is dynamically adjusted according to different index combinations, and the corresponding early warning mechanism is triggered, dynamic optimization of the production plan and multi-scenario early warning are realized, and the flexibility of the production plan and the stability of the production process are enhanced. BRIEF DESCRIPTION OF DRAWINGS

[0069] Figure 1 The flow chart of the cable production intelligent management and control method based on feature data analysis provided by the present application is shown in the figure.

[0070] Figure 2 The flow chart of the acquisition of the historical key feature data in the present application is shown in the figure.

[0071] Figure 3 The flow chart of the construction of the cable production quality index prediction model in the present application is shown in the figure.

[0072] Figure 4 The system framework diagram of the cable production intelligent management and control system based on feature data analysis provided by the present application is shown in the figure. DETAILED DESCRIPTION

[0073] The following description is used to disclose the present application so that those skilled in the art can implement the present application. The preferred embodiments in the following description are only as examples, and other obvious modifications can be thought of by those skilled in the art.

[0074] REFERENCE Figure 1 - Figure 4 As shown in the figure, the cable production intelligent management and control method based on feature data analysis comprises:

[0075] obtain cable structure information, determine cable type information, synchronously collect historical cable production whole-process data related to the cable type, the historical cable production whole-process data including historical equipment operation data, historical process parameter data, historical quality detection data, historical material data and historical environment data;

[0076] based on the historical cable production whole-process data, extract historical key feature data, construct a cable production quality index prediction model and a production equipment energy consumption prediction model;

[0077] obtain real-time equipment operation data, real-time process parameter data, real-time quality detection data, real-time material data and real-time environment data, then extract real-time key feature data and input the real-time key feature data into the cable production quality index prediction model and the production equipment energy consumption prediction model to obtain cable production quality indexes and production equipment energy consumption indexes;

[0078] based on the cable production quality indexes and the production equipment energy consumption indexes, dynamically adjust the cable production plan.

[0079] Further, obtain cable structure information, determine cable type information, specifically including:

[0080] obtain the cable structure information through drawing analysis;

[0081] decompose the cable structure information into quantified feature dimensions to form a feature vector set F = {f1, f2, …, fn}, each feature fn in the feature vector set corresponding to a physical property and a set parameter of the cable;

[0082] construct a cable type classification rule library, and match the feature vector set with a classification rule set in the cable type classification rule library one by one to determine the cable type information;

[0083] The construction steps of the cable type classification rule library are:

[0084] obtain cable production information, and construct a mapping relationship from structure features to cable types;

[0085] based on the mapping relationship from structure features to cable types, define a classification rule set R = {R1, R2, …, Rk}, wherein each rule Rk corresponds to a set of feature constraint conditions;

[0086] integrate the classification rule set to obtain the cable type classification rule library.

[0087] Specifically, let the cable structure feature set be F = {f1, f2, …, fn}, where each feature fn corresponds to a certain physical property or geometric parameter of the cable (such as conductor material, insulation layer thickness, number of shielding layers, sheath material, etc.). The feature vector can be represented as: x = [x1, x2, …, xn], where x n ] T , x n ∈ V n , where V n is the value space of feature fn (such as a discrete value set or a continuous interval). For discrete features (such as cable type enumeration values): let C = {c1, c2, …, cm} be a pre-defined set of cable types, and the type is determined by the mapping function φ: x → C. For continuous features (such as conductor diameter, insulation thickness): the numerical value xn ∈ R is obtained by the detection equipment, which needs to satisfy the process standard constraint xn ∈ [Ln, Un], where Ln and Un are the upper and lower limits of feature fn. Define the classification rule set R = {R1, R2, …, Rk}, where each rule Rk corresponds to a set of feature constraint conditions: → cable type is c k , where A k,i is the allowed value range of feature fn in rule Rk (discrete value set or continuous interval). For a given x, there is a unique j ∈ {1, 2, …, k} such that: x | = Rk (i.e. all conditions of rule Rk are satisfied). If the matching is successful, the cable type y = cj; if there is multiple matching or no matching, the manual verification mechanism needs to be triggered. The determined cable type y is used as the key identifier, which is associated to the subsequent links of the production process (such as process parameter configuration, equipment scheduling, etc.). Define the output function Type(x) = y, where y ∈ C, and x | = R k → y = c j .

[0088] Further, based on the historical cable production full-process data, the historical key feature data is extracted, specifically including:

[0089] Based on the historical cable production full-process data, the time span information and collection source information corresponding to the historical cable production full-process data are obtained;

[0090] The historical cable production full-process data is divided according to production factors, and a data dictionary is established;

[0091] The data in the data dictionary is cleaned and preprocessed to obtain an initial feature matrix;

[0092] Based on the normal distribution assumption, the 3σ principle is used to identify outliers in the initial feature matrix, and if |X i，j -μ j | > 3σ j, it is determined that the feature vector X corresponding to the feature j of the initial sample i is an abnormal value i，j is an abnormal value, and the abnormal value is removed;

[0093] The features in the initial feature matrix after removing the abnormal values are standardized by using the Z-score standardization method, the feature matrix is reconstructed, and the reconstructed feature matrix is taken as a basic feature matrix;

[0094] Based on the basic feature matrix, a historical equipment operation stability index and a historical process efficiency index are determined;

[0095] The feature vector X in the basic feature matrix is obtained i，j The Pearson correlation coefficient FR of the feature vector X in the basic feature matrix and the historical equipment operation stability index j The feature vector X in the basic feature matrix is obtained i，j The Pearson correlation coefficient TR of the feature vector X in the basic feature matrix and the historical process efficiency index j ;

[0096] A threshold τ (0.3, determined according to the 3σ principle) is set, and then the feature vectors with FR j >τ or TR j >τ in the basic feature matrix are retained, and the remaining feature vectors are removed, to obtain a historical key feature reference matrix;

[0097] The retained feature vectors in the historical key feature reference matrix are extracted, and the feature values contained in the retained feature vectors in the historical key feature reference matrix are sequentially obtained as historical key feature values;

[0098] Based on a principal component analysis method, a cumulative variance contribution rate CumVar(m) corresponding to the historical key feature values is obtained, the first m principal components with CumVar(m)≥θ are retained, to obtain a historical key feature standard vector, wherein θ can be 80%, and the final value is determined according to experiments from 80% to 99% in sequence;

[0099] The historical key feature standard vector is composed into a new feature matrix as historical key feature data.

[0100] Specifically, the data dictionary includes D={D 设备 , D 工艺 , D 质量 , D 物料 , D 环境 , …}, wherein D 设备 ={v i |v i ∈{speed, temperature, pressure}}, D 工艺 ={p j |p j{extrusion speed, vulcanization temperature}, and the rest by analogy. The historical equipment operation stability index Stab i = Var(X i，j , temperature), i.e., the variance of temperature. The historical process efficiency index Eff i = output length i / material consumption i.

[0101] Specific embodiments, for example, a cable production enterprise, collect historical production data for nearly 3 years, and clarify that the data comes from various sensors, quality detection instruments and material management systems on the production line. According to the production factors, the data types are divided to form a complete data dictionary. The collected feature matrix data is processed, and for the samples with missing values, the corresponding filling method is used according to the data type; the 3σ principle is used to detect and process outliers; Z-score standardization method is used to standardize all features, and a unified scale feature matrix is obtained. From the original features, device operation stability, process efficiency and other composite features are derived, and time domain and frequency domain features are calculated. By calculating the Pearson correlation coefficient and performing variance analysis, features with high and significant correlation with quality and energy consumption indicators are selected. Use principal component analysis for feature dimensionality reduction, and retain principal components with cumulative variance contribution rate of more than 80%. The selected key features form a new feature matrix, which is matched with the corresponding quality and energy consumption target variables. According to the production batch, the data is grouped to form a historical key feature data set. The data set is divided into training set, validation set and test set by stratified sampling. An index containing batch number and time stamp is established, and the data is stored in the enterprise's data warehouse for subsequent model training and production process analysis.

[0102] Further, the steps of constructing the cable production quality index prediction model specifically include:

[0103] The data in the historical key feature data is filtered by the feature importance ranking of the random forest to obtain an optimal feature subset;

[0104] The acquisition time span T1 corresponding to the optimal feature subset is obtained, and the feature values in the optimal feature subset are sorted according to the time stamp information in the acquisition time span T1 to obtain X s1,t1 , t1 ∈ T1, wherein X s1,t1 is the s1th feature value in the optimal feature subset at the t1th moment;

[0105] The mean, variance and extreme value of the s1th feature value within the acquisition time span T1 are obtained, or the frequency domain features of the s1th feature value within the acquisition time span T1 are extracted by Fourier transform;

[0106] the mean value, the variance and the extreme value of the s1th characteristic value in the collection time span T1, and the frequency domain features of the s1th characteristic value in the collection time span T1 are taken as model input parameters;

[0107] Based on the neural network, the historical key feature data is divided into a training set and a test set in a ratio of 7 to 3, the neural network model is trained synchronously, and after cross-validation and hyperparameter optimization, a cable production quality index prediction model is obtained.

[0108] Specifically, the expression of the model input parameters corresponding to the production equipment energy consumption prediction model is:

[0109]

[0110] In the formula, f1 is a model input parameter matrix corresponding to the production equipment energy consumption prediction model, σ1 and x max1 are the mean value, the variance and the extreme value of the s1th characteristic value corresponding to the production equipment energy consumption prediction model in the collection time span T1, and |X1(f1)| and |X1(f2)| are frequency domain characteristic values of the s1th characteristic value corresponding to the production equipment energy consumption prediction model in the collection time span T1.

[0111] Further, the construction steps of the production equipment energy consumption prediction model specifically include:

[0112] New features related to energy consumption are extracted from the historical key feature data, and the production equipment numbers corresponding to the new features are synchronously obtained;

[0113] Based on the production equipment numbers, the characteristic values in the historical key feature data corresponding to each production equipment number are determined, and the characteristic values in the historical key feature data obtained from the same production equipment are stored in the same data set, to obtain a production equipment energy consumption prediction subset to be selected;

[0114] The collection time span T2 corresponding to the production equipment energy consumption prediction subset to be selected is obtained, and the characteristic values in the production equipment energy consumption prediction subset to be selected are sorted according to the timestamp information in the collection time span T2, to obtain X s2,t2 , t2∈T2, wherein X s2,t2 is the s2th characteristic value in the production equipment energy consumption prediction subset to be selected at the t2th moment;

[0115] The mean value, the variance and the extreme value of the s2th characteristic value in the collection time span T2 are obtained, or the frequency domain features of the s2th characteristic value in the collection time span T2 are extracted through Fourier transform;

[0116] a mean value, a variance and an extreme value of the s2th characteristic value within the collection time span T2, and a frequency domain feature of the s2th characteristic value within the collection time span T2 are taken as model input parameters;

[0117] The historical key feature data of the same production equipment number is divided into a data set, and a production equipment energy consumption prediction first data set is obtained;

[0118] Based on linear regression, the production equipment energy consumption prediction first data set corresponding to each production equipment number is divided into a training set and a test set according to an 8:2 ratio, and a linear regression network model is trained synchronously. After cross-validation and hyperparameter optimization, a production equipment energy consumption prediction model is obtained.

[0119] Specifically, the expression of the model input parameters corresponding to the production equipment energy consumption prediction model is:

[0120]

[0121] In the formula, f2 is a model input parameter matrix corresponding to the production equipment energy consumption prediction model, σ2 and x max2 are a mean value, a variance and an extreme value of the s2th characteristic value within the collection time span T2, and |X2(f1)| and |X2(f2)| are frequency domain feature values of the s2th characteristic value within the collection time span T2.

[0122] Further, real-time key feature data is extracted and input into the cable production quality index prediction model and the production equipment energy consumption prediction model to obtain cable production quality indexes and production equipment energy consumption indexes. Specifically, the method comprises:

[0123] The production equipment number corresponding to the real-time key feature data is obtained, and then the real-time key feature data is decomposed into first real-time key feature data required by the cable production quality index prediction model and second real-time key feature data required by the production equipment energy consumption prediction model;

[0124] The first real-time key feature data and the second real-time key feature data are input into the cable production quality index prediction model and the production equipment energy consumption prediction model respectively, and the cable production quality indexes and the production equipment energy consumption indexes output by the cable production quality index prediction model and the production equipment energy consumption prediction model are obtained.

[0125] Specifically, the feature decomposition of real-time key feature data is to divide the real-time feature data into two categories according to the pre-defined feature-model mapping relationship, for example, process quality related features-extrusion temperature, pulling speed, insulation material melt index, etc.-quality index prediction model, equipment energy consumption related features-equipment current, voltage, running time, idle rate, etc.-energy consumption index prediction model.

[0126] Further, based on the cable production quality index and the production equipment energy consumption index, the cable production plan is dynamically adjusted, specifically including:

[0127] The cable production quality index threshold and the production equipment energy consumption index threshold are set, the prediction time period corresponding to the cable production quality index and the production equipment energy consumption index is obtained, and based on the historical production data of the same type of cable, for the set cable production quality index threshold, the mean (μ) and the standard deviation (σ) of the quality index are calculated, and the dynamic threshold is set in combination with the process stability. For example: the mean ± 2σ of the qualified products is used as the internal control warning line, and the mean ± 3σ is used as the forced adjustment threshold. For the production equipment energy consumption index threshold, the historical energy consumption data of the same type of equipment is analyzed, a "production scale-energy consumption" regression model is established, and the upper limit of energy consumption under different production capacity is set. For example: when the production capacity is Q, the energy consumption threshold E thres =aQ+b (a, b are fitting coefficients). The acquisition of the prediction time period is based on the fixed time period of the production cycle. For single batch production scene: according to the cable production process time (such as extrusion, vulcanization, cooling, etc.), the prediction time period is determined as the remaining production time of the current batch. For example, the total production cycle of a certain cable batch is 8 hours, and 5 hours have been produced, and the prediction time period is set to the future 3 hours. For continuous production scene: divide the prediction time period according to the fixed time window (such as 1 hour, 4 hours, 8 hours), and adapt to the periodic scheduling needs of the production line.

[0128] If the cable production quality index is greater than the cable production quality index threshold and the production equipment energy consumption index is greater than the production equipment energy consumption index threshold within the prediction time period, an emergency warning mechanism is triggered immediately, and an alarm information is sent to the production management personnel;

[0129] If the cable production quality index is greater than the cable production quality index threshold and the production equipment energy consumption index is less than or equal to the production equipment energy consumption index threshold within the prediction time period, a quality warning program is started, and a quality abnormality prompt is issued to the quality control department and the relevant production personnel;

[0130] If the cable production quality index is less than or equal to the cable production quality index threshold and the production equipment energy consumption index is greater than the production equipment energy consumption index threshold within the prediction time period, an energy consumption warning notice is issued;

[0131] If the cable production quality index is less than or equal to the cable production quality index threshold and the production equipment energy consumption index is less than or equal to the production equipment energy consumption index threshold within the prediction time period, the current production plan and production process are maintained unchanged, and production continues according to the established production arrangement.

[0132] Specifically, if the cable production quality index is greater than the cable production quality index threshold and the production equipment energy consumption index is greater than the production equipment energy consumption index threshold within the prediction time period:

[0133] An emergency warning mechanism is immediately triggered, and an alarm message is sent to the production manager, detailing the quality and energy consumption exceeding the standard and potential risks.

[0134] The current production task is suspended, and professionals from process, equipment, and quality departments form a special team to conduct a comprehensive investigation of the production process. Using root cause analysis tools, combined with historical production data and real-time process parameters, the root causes of the simultaneous quality and energy consumption exceeding the standard are analyzed in multiple dimensions such as raw material quality, equipment operating status, process parameter settings, and operator skills.

[0135] According to the analysis results, a comprehensive rectification plan is developed. If there is a problem with the raw materials, replace them with qualified batches immediately and isolate and retest the products produced using the batch of raw materials; if equipment failure or performance decline causes the problem, arrange professional maintenance personnel to repair and debug the equipment, and replace key components if necessary; if the process parameters are unreasonable, optimize the process parameter settings and conduct small-batch trial production to verify the adjustment effect.

[0136] Subsequent production plans are significantly adjusted, and the production cycle is extended to ensure sufficient time for rectification and quality verification; the order delivery time is re-evaluated, and if it cannot be delivered on time, communicate with the customer in a timely manner, explain the situation and negotiate a solution; at the same time, consider increasing the frequency and items of quality detection to strengthen the monitoring of the production process and ensure that the product quality meets the standards.

[0137] If the cable production quality index is greater than the cable production quality index threshold and the production equipment energy consumption index is less than or equal to the production equipment energy consumption index threshold within the prediction time period:

[0138] Start the quality warning program and send a quality abnormality prompt to the quality control department and relevant production personnel, clearly indicating the specific content of the quality index exceeding the standard and the potential impact.

[0139] The current production equipment operating status is temporarily maintained to avoid additional impacts on energy consumption due to equipment adjustments. At the same time, arrange quality detection personnel to conduct sampling re-inspection of the produced products to further confirm the scope and severity of the quality problem.

[0140] The organization process and quality department personnel conduct in-depth analysis of the production process, compare historical production data and standard process parameters, and find the process link that causes quality problems. For example, check whether the extrusion temperature and pressure are stable, and whether the strand pitch meets the requirements. According to the analysis results, fine-tune the process parameters, such as appropriately reducing the extrusion speed, adjusting the mold temperature, etc., and after adjustment, carry out small batch trial production to verify the quality improvement effect.

[0141] If the quality problem has a greater impact on product performance, the produced unqualified products need to be isolated, reworked or scrapped; if the impact is small, quality monitoring can be strengthened in subsequent production to ensure that similar problems do not occur again. At the same time, the production plan is partially adjusted, and the quality detection link and frequency are increased to ensure that the subsequent product quality meets the standards, but the production rhythm and energy consumption level are kept stable as much as possible.

[0142] If the cable production quality index is less than or equal to the cable production quality index threshold value within the predicted time period, and the production equipment energy consumption index is greater than the production equipment energy consumption index threshold value:

[0143] An energy consumption warning notice is issued to remind the device management department and production personnel to pay attention to the abnormal situation of device energy consumption and analyze the reasons for the excessive energy consumption.

[0144] The current production task is maintained normally, but the change of production quality index is closely monitored to prevent the product quality from being affected by adjusting energy consumption measures. The device maintenance personnel are arranged to conduct a comprehensive check on the high-energy-consumption equipment, including the running parameters, lubrication condition, electrical system, etc. of the equipment, to check whether there are equipment failures, low running efficiency or energy waste, etc.

[0145] According to the inspection results, targeted energy-saving measures are taken. If the equipment has a fault that causes increased energy consumption, it is repaired in time; if the equipment running parameters are unreasonable, the speed, power and other parameters of the equipment are adjusted to optimize the equipment running efficiency; if there are loopholes in energy management, unnecessary energy-consuming equipment is turned off or the idle running time is reduced. In the process of taking energy-saving measures, the changes of device energy consumption and production quality index are monitored in real time to ensure that the quality is not affected.

[0146] The energy distribution in the production plan is re-optimized, and the running time and load distribution of each device are reasonably adjusted to reduce the overall energy consumption level while ensuring product quality. At the same time, the implementation effect of the energy-saving measures is recorded to provide experience reference for energy consumption management in the subsequent production process.

[0147] If the cable production quality index is less than or equal to the cable production quality index threshold value within the predicted time period, and the production equipment energy consumption index is less than or equal to the production equipment energy consumption index threshold value:

[0148] Maintain the current production plan and production process unchanged, continue to produce according to the established production arrangement.

[0149] Strengthen the daily monitoring of the production process, continuously collect quality indicators and energy consumption indicators data, observe the trend of change, and ensure the stability and continuity of the production process. At the same time, regularly analyze the production data, summarize the production experience, and find the space for further optimization of production process and improvement of production efficiency.

[0150] Carry out preventive maintenance work, regularly maintain and check the production equipment, and find and handle potential equipment fault hidden dangers in advance to ensure that the equipment is always in good working condition to maintain stable quality and energy consumption level.

[0151] According to the production situation and market demand, timely fine-tune the production plan, such as reasonably arrange equipment maintenance time, optimize order production sequence, etc., on the basis of ensuring quality and energy consumption standard, improve production efficiency and enterprise economic benefit.

[0152] Further, an intelligent cable production management and control system based on feature data analysis is proposed, which is used to realize the management and control method of any one of the above, comprising:

[0153] The acquisition module is used to acquire cable structure information, used to collect historical cable production whole-process data related to cable type, used to acquire real-time equipment operation data, real-time process parameter data, real-time quality detection data, real-time material data and real-time environment data;

[0154] The data processing module is used to determine the cable type information, based on the historical cable production whole-process data, extract the historical key feature data, construct the cable production quality index prediction model and the production equipment energy consumption prediction model, extract the real-time key feature data, and input it into the cable production quality index prediction model and the production equipment energy consumption prediction model, and acquire the cable production quality index and the production equipment energy consumption index;

[0155] The management and control module is used to dynamically adjust the cable production plan based on the cable production quality index and the production equipment energy consumption index;

[0156] The display module is used to present the process and results of the intelligent cable production management and control to the user.

[0157] Further, the acquisition module comprises:

[0158] The first acquisition unit is used to acquire cable structure information;

[0159] The second acquisition unit is used to collect historical cable production whole-process data related to cable type;

[0160] The third acquisition unit is configured to acquire real-time equipment operation data, real-time process parameter data, real-time quality detection data, real-time material data and real-time environment data.

[0161] Further, the data processing module comprises:

[0162] The first processing unit is configured to determine the cable type information.

[0163] The second processing unit is configured to extract historical key feature data based on historical cable production full-process data, and construct a cable production quality index prediction model and a production equipment energy consumption prediction model.

[0164] The third processing unit is configured to extract real-time key feature data, and input the real-time key feature data into the cable production quality index prediction model and the production equipment energy consumption prediction model to obtain cable production quality indexes and production equipment energy consumption indexes.

[0165] The basic principles, main features and advantages of the present application are shown and described above. It should be understood by those skilled in the art that the present application is not limited by the above examples, and the above examples and descriptions in the specification are only the principles of the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the claimed present application. The scope of protection claimed by the present application is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent control of cable production based on feature data analysis, characterized in that, include: Obtain cable structure information, determine cable type information, and simultaneously collect historical cable production process data related to the cable type. The historical cable production process data includes historical equipment operation data, historical process parameter data, historical quality inspection data, historical material data, and historical environmental data. Based on historical data of the entire cable production process, key historical feature data are extracted to construct a cable production quality index prediction model and a production equipment energy consumption prediction model. The system acquires real-time equipment operation data, real-time process parameter data, real-time quality inspection data, real-time material data, and real-time environmental data. Then, it extracts real-time key feature data and inputs it into the cable production quality index prediction model and the production equipment energy consumption prediction model to obtain cable production quality index and production equipment energy consumption index. Based on cable production quality indicators and production equipment energy consumption indicators, the cable production plan is dynamically adjusted. The acquisition of cable structure information and determination of cable type information specifically includes: Cable structure information is obtained through drawing analysis; The cable structure information is decomposed into quantified feature dimensions to form a feature vector set F = {f1, f2, ..., fn}, where each feature fn in the feature vector set corresponds to the physical properties and set parameters of the cable. A cable type classification rule base is constructed, and the feature vector set is matched one by one with the classification rule set in the cable type classification rule base to determine the cable type information; The steps for constructing the cable type classification rule base are as follows: Obtain cable production information and construct a mapping relationship from structural features to cable types; Based on the mapping relationship from structural features to cable type, a set of classification rules R = {R1, R2, ..., Rk} is defined, where each rule Rk corresponds to a set of feature constraints. The set of classification rules is integrated to obtain a cable type classification rule library; The construction steps of the cable production quality index prediction model specifically include: The optimal feature subset is obtained by filtering the data from historical key feature data through feature importance ranking using random forest; Obtain the collection time span T1 corresponding to the optimal feature subset, and sort the feature values ​​in the optimal feature subset according to the timestamp information in the collection time span T1 to obtain X. s1,t1 t1∈T1, where X s1,t1 This represents the s1st eigenvalue in the optimal feature subset at time t1. Obtain the mean, variance, and extreme values ​​of the s1st eigenvalue within the acquisition time span T1, or extract the frequency domain features of the s1st eigenvalue within the acquisition time span T1 through Fourier transform; The mean, variance, and extreme values ​​of the s1st eigenvalue within the acquisition time span T1, as well as the frequency domain characteristics of the s1st eigenvalue within the acquisition time span T1, are used as model input parameters. Based on neural networks, historical key feature data are divided into training and testing sets in a 7:3 ratio. The neural network model is trained simultaneously. After cross-validation and hyperparameter optimization, a cable production quality index prediction model is obtained. The steps for constructing the energy consumption prediction model for the production equipment specifically include: Extract new features related to energy consumption from historical key feature data, and simultaneously obtain the production equipment number corresponding to the new features; Based on the production equipment number, the feature values ​​in the historical key feature data corresponding to each production equipment number are determined, and the feature values ​​in the historical key feature data obtained from the same production equipment are stored in the same dataset to obtain the energy consumption prediction subset of the candidate production equipment. Obtain the data collection time span T2 corresponding to the energy consumption prediction subset of the candidate production equipment, and sort the feature values ​​in the energy consumption prediction subset of the candidate production equipment according to the timestamp information in the data collection time span T2 to obtain X. s2,t2 t2∈T2, where X s2,t2 The s2nd characteristic value in the subset of energy consumption predictions for candidate production equipment at time t2; Obtain the mean, variance, and extreme values ​​of the s2nd eigenvalue within the acquisition time span T2, or extract the frequency domain features of the s2nd eigenvalue within the acquisition time span T2 through Fourier transform; The mean, variance, and extreme values ​​of the s2nd eigenvalue within the acquisition time span T2, as well as the frequency domain characteristics of the s2nd eigenvalue within the acquisition time span T2, are used as model input parameters. Historical key feature data with the same production equipment number are grouped into one dataset to obtain the first dataset for production equipment energy consumption prediction. Based on linear regression, the first dataset for predicting the energy consumption of each production equipment number is divided into a training set and a test set in an 8:2 ratio. The linear regression network model is trained simultaneously, and after cross-validation and hyperparameter optimization, the production equipment energy consumption prediction model is obtained.

2. The intelligent control method for cable production based on feature data analysis according to claim 1, characterized in that, The extraction of key historical feature data based on historical cable production process data specifically includes: Based on historical cable production process data, obtain the time span information and data collection source information corresponding to the historical cable production process data; The historical cable production process data was divided according to production factors to establish a data dictionary; Data cleaning and preprocessing are performed on the data in the data dictionary to obtain the initial feature matrix; Based on the assumption of normal distribution, the 3σ principle is used to identify outliers in the initial feature matrix. If |X i，j -μ j |>3σ j Then determine the feature vector X corresponding to feature j of the initial sample i. i，j If an outlier is identified, it will be removed. The Z-score standardization method is used to standardize the features in the initial feature matrix after outlier removal, reconstruct the feature matrix, and use the reconstructed feature matrix as the base feature matrix. Based on the basic feature matrix, historical equipment operation stability indicators and historical process efficiency indicators are determined. Obtain the eigenvector X from the basic feature matrix i，j Pearson correlation coefficient FR with historical equipment operational stability indicators j Obtain the eigenvector X from the basic feature matrix. i，j Pearson correlation coefficient TR with historical process efficiency indicators j ; Set a threshold τ, then extract FR from the basic feature matrix. j >τ or TR j The eigenvectors of >τ are retained, and the remaining eigenvectors are removed to obtain the historical key feature reference matrix; Extract the feature vectors retained in the historical key feature reference matrix, and then obtain the feature values ​​contained in the feature vectors retained in the historical key feature reference matrix as historical key feature values. Based on the principal component analysis method, the cumulative variance contribution rate CumVar(m) corresponding to the historical key feature values ​​is obtained. The first m principal components with CumVar(m)≥θ are retained to obtain the standard vector of historical key features. The standard vectors of historical key features are used to form a new feature matrix, which is then used as historical key feature data.

3. The intelligent control method for cable production based on feature data analysis according to claim 1, characterized in that, The process of extracting real-time key feature data and inputting it into the cable production quality index prediction model and the production equipment energy consumption prediction model to obtain cable production quality indicators and production equipment energy consumption indicators specifically includes: Obtain the production equipment number corresponding to the real-time key feature data, and then decompose the real-time key feature data. Decompose the data required by the cable production quality index prediction model into the first real-time key feature data, and decompose the data required by the production equipment energy consumption prediction model into the second real-time key feature data. The first real-time key feature data and the second real-time key feature data are respectively input into the cable production quality index prediction model and the production equipment energy consumption prediction model to obtain the cable production quality index and the production equipment energy consumption index output by the cable production quality index prediction model and the production equipment energy consumption prediction model.

4. The intelligent control method for cable production based on feature data analysis according to claim 1, characterized in that, The dynamic adjustment of the cable production plan based on cable production quality indicators and production equipment energy consumption indicators specifically includes: Set thresholds for cable production quality indicators and energy consumption indicators for production equipment, and obtain the prediction time periods corresponding to the cable production quality indicators and energy consumption indicators for production equipment. If the cable production quality index exceeds the cable production quality index threshold and the production equipment energy consumption index exceeds the production equipment energy consumption index threshold within the predicted time period, the emergency early warning mechanism will be triggered immediately to send an alarm message to the production management personnel. If the cable production quality index is greater than the cable production quality index threshold during the predicted time period, and the production equipment energy consumption index is less than or equal to the production equipment energy consumption index threshold, then the quality early warning procedure will be activated to issue a quality abnormality warning to the quality control department and relevant production personnel. If the cable production quality index is less than or equal to the cable production quality index threshold during the predicted time period, and the production equipment energy consumption index is greater than the production equipment energy consumption index threshold, an energy consumption early warning notice will be issued. If the cable production quality index is less than or equal to the cable production quality index threshold and the production equipment energy consumption index is less than or equal to the production equipment energy consumption index threshold during the predicted time period, then the current production plan and production process will remain unchanged, and production will continue according to the established production schedule.

5. A cable production intelligent control system based on feature data analysis, used to implement the control method as described in any one of claims 1-4, characterized in that, include: The data acquisition module is used to acquire cable structure information, collect historical cable production process data related to cable type, and acquire real-time equipment operation data, real-time process parameter data, real-time quality inspection data, real-time material data, and real-time environmental data. The data processing module is used to determine cable type information, extract historical key feature data based on historical cable production process data, construct cable production quality index prediction model and production equipment energy consumption prediction model, extract real-time key feature data and input it into the cable production quality index prediction model and production equipment energy consumption prediction model to obtain cable production quality index and production equipment energy consumption index. The control module is used to dynamically adjust the cable production plan based on cable production quality indicators and production equipment energy consumption indicators. The display module is used to present the process and results of intelligent control of cable production to the user.

6. The intelligent control system for cable production based on feature data analysis according to claim 5, characterized in that, The acquisition module includes: The first acquisition unit is used to acquire cable structure information; The second acquisition unit is used to collect historical data on the entire cable production process related to the cable type. The third acquisition unit is used to acquire real-time equipment operation data, real-time process parameter data, real-time quality inspection data, real-time material data, and real-time environmental data.

7. The intelligent control system for cable production based on feature data analysis according to claim 5, characterized in that, The data processing module includes: A first processing unit is used to determine cable type information; The second processing unit is used to extract historical key feature data based on historical cable production process data, and to construct a cable production quality index prediction model and a production equipment energy consumption prediction model. The third processing unit is used to extract real-time key feature data and input it into the cable production quality index prediction model and the production equipment energy consumption prediction model to obtain the cable production quality index and the production equipment energy consumption index.

Images