Fiber selection and distribution optimization method for guiding optical cable production by industrial big data

Abstract

The invention discloses a fiber selection and distribution optimization method for guiding optical cable production by industrial big data, and relates to the field of optical cable manufacturing, and the method comprises the steps: collecting optical fiber performance parameters, production equipment parameters, environment parameters and dynamic resource data; denoising, classifying and normalizing the acquired optical fiber performance parameters, production equipment parameters, environmental parameters and dynamic resource data; constructing an optical cable fiber selection and distribution optimization model through statistical analysis and machine learning, and predicting the influence of different fiber selection and distribution schemes on the optical cable performance; and a multi-target optimization algorithm is adopted to calculate an optimal scheme of a multi-target optical fiber arrangement sequence and an assembly mode. According to the invention, through two innovation modules of multi-source data fusion acquisition and machine learning optimization, a complete intelligent fiber selection and distribution system is constructed, and in a data acquisition stage, optical fiber performance parameters, equipment operation parameters, environment states and supply chain data are synchronously acquired.

Description

Technical Field

[0001] The present invention relates to the field of optical cable manufacturing, and in particular to a method for optimizing fiber selection in optical cable production guided by industrial big data. Background Art

[0002] In the field of optical cable manufacturing, traditional fiber selection and matching technology mainly relies on manual experience and a fixed rule base. By establishing a fiber performance parameter database and combining it with preset matching rules to generate a fiber matching plan, this technology uses a decision-making system based on a rule engine to compare key parameters such as fiber attenuation and dispersion with predefined process standards, and output a fiber matching combination that meets basic technical indicators. Some advanced systems have introduced simple linear regression models to predict the impact of specific arrangements on transmission performance. Existing technologies can achieve basic fiber matching functions and maintain a relatively stable yield rate in a conventional production environment.

[0003] Although existing technologies have achieved basic automation, their adaptability in dynamic production environments is significantly limited. Specifically, the ability to coordinate and optimize fiber distribution plans and real-time process parameters is insufficient: when production equipment status (such as fluctuations in coating machine speed), environmental conditions (temperature and humidity changes), or supply chain resources (differences in fiber batches) change dynamically, it is difficult for a fixed rule base to quickly generate an optimal solution. In the event of sudden temperature and humidity fluctuations, traditional methods cannot dynamically adjust the fiber assembly sequence to compensate for the additional attenuation caused by environmental factors. This static decision-making mode causes material utilization to fluctuate by 15%-20%, and it is difficult to meet the stringent performance consistency requirements of military-grade products. Existing technologies have not yet effectively solved the problem of real-time coordination between multi-dimensional production data and fiber distribution decisions. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a fiber selection optimization method for optical cable production guided by industrial big data to solve the problem of insufficient adaptability to dynamic production environments caused by static rule bases in the prior art.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a method for optimizing fiber selection in optical cable production guided by industrial big data, which includes collecting optical fiber performance parameters, production equipment parameters, environmental parameters, and dynamic resource data;

[0008] De-noising, classification and normalization of collected optical fiber performance parameters, production equipment parameters, environmental parameters and dynamic resource data;

[0009] Build an optical fiber selection optimization model through statistical analysis and advanced machine learning to predict the impact of different fiber selection schemes on optical cable performance;

[0010] Use multi-objective optimization algorithm to calculate the optimal solution for multi-objective optical fiber arrangement sequence and combination mode;

[0011] Apply the optimal solution of multi-objective optical fiber arrangement sequence and combination method to the production line to obtain production data, and use quality inspection equipment to evaluate the fiber matching effect to obtain optimization results;

[0012] Enter production data and optimization results into the knowledge base to form a dynamic optimization plan.

[0013] As a preferred solution of the method for optimizing the selection of optical fibers for guiding optical cable production using industrial big data of the present invention, the method includes collecting optical fiber performance parameters, production equipment parameters, environmental parameters and dynamic resource data, including the following steps:

[0014] Collect fiber attenuation values ​​and output structured data packets, which are measured using a dispersion analyzer to generate parameter groups;

[0015] Based on the parameter group, the mode field diameter is calculated by the far-field scanning method to obtain the fiber performance parameters;

[0016] Collect coating speed output time series data stream, use Kalman filter to process tension sensor raw data, and output purified tension sequence;

[0017] Record the temperature of the three zones of the sheath extruder based on the purification tension sequence and generate production equipment parameters;

[0018] Use SHT35 sensor to record environmental RH / T data, output environmental vector, bind vibration spectrum, and obtain environmental parameters;

[0019] Read the fiber reel number and length, output the inventory list, call the supplier API to obtain OTD data based on the inventory type code, and generate dynamic resource data.

[0020] As a preferred solution of the method for optimizing the selection of optical fibers for guiding optical cable production using industrial big data of the present invention, the collected optical fiber performance parameters, production equipment parameters, environmental parameters and dynamic resource data are subjected to denoising, classification and normalization, including the following steps:

[0021] Eliminate outliers, noise data, and complete data on collected optical fiber performance parameters, production equipment parameters, environmental parameters, and dynamic resource data;

[0022] Interpolate or predict missing data, standardize data of different dimensions, classify and summarize data, and organize and classify them according to fiber type, production batch, etc.

[0023] As a preferred solution of the method for optimizing fiber selection in optical cable production guided by industrial big data of the present invention, an optical cable fiber selection optimization model is constructed by statistical analysis and advanced machine learning to predict the impact of different fiber selection schemes on optical cable performance, including the following steps:

[0024] Convert the fiber performance parameters, production equipment parameters, environmental parameters and dynamic resource data after denoising, classification and normalization into feature vectors;

[0025] According to the tolerance level of technical indicators, each production batch is labeled with a quality grade label and a label data set is output;

[0026] Form a Pearson correlation coefficient matrix based on the label data set to obtain the feature subset;

[0027] Based on the fiber performance parameters in the feature subset, the K-means algorithm is used to divide the fibers into k groups and output the clustering results;

[0028] Based on the parameters of the Yuejin algorithm, a support vector machine is used to build an optical cable fiber selection optimization model, and a comprehensive objective function is used to train the optical cable fiber selection optimization model, and the trained optical cable fiber selection optimization model is output;

[0029] The feature vector is input into the trained optical cable fiber selection optimization model to predict the impact of different fiber selection schemes on optical cable performance and obtain the predicted value.

[0030] As a preferred solution of the fiber selection optimization method for guiding optical cable production with industrial big data of the present invention, a multi-objective optimization algorithm is used to calculate the optimal solution for the multi-objective optical fiber arrangement sequence and combination mode, including the following steps:

[0031] Based on the comprehensive objective function, the feature vector is converted into a four-dimensional target vector and the optimization target set is output;

[0032] Based on the fiber waste length records, the waste matrix is ​​obtained according to the factory tolerance and tolerance multiplication factor. According to the fiber substitutability rule, the substitution vector is generated by counting the supplier material substitution records.

[0033] Based on the waste matrix and substitution vector, constraints and the inter-substitutability vector of the same type of optical fiber are set. Based on the eigenvector, multiple sets of performance parameters corresponding to the same optical fiber ID are integrated into row vectors to obtain the optical fiber performance matrix.

[0034] Using 12-bit binary coding rules, the fiber performance matrix is ​​encoded into the initial population;

[0035] Calculate the fitness of each individual in the initial population, output the scored population, use tournament selection and single-point crossover to process the scored population, and output the offspring population;

[0036] Perform bit flipping on the offspring population using the mutation rate and output a new generation of population;

[0037] The fitness of the new generation population is sorted by genetic algorithm to obtain the optimal solution for the multi-objective optical fiber arrangement sequence and combination method.

[0038] As a preferred solution of the method for optimizing the selection of optical fibers for guiding optical cable production using industrial big data of the present invention, the optimal solution of the multi-objective optical fiber arrangement sequence and combination mode is applied to the production line to obtain production data, including the following steps:

[0039] The optimal solution for the arrangement sequence and combination of multiple target optical fibers is converted into equipment control parameters according to the process mapping table and output as an instruction set;

[0040] A structured production work order is obtained based on the instruction set, and the speed parameters in the structured production work order are sent to the coating machine through the industrial bus to adjust the speed to the target value in real time;

[0041] According to the tension parameters in the structured production work order, the tension of the stranding equipment is controlled to the target value through servo control;

[0042] Adjust the sheath extruder temperature through PID algorithm and output the temperature adjustment value;

[0043] Based on environmental sensor data, the workshop humidity is dynamically adjusted, the optical cable attenuation value is scanned every 100 meters, and the deviation of the predicted value is used to obtain production data.

[0044] As a preferred solution of the method for optimizing fiber selection for optical cable production guided by industrial big data of the present invention, wherein: evaluating the fiber matching effect by quality inspection equipment to obtain the optimization result includes the following steps:

[0045] Use an optical time domain reflectometer to scan the optical cable at 100-meter intervals to calculate the measured attenuation value;

[0046] Acquire fiber alignment images using an optical microscope and output an offset dataset;

[0047] Compare the measured attenuation value with the predicted value to form a standardized deviation and output a deviation report;

[0048] Calculate the standard deviation based on the data set and output a stability score;

[0049] Through the compensation algorithm, the tension adjustment instruction is obtained;

[0050] The optical cable selection fiber optimization model weights are updated based on the deviation report and stability score to obtain the optimization results.

[0051] As a preferred solution of the fiber selection optimization method for guiding optical cable production with industrial big data described in the present invention, the production data and optimization results are entered into a knowledge base to form a dynamic optimization plan, including the following steps: the production data and optimization results of each production process are entered into the production knowledge base to form a reusable fiber selection strategy, supporting the production needs of different personalities and types of optical cables, and forming a dynamic optimization plan.

[0052] In a second aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the method for optimizing fiber selection for optical cable production guided by industrial big data as described in the first aspect of the present invention is implemented.

[0053] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the method for optimizing fiber selection in optical cable production guided by industrial big data as described in the first aspect of the present invention.

[0054] The present invention has the following beneficial effects: Through two innovative modules, multi-source data fusion and acquisition, and Yujin machine learning optimization, a complete intelligent fiber selection system is constructed. During the data acquisition phase, fiber performance parameters, equipment operating parameters, environmental status, and supply chain data are simultaneously acquired, and Kalman filtering and far-field scanning methods are used to ensure data accuracy. During the optimization decision-making phase, a hybrid model is constructed based on the Yujin algorithm. By defining a three-dimensional demand vector and a comprehensive objective function, an improved genetic algorithm is combined to achieve a multi-objective Pareto optimal solution search. Ultimately, dynamic optimization capabilities are achieved through closed-loop feedback from the knowledge base. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0056] Figure 1 Flowchart of the fiber selection optimization method for guiding optical cable production based on industrial big data.

[0057] Figure 2 Schematic diagram of the digital process for optical cable production.

[0058] Figure 3 This is the optical cable production process and data collection diagram.

[0059] Figure 4 This is the architecture diagram of the fiber data acquisition and analysis system.

[0060] Figure 5 This is a schematic diagram of the fiber selection knowledge base construction process and optimization model structure. DETAILED DESCRIPTION

[0061] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0062] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0063] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0064] Reference Figures 1 to 5 , is an embodiment of the present invention, which provides a method for optimizing fiber selection in optical cable production guided by industrial big data, comprising the following steps:

[0065] S1. Collect optical fiber performance parameters, production equipment parameters, environmental parameters and dynamic resource data.

[0066] S1.1. Collect the optical fiber attenuation value and output a structured data packet. Use a dispersion analyzer to measure and generate a parameter group.

[0067] Furthermore, an optical time domain reflectometer is used to scan the optical fiber at a wavelength of 1550nm, and a dispersion analyzer is used to measure the dispersion coefficients D1310 and D1550 at wavelengths of 1310nm and 1550nm. The measurement results are associated with the fiber ID in the structured data packet to expand and generate a parameter group in the format of {fiber ID:α,D1310,D1550}.

[0068] S1.2. Based on the parameter group, the mode field diameter is calculated by the far-field scanning method to obtain the optical fiber performance parameters.

[0069] Furthermore, a parameter group including the fiber ID, attenuation value α, and dispersion coefficients D1310 and D1550 is used as input, and a far-field scanning device is used to perform a two-dimensional scan of the laser spot emitted from the fiber end face. The light intensity distribution at different radial positions is recorded, and the measured light intensity distribution data is Gaussian fitted to solve the fiber mode field diameter MFD. The mode field diameter MFD is merged with the original data in the parameter group to generate a fiber performance parameter set containing complete fiber performance parameters.

[0070] S1.3. Collect the coating speed output time series data stream, use Kalman filtering to process the tension sensor raw data, and output the purified tension sequence.

[0071] Furthermore, a laser velocimeter is installed at the outlet of the coating machine to record the coating speed v in real time at a fixed sampling frequency, and generate a coating speed time series data stream {v(t)} corresponding to the timestamp and speed value; the timestamp t of the coating speed time series data stream is used as a reference, and the tension sensor raw data T_raw is synchronously collected; the Kalman filter algorithm is applied to process T_raw, and the filtering formula is T_filtered = K×(zH×T_prev), where K is the gain coefficient, z is the observation value, H is the state transfer matrix, and T_prev is the estimated value at the previous moment; the filtered tension data is associated with the timestamp to generate a purified tension sequence.

[0072] S1.4. Record the temperatures of the three zones of the sheath extruder based on the purification tension sequence and generate production equipment parameters.

[0073] Furthermore, taking the timestamp t of the purification tension sequence {T_filtered(t)} as the synchronization benchmark, thermocouple temperature sensors are installed in the feed zone, melting zone and die zone of the sheath extruder respectively, and the temperatures of the three zones Temp1, Temp2 and Temp3 are collected in real time; the collected temperature data is aligned with the timestamp of the purification tension sequence to generate a production equipment parameter set containing the complete equipment operation status, and its data structure is {timestamp t, purification tension T_filtered(t), feed zone temperature Temp1(t), melting zone temperature Temp2(t), die zone temperature Temp3(t)}.

[0074] S1.5. Use the SHT35 sensor to record the environmental RH / T data, output the environmental vector, bind the vibration spectrum, and obtain the environmental parameters.

[0075] Furthermore, SHT35 temperature and humidity sensors are installed at key workstations of the optical cable production line to synchronously measure the ambient relative humidity RH and ambient temperature T at a fixed sampling interval to generate an environmental vector {RH(t), T(t)}. A three-axis acceleration sensor is installed at the same location to collect vibration spectrum data Vrms(f) in the 0-1kHz frequency band. The environmental vector and the vibration spectrum data are precisely aligned at timestamp t to form a complete set of environmental parameters, whose data structure is {timestamp t: ambient relative humidity RH(t), ambient temperature T(t), vibration spectrum Vrms(f)}.

[0076] S1.6. Read the fiber reel number and length, output the inventory list, call the supplier API based on the inventory type code to obtain OTD data, and generate dynamic resource data.

[0077] Furthermore, optical fiber inventory information is obtained through the enterprise ERP interface, and the optical fiber reel number, length L and type code are extracted to generate a structured inventory list {reel number: type code, L}. Based on the type code in the inventory list, an API request is sent to the supplier system to obtain the on-time delivery rate OTD of the corresponding optical fiber type; the OTD data is combined with the inventory list to calculate the availability rate R_avail = L_usable / L_total × 100%, and finally generate dynamic resource data.

[0078] S2. De-noise, classify and normalize the collected optical fiber performance parameters, production equipment parameters, environmental parameters and dynamic resource data.

[0079] S2.1. Eliminate outliers, noise data, and complete data on the collected optical fiber performance parameters, production equipment parameters, environmental parameters, and dynamic resource data.

[0080] Furthermore, the 3σ-ROBUST algorithm is first used to process the attenuation value α and dispersion coefficient D1310 / D1550 in the optical fiber performance parameters, calculate the median μ and 1.483 times the median absolute deviation σ, and eliminate data points outside the range of μ±3σ. The coating speed v and stranding tension T_filtered in the production equipment parameters are subjected to sliding average filtering, with the window width set to 5 sampling points. For the relative humidity RH and ambient temperature T in the environmental parameters, wavelet transform denoising is used, and the number of decomposition layers is set to 3. For the on-time delivery rate OTD in the dynamic resource data, KNN missing value filling is performed, with the number of nearest neighbors k=3 and the weight wi=1 / distance squared. Finally, the cleaned complete data set is output.

[0081] S2.2. Interpolate or predict missing data, standardize data of different dimensions, classify and summarize data, and organize and classify them by fiber type, production batch, etc.

[0082] Furthermore, for the missing mode field diameter (MFD) values ​​in the optical fiber performance parameters, the cubic spline interpolation method is used to fill in the missing values ​​based on the known MFD data of the adjacent optical fiber IDs. For the missing sheath extruder temperature (Temp2) in the production equipment parameters, the ARIMA time series model is used to predict and fill in the missing values ​​based on the Temp1 and Temp3 data of the previous and subsequent time periods. Maximum and minimum normalization processing is performed to map the coating speed (v) to the [0,1] interval, and logarithmic normalization processing is used for the dispersion coefficient (D1310 / D1550). The data are divided into categories such as G.652D and G.657A according to the optical fiber type code, and then sorted by the production batch number within each category to finally generate a structured data set.

[0083] S3. Build an optical cable fiber selection optimization model through statistical analysis and advanced machine learning to predict the impact of different fiber selection schemes on optical cable performance.

[0084] S3.1. Convert the optical fiber performance parameters, production equipment parameters, environmental parameters, and dynamic resource data after denoising, classification, and normalization into feature vectors.

[0085] Furthermore, the fiber performance parameters, production equipment parameters, environmental parameters and dynamic resource data after denoising, classification and normalization are aligned and mapped according to the unified timestamp, fiber ID and fiber type code to construct a feature vector for prediction; the feature vector includes the attenuation value α, dispersion coefficient D1310, dispersion coefficient D1550, mode field diameter MFD in the fiber performance parameters, the purification tension T_filtered(t), feed zone temperature Temp1(t), melting zone temperature Temp2(t), die zone temperature Temp3(t) in the production equipment parameters, and the relative humidity R in the environmental parameters. H(t), ambient temperature T(t), vibration spectrum Vrms(f), fiber reel length L, on-time delivery rate OTD and availability rate R_avail in dynamic resource data; all of the above parameters have completed outlier elimination, noise data cleaning and missing completion operations, and after normalization processing, they are uniformly organized according to the same dimension to form a numerical input set. Before training the optical cable fiber selection optimization model, the input set is grouped according to the fiber type code (such as G.652D, G.657A) and production batch number to construct a multi-sample, multi-dimensional feature combination, and finally form an input feature vector set for evaluating the impact of the fiber selection scheme on the optical cable performance.

[0086] S3.2. Label each production batch with a quality grade label based on the tolerance level of the technical indicators and output the label dataset.

[0087] Furthermore, the performance parameter set of all optical fibers in each production batch is first compared item by item with the technical indicator tolerance level L = [l_U, l_Q, l_H]; among them, the attenuation value α, dispersion coefficient D1310, dispersion coefficient D1550, mode field diameter MFD are used as optical fiber performance parameter indicators, the feed zone temperature Temp1, melting zone temperature Temp2, die zone temperature Temp3 and purification tension T_filtered are used as production equipment parameter indicators, the relative humidity RH and ambient temperature T are used as environmental parameter indicators, the on-time delivery rate OTD and availability rate R_avail are used as dynamic information. Source data indicators; each indicator is matched and judged with the level interval [l_U, l_Q, l_H] in the technical indicator tolerance level L. If all key indicators in a production batch are within the l_H interval, the quality level of the batch is marked as "H"; if all are within the l_Q interval or part are within the l_Q interval and the rest are within the l_H interval, it is marked as "Q"; if any indicator falls into the l_U interval, it is marked as "U"; after the above judgment is completed, the production batch number and the corresponding quality level label are combined into a structured label data set, and the output format is {production batch number: quality level}.

[0088] S3.3. Form a Pearson correlation coefficient matrix based on the label data set to obtain a feature subset.

[0089] Furthermore, the quality grade labels corresponding to each production batch number are numerically mapped, for example, the quality grades "H", "Q", and "U" are mapped to the values ​​3, 2, and 1 respectively; the mapped quality grade values ​​are then used as dependent variables, and the fiber performance parameters, production equipment parameters, environmental parameters, and dynamic resource data of the corresponding batch are input as independent variables to form a symmetric Pearson correlation coefficient matrix with all input variables as rows and columns, in which each element represents the degree of linear correlation between any two variables. In the correlation coefficient matrix, the independent variables whose absolute value of the correlation coefficient corresponding to the quality grade value is greater than the specified threshold are selected as important features to form a feature subset.

[0090] S3.4. Based on the optical fiber performance parameters in the feature subset, the K-means algorithm is used to divide the optical fibers into k groups and output the clustering results.

[0091] Furthermore, the K-means algorithm was used to cluster the fiber performance parameters in the feature subset as input variables. Four indicators, including attenuation α, dispersion coefficient D1310, dispersion coefficient D1550, and mode field diameter (MFD), were selected. First, these fiber performance parameters were constructed as multidimensional vectors according to their characteristic dimensions, and all vector data was normalized to eliminate dimensional differences. The number of clusters was set to k, and k initial cluster centers were randomly initialized in the fiber performance parameter vector space. The Euclidean distance of each fiber sample to each cluster center was calculated, and the sample was assigned to the group to which the closest cluster center belonged. The mean of all fiber samples in each group was used as the new cluster center. The distance calculation, group assignment, and center update operations were repeated until all cluster center positions converged and remained unchanged or the maximum number of iterations was reached. After clustering was completed, the group label corresponding to each fiber was output and associated with the fiber ID, ultimately outputting the clustering results.

[0092] S3.5. Based on the parameters of the advance algorithm, a support vector machine is used to construct an optical cable fiber selection optimization model, and a comprehensive objective function is used to train the optical cable fiber selection optimization model, and the trained optical cable fiber selection optimization model is output.

[0093] Furthermore, the groups to which each optical fiber belongs, its corresponding optical fiber performance parameters, production equipment parameters, environmental parameters and dynamic resource data are uniformly constructed into a training sample vector set, and the optical cable performance evaluation index is used as the output label. The support vector machine method is used to construct an optical cable fiber selection optimization model; in the support vector machine training process, the kernel function type is selected and hyperparameters such as the penalty parameter C and the kernel function coefficient γ are set, the samples are mapped and the optimal hyperplane is constructed to achieve classification or regression between samples in high-dimensional space; in the training process, a comprehensive objective function including the mean optical cable attenuation, composite dispersion offset and structural stability targets is used as the optimization criterion, and the support vector machine is trained by minimizing the comprehensive objective function, and finally the trained optical cable fiber selection optimization model is obtained.

[0094] S3.6. Input the feature vector into the trained optical cable fiber selection optimization model to predict the impact of different fiber selection schemes on optical cable performance and obtain a predicted value.

[0095] Furthermore, based on the trained optical cable fiber selection optimization model, the feature vector corresponding to the optical fiber selection scheme to be evaluated is used as input, including the optical fiber performance parameters, production equipment parameters, environmental parameters and dynamic resource data of each optical fiber, to ensure that the input dimensions are consistent with those during the training of the optical cable fiber selection optimization model; by mapping the input feature vectors to the support vector space item by item, the kernel function similarity with the support vector in training is calculated, and the output result is calculated by combining the support vector set and bias term obtained through training; the output result is the predicted comprehensive performance index value of the optical cable, including the normalized average attenuation of the optical cable, the dispersion stability index and the structural consistency index, and the overall performance value J_pred is calculated by the comprehensive objective function, which reflects the predicted impact of the current fiber selection scheme on the optical cable performance, and finally forms a predicted value.

[0096] S4. Use a multi-objective optimization algorithm to calculate the optimal solution for the multi-objective optical fiber arrangement sequence and combination method.

[0097] S4.1. Convert the feature vector into a four-dimensional target vector based on the comprehensive objective function and output the optimization target set.

[0098] Furthermore, the various indicators in the input feature vector are transformed according to the definition structure of the comprehensive objective function to form a four-dimensional optimization target vector; the comprehensive objective function consists of four parts: average cable attenuation, dispersion coefficient fluctuation value, structural consistency index and predicted comprehensive performance value. Specifically, the normalized average cable attenuation is recorded as A_norm, which represents the optical transmission loss performance, the standard deviation of the dispersion coefficients D1310 and D1550 is recorded as D_var, which represents the optical signal waveform retention capability, and the in-batch standard deviation of the optical fiber tension and mode field diameter is jointly expressed as the structural consistency index S_cons, which reflects the manufacturing uniformity; the predicted comprehensive performance value J_pred output by the trained optical cable fiber selection optimization model is used as the fourth-dimensional performance target; the above four indicators are constructed into an optimization target vector according to a unified dimension, and all fiber selection schemes to be evaluated are processed one by one, and finally a complete set of optimization targets is output.

[0099] S4.2. Based on the fiber waste length record, the waste matrix is ​​obtained according to the factory tolerance and the tolerance multiplication factor. According to the fiber substitutability rule, the substitution vector is generated by counting the supplier material substitution records.

[0100] Furthermore, classification statistics are performed according to different optical fiber types and production batches. Combined with the factory tolerance and the tolerance multiplier factor, the tolerance upper limit L_max = L_tol × M_f of each optical fiber is calculated to determine whether the actual waste length exceeds the tolerance range; if it exceeds, it is marked as a waste item under the corresponding optical fiber type and batch index, forming a two-dimensional waste matrix, where the rows represent the optical fiber type, the columns represent the production batch, and the matrix elements are the differences between the actual waste length and the tolerance upper limit; while constructing the waste matrix W, based on the optical fiber substitutability rule, the material substitution records from the supplier are counted, the pairing information of the main material and the substitute material in the substitution relationship is extracted, and the substitution path is generated according to the optical fiber ID; based on the principles of the same type, similar attenuation value, and close dispersion parameters, the acceptable substitute optical fiber IDs of each type of optical fiber are extracted, and the material substitution frequency is counted and sorted, and finally a structured substitution vector is formed.

[0101] S4.3. Based on the waste matrix and the substitution vector, set the constraint conditions and the intersubstitutability vector of the same type of optical fiber. Based on the eigenvector, integrate the multiple groups of performance parameters corresponding to the same optical fiber ID into row vectors to obtain the optical fiber performance matrix.

[0102] Furthermore, based on the fiber ID marked as out of tolerance in the waste matrix, the available fiber ID of the same type in the corresponding alternative vector is located; based on conditions such as consistent fiber type, close dispersion coefficient D1310 and D1550, similar attenuation value α and mode field diameter MFD, alternative fibers are screened, and all alternative relationships are mapped and corrected using the mutual substitution vector. After completing the screening of alternative fiber IDs, based on the feature vector content, all fiber performance parameters corresponding to each fiber ID are extracted, including multiple groups of time-series acquisition or measurement data such as attenuation value α, dispersion coefficient D1310 and D1550, and mode field diameter MFD, and multiple groups of performance parameters with the same fiber ID are integrated in a dimensionally aligned manner to form a row vector indexed by the fiber ID; finally, the row vectors of all fiber IDs are arranged in ascending order by index number and combined into a fiber performance matrix.

[0103] S4.4. Use a 12-bit binary encoding rule to encode the fiber performance matrix into an initial population.

[0104] Furthermore, the indicators in each row of the optical fiber performance vector are coded and converted according to the 12-bit binary coding rule. First, the maximum and minimum value range of each performance indicator is determined, and each indicator value is mapped to 0 to 21 according to the range. 2-1, and then convert the corresponding integer value into a 12-bit binary code segment. For example, the normalized value of the mode field diameter MFD is 0.65, the mapping integer is 2662, and the binary value is "101001100110". In this way, all the indicators in each row of the fiber performance vector are encoded into corresponding 12-bit binary strings in sequence, and spliced ​​in the order of indicators to generate a complete coding sequence with a length of 12×n, where n is the number of indicators contained in the fiber performance vector. The coding sequence corresponding to each fiber ID is regarded as an individual, and the coding results of all fiber IDs are aggregated into the initial population.

[0105] S4.5. Calculate the fitness of each individual in the initial population, output the scored population, use tournament selection and single-point crossover to process the scored population, and output the offspring population.

[0106] Furthermore, the optical fiber performance parameters corresponding to the individual are decoded and restored using the optical cable selection fiber optimization model, the corresponding comprehensive performance index value is calculated, and the comprehensive performance index is converted into a fitness score according to the preset fitness function to form a scored population, in which each individual coding sequence is associated with a fitness score. The tournament selection method is used to randomly select several individuals to form a competition group, compare the fitness scores, and select individuals with higher fitness to enter the next step. The selected individual coding sequences are paired with a single-point crossover operation to determine a random crossover point, and the binary sequences of the two individual codes after the point are exchanged to generate a new offspring coding sequence. All offspring codes generated by the crossover operation are collected to form an offspring population.

[0107] S4.6. Perform bit flipping on the offspring population using the mutation rate and output the new generation population.

[0108] Furthermore, the binary code is traversed bit by bit according to the preset mutation rate, and a random probability judgment is performed on each bit. When the random value is less than the mutation rate, the binary value of the bit is flipped from "0" to "1", or from "1" to "0". After completing the bit-by-bit mutation operation on all individuals, a new generation population containing the mutated individuals is generated.

[0109] S4.7. Use genetic algorithm to sort the fitness of the new generation population and obtain the optimal solution for the multi-objective optical fiber arrangement sequence and combination method.

[0110] Furthermore, based on the coding sequence of each individual in the new generation population, the optical cable fiber selection optimization model is used to calculate its corresponding comprehensive performance index, and the comprehensive performance index is converted into a multi-objective fitness score; the new generation population is sorted according to the fitness score, and several individual coding sequences with the best performance in multi-objective optimization indicators are screened out; based on the sorting results, the optimal solution for the optical fiber arrangement sequence and combination method is determined, so that multiple objectives such as optical fiber performance, waste rate, mutual substitutability and production constraints are optimally balanced, and finally the optimal solution for the multi-objective optical fiber arrangement sequence and combination method is output.

[0111] S5. Apply the optimal solution of the arrangement sequence and combination method of the multi-target optical fibers to the production line to obtain production data.

[0112] S5.1. Convert the optimal solution of the arrangement sequence and combination mode of the multi-target optical fibers into equipment control parameters according to the process mapping table and output an instruction set.

[0113] Furthermore, based on the optimal solution for multi-objective optical fiber arrangement sequence and assembly method, the process mapping table is consulted, and each optical fiber arrangement position and combination information is mapped to a specific equipment control parameter. According to the mapping rules, the optical fiber arrangement sequence and assembly method are converted into executable control instructions one by one to form an instruction set containing equipment operation instructions. The instruction set includes the optical fiber placement sequence, assembly method parameters and related process requirements, ensuring that the equipment can accurately execute optical fiber arrangement and assembly, and finally output a complete equipment control parameter instruction set.

[0114] S5.2. A structured production work order is obtained based on the instruction set, and the speed parameters in the structured production work order are sent to the coating machine through the industrial bus, and the speed is adjusted to the target value in real time.

[0115] Furthermore, based on the instruction set, a structured production work order is generated, which includes the speed parameters of the optical fiber coating process; the speed parameters in the structured production work order are sent to the coating machine through the industrial bus. After receiving the speed parameters, the coating machine adjusts the operating speed to the target speed value in real time; the speed adjustment process continuously monitors the current speed to ensure that it is consistent with the target speed, thereby achieving precise speed control in the optical fiber coating process.

[0116] S5.3. According to the tension parameters in the structured production work order, the tension of the stranding equipment is controlled to the target value through servo control.

[0117] Furthermore, based on the tension parameters in the structured production work order, the servo controller obtains the target tension value, collects the current tension signal of the stranding equipment in real time, compares the current tension with the target tension, calculates the error, and adjusts the drive output of the stranding equipment through the servo control algorithm, continuously adjusting the tension of the stranding equipment to the target value, ensuring that the tension during the stranding process is stable and meets the process requirements, thereby achieving precise control and dynamic adjustment of the tension.

[0118] S5.4. Adjust the temperature of the sheath extruder through the PID algorithm and output the temperature adjustment value.

[0119] Furthermore, based on the temperature parameters of the three zones of the sheath extruder collected in the structured production work order, the deviation between the set temperature and the real-time measured temperature is calculated, and the proportional-integral-differential (PID) algorithm is used to process the temperature deviation. The adjustment amounts of the proportional term, integral term and differential term are calculated respectively, and the temperature adjustment amount output signal is obtained comprehensively. The heating or cooling device is driven to adjust the temperature of the three zones of the sheath extruder, and the temperature deviation is corrected by dynamic feedback to ensure that the temperature parameters are stable near the target set value, meet the process requirements, and realize precise adjustment and dynamic control of the temperature.

[0120] S5.5. Based on the environmental sensor data, the workshop humidity is dynamically adjusted, the optical cable attenuation value is scanned every 100 meters, the deviation of the predicted value is calculated, and the production data is obtained.

[0121] Furthermore, based on the ambient relative humidity and ambient temperature in the environmental parameter set, by real-time collection of environmental sensor data, the operating status of the workshop humidity facilities is dynamically adjusted using a feedback control algorithm to adjust the humidity to within the target range. The optical fiber attenuation value is scanned every 100 meters on the optical cable production line, and the actual attenuation data is collected and compared with the attenuation value predicted by the optical cable selection fiber optimization model to calculate the predicted deviation; the environmental adjustment process data and the optical cable attenuation prediction deviation are recorded synchronously to form production data containing environmental adjustment effects and optical cable performance deviation information.

[0122] S6. Evaluate the fiber matching effect through quality inspection equipment to obtain optimized results.

[0123] S6.1. Use an optical time domain reflectometer to scan the optical cable at 100-meter intervals and calculate the measured loss value.

[0124] Specifically, the expression is,

[0125]

[0126] Among them, α actual is the measured attenuation value, L is the test section length, P out is the output optical power, P in is the test segment length.

[0127] S6.2. Acquire an image of the fiber arrangement using an optical microscope and output an offset dataset.

[0128] Specifically, the expression is,

[0129]

[0130] Where Δd is the offset data set, (x i y i ) is the measured core position, and (x0y0) is the theoretical core position.

[0131] S6.3. Compare the measured attenuation value with the predicted value to form a standardized deviation and output a deviation report.

[0132] Specifically, the expression is,

[0133]

[0134] Where δ is the standardized deviation, L threshold is the technical indicator level, α pred is the predicted value.

[0135] S6.4. Calculate the standard deviation based on the data set and output a stability score.

[0136] Specifically, the expression is,

[0137]

[0138] Among them, σ d is the standard deviation of the data set, N is the number of samples, μ d is the average offset, Δd i is a single offset and i is a sample index.

[0139] S6.5. Obtain tension adjustment instructions through compensation algorithm.

[0140] Specifically, the expression is,

[0141] ΔT=0.2×(σ d -0.5);

[0142] Among them, ΔT is the tension adjustment instruction.

[0143] S6.6. Update the weights of the optical cable selection optimization model based on the deviation report and stability score to obtain the optimization results.

[0144] Furthermore, based on the optical cable attenuation prediction deviation report and the stability score of the optical cable comprehensive performance, the deviation value and stability score are used as feedback signals, combined with the weight adjustment mechanism in the support vector machine, and the weight parameters in the optical cable fiber selection optimization model are iteratively updated through the gradient descent method; in each iteration, the weight adjustment amount is calculated to minimize the prediction error and improve the performance stability index until the preset convergence condition is met or the maximum number of training rounds is reached; after completing the weight update, the optimization result is obtained.

[0145] S7. Enter the production data and optimization results into the knowledge base to form a dynamic optimization plan.

[0146] S7.1. Enter the production data and optimization results of each production process into the production knowledge base to form a reusable fiber selection strategy, support the production needs of different personalities and types of optical cables, and form a dynamic optimization plan.

[0147] Furthermore, the production data of each production process and the optimization results of the corresponding optical cable fiber selection optimization model after training are structured and entered into the production knowledge base to form knowledge entries containing production batch information, optical fiber performance parameters, production equipment parameters, environmental parameters, dynamic resource data and optimization solutions; based on the production knowledge base, according to the personalized parameters and optical cable type codes of different customer needs, historical optimization solutions are quickly matched and called to support dynamic response to diversified optical cable production needs; by continuously updating the production knowledge base, the reuse and iterative optimization of the fiber selection strategy are realized, and the dynamic optimization and adaptive adjustment of the production plan are promoted to ensure that the fiber selection plan in the production process can meet the performance and quality requirements of different optical cable products.

[0148] This embodiment also provides a computer device, which is suitable for the method of optimizing fiber selection for optical cable production guided by industrial big data, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement the method of optimizing fiber selection for optical cable production guided by industrial big data as proposed in the above embodiment.

[0149] The computer device may be a terminal, comprising a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner may be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a button, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse.

[0150] This embodiment also provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the fiber selection optimization method for guiding optical cable production with industrial big data as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, disk or optical disk.

[0151] In summary, this invention utilizes two innovative modules: multi-source data fusion and acquisition, and Yujin machine learning optimization, to construct a complete intelligent fiber selection system. During the data acquisition phase, fiber performance parameters, equipment operating parameters, environmental status, and supply chain data are simultaneously acquired, employing Kalman filtering and far-field scanning to ensure data accuracy. During the optimization decision-making phase, a hybrid model is constructed based on the Yujin algorithm. By defining a three-dimensional demand vector and a comprehensive objective function, an improved genetic algorithm is combined to achieve multi-objective Pareto optimal solution search. Ultimately, dynamic optimization capabilities are achieved through closed-loop feedback from the knowledge base.

[0152] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A fiber selection optimization method for optical cable production guided by industrial big data, characterized by: include, Collect optical fiber performance parameters, production equipment parameters, environmental parameters and dynamic resource data; De-noising, classification and normalization of collected optical fiber performance parameters, production equipment parameters, environmental parameters and dynamic resource data; Build an optical fiber selection optimization model through statistical analysis and advanced machine learning to predict the impact of different fiber selection schemes on optical cable performance; Use multi-objective optimization algorithm to calculate the optimal solution for multi-objective optical fiber arrangement sequence and combination mode; Apply the optimal solution of multi-objective optical fiber arrangement sequence and combination method to the production line to obtain production data, and evaluate the fiber matching effect through quality inspection equipment to obtain optimization results; Enter production data and optimization results into the knowledge base to form a dynamic optimization plan.

2. The method for optimizing fiber selection in optical cable production guided by industrial big data according to claim 1, characterized in that: Collecting optical fiber performance parameters, production equipment parameters, environmental parameters and dynamic resource data includes the following steps: Collect fiber attenuation values ​​and output structured data packets, which are measured using a dispersion analyzer to generate parameter groups; Based on the parameter group, the mode field diameter is calculated by the far-field scanning method to obtain the fiber performance parameters; Collect coating speed output time series data stream, use Kalman filter to process tension sensor raw data, and output purified tension sequence; Record the temperature of the three zones of the sheath extruder based on the purification tension sequence and generate production equipment parameters; Use SHT35 sensor to record environmental RH / T data, output environmental vector, bind vibration spectrum, and obtain environmental parameters; Read the fiber reel number and length, output the inventory list, call the supplier API to obtain OTD data based on the inventory type code, and generate dynamic resource data.

3. The method for optimizing fiber selection in optical cable production guided by industrial big data according to claim 2, characterized in that: The collected optical fiber performance parameters, production equipment parameters, environmental parameters and dynamic resource data are subjected to denoising, classification and normalization, including the following steps: Eliminate outliers, noise data, and complete data on collected optical fiber performance parameters, production equipment parameters, environmental parameters, and dynamic resource data; Interpolate or predict missing data, standardize data of different dimensions, classify and summarize data, and organize and classify them according to fiber type, production batch, etc.

4. The method for optimizing fiber selection in optical cable production guided by industrial big data according to claim 3, characterized in that: Through statistical analysis and advanced machine learning, an optical fiber selection optimization model is constructed to predict the impact of different optical fiber selection schemes on optical cable performance, including the following steps: Convert the fiber performance parameters, production equipment parameters, environmental parameters and dynamic resource data after denoising, classification and normalization into feature vectors; According to the tolerance level of technical indicators, each production batch is labeled with a quality grade label and a label data set is output; Form a Pearson correlation coefficient matrix based on the label data set to obtain the feature subset; Based on the fiber performance parameters in the feature subset, the K-means algorithm is used to divide the fibers into k groups and output the clustering results; Based on the parameters of the Yuejin algorithm, a support vector machine is used to build an optical cable fiber selection optimization model, and a comprehensive objective function is used to train the optical cable fiber selection optimization model, and the trained optical cable fiber selection optimization model is output; The feature vector is input into the trained optical cable fiber selection optimization model to predict the impact of different fiber selection schemes on optical cable performance and obtain the predicted value.

5. The method for optimizing fiber selection in optical cable production guided by industrial big data according to claim 4, characterized in that: The multi-objective optimization algorithm is used to calculate the optimal solution for the multi-objective optical fiber arrangement sequence and combination mode, including the following steps: Based on the comprehensive objective function, the feature vector is converted into a four-dimensional target vector and the optimization target set is output; Based on the fiber waste length records, the waste matrix is ​​obtained according to the factory tolerance and tolerance multiplication factor. According to the fiber substitutability rule, the substitution vector is generated by counting the supplier material substitution records. Based on the waste matrix and substitution vector, constraints and the inter-substitutability vector of the same type of optical fiber are set. Based on the eigenvector, multiple sets of performance parameters corresponding to the same optical fiber ID are integrated into row vectors to obtain the optical fiber performance matrix. Using 12-bit binary coding rules, the fiber performance matrix is ​​encoded into the initial population; Calculate the fitness of each individual in the initial population, output the scored population, use tournament selection and single-point crossover to process the scored population, and output the offspring population; Perform bit flipping on the offspring population using the mutation rate and output a new generation of population; The fitness of the new generation population is sorted by genetic algorithm to obtain the optimal solution for the multi-objective optical fiber arrangement sequence and combination method.

6. The method for optimizing fiber selection in optical cable production guided by industrial big data according to claim 5, characterized in that: Apply the optimal solution of multi-target optical fiber arrangement sequence and combination method to the production line to obtain production data. The following steps are included: The optimal solution for the arrangement sequence and combination of multiple target optical fibers is converted into equipment control parameters according to the process mapping table and output as an instruction set; A structured production work order is obtained based on the instruction set, and the speed parameters in the structured production work order are sent to the coating machine through the industrial bus to adjust the speed to the target value in real time; According to the tension parameters in the structured production work order, the tension of the stranding equipment is controlled to the target value through servo control; Adjust the sheath extruder temperature through PID algorithm and output the temperature adjustment value; Based on environmental sensor data, the workshop humidity is dynamically adjusted, the optical cable attenuation value is scanned every 100 meters, and the deviation of the predicted value is used to obtain production data.

7. The method for optimizing fiber selection in optical cable production guided by industrial big data according to claim 6, characterized in that: The fiber matching effect is evaluated by quality inspection equipment to obtain the optimization results, including the following steps: Use an optical time domain reflectometer to scan the optical cable at 100-meter intervals to calculate the measured attenuation value; Acquire fiber alignment images using an optical microscope and output an offset dataset; Compare the measured attenuation value with the predicted value to form a standardized deviation and output a deviation report; Calculate the standard deviation based on the data set and output a stability score; Through the compensation algorithm, the tension adjustment instruction is obtained; The optical cable selection fiber optimization model weights are updated based on the deviation report and stability score to obtain the optimization results.

8. The method for optimizing fiber selection in optical cable production guided by industrial big data according to claim 7, characterized in that: Enter the production data and optimization results into the knowledge base to form a dynamic optimization plan, including the following steps: The production data and optimization results of each production process are entered into the production knowledge base to form a reusable fiber selection strategy, support the production needs of different personalities and types of optical cables, and form a dynamic optimization plan.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method for optimizing fiber selection for optical cable production guided by industrial big data are implemented as described in any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for optimizing fiber selection for optical cable production guided by industrial big data according to any one of claims 1 to 8 are implemented.

Images