A data processing method and system for the research and development of plastic products

Abstract

The present invention discloses a method and system for processing R & D data of plastic products, specifically relating to the technical field of plastic products; by comprehensively analyzing the extracted material property complexity features and market feedback real-time features through fuzzy logic, the reliability of the product prediction model can be accurately evaluated. According to the prediction results of the model, the results are further divided into accurate prediction results, possibly accurate prediction results, and inaccurate prediction results, and corresponding processing measures are taken according to different categories, which not only improves the R & D efficiency, but also ensures the market adaptability and quality of the products, helps enterprises better meet market demands, and thus enhances the overall competitiveness.

Description

Technical Field

[0001] The present invention relates to the technical field of plastic products, and particularly relates to a method and system for processing research and development data of plastic products. Background Art

[0002] Processing research and development data for plastic products refers to collecting, organizing, and analyzing data related to material properties, processing technologies, product designs, etc. during the research and development process of plastic products. This data can include experimental results, market research, user feedback, and monitoring data during the production process. Through data processing, the research and development team can identify potential areas for improvement, optimize material formulations and production processes to improve product performance and market competitiveness.

[0003] During the data processing process, technologies such as statistical analysis, data visualization, and machine learning are usually used. These technologies can help researchers extract valuable information from complex data sets, predict the performance of products in actual use, and quickly verify design hypotheses. Ultimately, through effective data processing, enterprises can develop plastic products that meet market demands more efficiently, thereby improving the overall research and development efficiency and product quality. However, when constructing a prediction model, if too much attention is paid to the details in the training data, the model may capture noise rather than real patterns. Although the model performs excellently on the training data, in actual applications, the prediction results may be inaccurate, leading the research and development team to misevaluate the reliability and performance of the product. Summary of the Invention

[0004] The purpose of the present invention is to provide a method and system for processing research and development data of plastic products to solve the deficiencies in the background art.

[0005] To achieve the above purpose, the present invention provides the following technical solutions: A method for processing research and development data of plastic products, comprising the following steps:

[0006] S1: Collect research and development data of plastic products from different data sources, where the research and development data of plastic products includes material property data and market feedback data, and preprocess the obtained research and development data of plastic products;

[0007] S2: Extract features from the preprocessed research and development data of plastic products, and respectively extract the material property complexity feature in the material property data and the market feedback real-time feature in the market feedback data;

[0008] S3: Based on the extracted material property complexity feature and market feedback real-time feature, determine the accuracy of the product prediction model through comprehensive analysis using fuzzy logic;

[0009] S4: According to the prediction results, divide the prediction results of the product prediction model into accurate prediction results, possible accurate prediction results, and inaccurate prediction results, and perform corresponding processing.

[0010] Preferably, in S2, generate a material property complexity index based on the material property complexity characteristics in the extracted material property data. The method for obtaining the material property complexity index is as follows:

[0011] Obtain the property data of all materials, construct a data set, perform standardization processing on the data, and calculate the covariance matrix for the standardized data. The expression is: ; where C is the covariance matrix, X′ is the standardized data matrix, n is the number of samples, T is the matrix transpose, calculate the eigenvalues and eigenvectors of the covariance matrix to identify the principal components: ; λ is the eigenvalue, I is the identity matrix, select the principal components according to the eigenvalue size, select the eigenvalues with a cumulative variance ratio reaching 85% - 95% as the principal components, project the standardized data onto the selected principal components to obtain the coordinates Z of each sample in the principal component space, and calculate the material property complexity index through the scores of the selected principal components. The expression is: ; where p is the number of selected principal components, is the eigenvalue of the kth principal component, and MP is the material property complexity index.

[0012] Preferably, in S2, generate a feedback frequency fluctuation index based on the market feedback timeliness characteristics in the extracted market feedback data. The method for obtaining the feedback frequency fluctuation index is as follows:

[0013] Obtain the market feedback data within the H time period, select the time window W, and within this time window, count the number of feedbacks, count the number of received feedbacks for each time unit to generate a feedback frequency sequence: ; calculate the mean value of the feedback frequency within the W time window. The expression is: ; where MF is the mean value of the feedback frequency, calculate the fluctuation value of the feedback frequency. The expression is: ; where is the fluctuation value of the feedback frequency, and perform a weighted average summation calculation on the fluctuation values of the feedback frequencies within each time window to obtain the feedback frequency fluctuation index.

[0014] Preferably, in S3, based on the extracted material property complexity characteristics and market feedback timeliness characteristics, determine the accuracy of the product prediction model through comprehensive analysis using fuzzy logic. Specifically:

[0015] Take the material property complexity index MP and the feedback frequency fluctuation index KL as the input items of fuzzy logic, and take the accuracy level F of the product prediction model as the output item of fuzzy logic;

[0016] For the material property complexity index, the feedback frequency fluctuation index, and the accuracy level of the product prediction model, divide them into fuzzy sets;

[0017] According to the relationship between the input variables, formulate fuzzy rules to determine the value of the output variable;

[0018] Convert the actual material property complexity index and the feedback frequency fluctuation index into fuzzy values;

[0019] Evaluate the fuzzy input according to the fuzzy rules, and aggregate the output fuzzy sets of all rules. Usually, the maximum aggregation method is used to form the output fuzzy set;

[0020] Convert the fuzzy output value into a clear output value, and the final result obtained is the accuracy level of the product prediction model.

[0021] Preferably, in S4, for the possible accuracy prediction results, according to the change of the model prediction accuracy, dynamically adjust the product. Specifically:

[0022] Calculate the abnormal degree value of the model prediction accuracy. The expression is: : In the formula, TP is the true positive example, TN is the true negative example, FP is the false positive example, FN is the false negative example, and A is the abnormal degree value; calculate the adjustment coefficient. The expression is: ; In the formula, AF is the adjustment coefficient, K is a constant, is the target prediction accuracy rate, and the adjusted material characteristics are: ; M is the current material characteristic, is the adjustment amplitude, indicating the amount of each adjustment.

[0023] The present invention also provides a data processing system for plastic product R & D, including a data acquisition module, a feature extraction module, an accuracy analysis module, and a processing module;

[0024] Data acquisition module: Collect plastic product R & D data from different data sources. The plastic product R & D data includes material property data and market feedback data, and preprocess the obtained plastic product R & D data;

[0025] Feature extraction module: Extract features from the preprocessed plastic product R & D data, and respectively extract the material property complexity feature in the material property data and the market feedback real-time feature in the market feedback data;

[0026] Accuracy Analysis Module: Based on the extracted material property complexity features and market feedback timeliness features, after comprehensively analyzing them using fuzzy logic, it determines the accuracy of the product prediction model;

[0027] Processing Module: According to the prediction results, it classifies the prediction results of the product prediction model into accurate prediction results, possibly accurate prediction results, and inaccurate prediction results, and performs corresponding processing.

[0028] In the above technical solution, the technical effects and advantages provided by the present invention:

[0029] 1. By making full use of technologies such as statistical analysis, data visualization, and machine learning, the present invention can effectively extract valuable information from complex datasets. This process includes comprehensive collection and preprocessing of material property data and market feedback data, feature extraction, fuzzy logic analysis, and classification processing of prediction results, thus significantly improving the R & D efficiency and market adaptability of products. By generating the material property complexity index and feedback frequency fluctuation index, the R & D team can more accurately evaluate product performance and make necessary dynamic adjustments to ensure that decisions in the R & D process are based on reliable data analysis.

[0030] 2. Through the data acquisition module, feature extraction module, accuracy analysis module, and processing module, each module cooperates with each other to form an efficient workflow. Such a design not only improves the automation and intelligence level of data processing but also provides enterprises with the ability to flexibly respond to market changes. Through accurate model prediction, enterprises can quickly respond to market demands, reduce R & D risks, and ultimately promote the improvement of the quality and performance of plastic products, thereby enhancing market competitiveness. Brief Description of the Drawings

[0031] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required to be used in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0032] Figure 1 It is a method flow chart of a method for processing R & D data of plastic products according to the present invention.

[0033] Figure 2 It is a module schematic diagram of a system for processing R & D data of plastic products according to the present invention. Detailed Embodiments

[0034] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0035] Example 1. Refer to Figure 1 and 2 as shown. A method for processing research and development data of plastic products in this embodiment includes the following steps:

[0036] S1: Collect research and development data of plastic products from different data sources. The research and development data of plastic products includes material property data and market feedback data, and preprocess the obtained research and development data of plastic products.

[0037] S2: Extract features from the preprocessed research and development data of plastic products, and respectively extract the material property complexity feature in the material property data and the market feedback timeliness feature in the market feedback data.

[0038] S3: Based on the extracted material property complexity feature and market feedback timeliness feature, determine the accuracy of the product prediction model through comprehensive analysis using fuzzy logic.

[0039] S4: According to the prediction results, divide the prediction results of the product prediction model into accurate prediction results, possibly accurate prediction results, and inaccurate prediction results, and perform corresponding processing.

[0040] Among them, in S1, collect research and development data of plastic products from different data sources. The research and development data of plastic products includes material property data and market feedback data, and preprocess the obtained research and development data of plastic products. Specifically:

[0041] Collect research and development data of plastic products from different data sources. The research and development data of plastic products includes material property data and market feedback data. Among them, the material property data can be collected through laboratory tests, literature reviews, industry standards, databases, and other channels. The data usually includes the physical properties of the material (such as density, strength, ductility, etc.), chemical properties (such as corrosion resistance, thermal stability, etc.), processing properties (such as melt index, molding temperature, etc.), etc.

[0042] The market feedback data can be obtained through user surveys, customer reviews, market research reports, social media analysis, etc. The data usually includes consumers' satisfaction with plastic products, usage experience, function evaluation, purchase intention, price sensitivity, etc.

[0043] After collecting material property data and market feedback data, the following preprocessing steps are required:

[0044] Check whether there are missing values in the dataset and take appropriate measures to handle them. Options include deleting records with missing values, filling missing values with the mean / median, or using interpolation for estimation. Identify and remove outliers in the data to ensure the overall quality of the dataset. For example, detect outliers through box plots or the Z-score method. For features with different units in the material property data, standardization or normalization is required so that different features can be compared on the same scale. The market feedback data may contain categorical data (such as the grading of user satisfaction), which needs to be encoded (such as one-hot encoding) for subsequent analysis.

[0045] Identify and select features relevant to the R & D goal. Methods such as correlation analysis and principal component analysis (PCA) can be used to extract important features and reduce the data dimension. Combine different features to generate new features. For example, combine the physical properties of materials with the user evaluations in the market feedback to create a comprehensive scoring feature. Integrate the material property data and the market feedback data to form a unified dataset. Connection can be made according to key fields such as product model and batch. Ensure that the integrated dataset is consistent in format, structure, and unit for subsequent analysis and modeling.

[0046] Use data visualization tools (such as scatter plots, heat maps, bar charts, etc.) to analyze the preprocessed data and identify potential trends, patterns, and relationships. In the final stage of preprocessing, check the integrity and consistency of the data to ensure its reliability in subsequent analysis and modeling.

[0047] S2: Extract features from the preprocessed R & D data of plastic products, and extract the material property complexity feature from the material property data and the market feedback timeliness feature from the market feedback data respectively.

[0048] Before feature extraction, it is first necessary to deeply understand the structure, type, and information contained in the data. The main data sources include: Material property data: such as strength, toughness, thermal stability, density, etc. Market feedback data: such as customer ratings, user reviews, sales data, market trends, etc.

[0049] Extraction of basic performance indicators: Extract the basic physical and chemical properties of each material (such as tensile strength, elastic modulus, melt index, etc.) as basic features.

[0050] Performance Composite Features: Synthetic Features: Combine multiple basic performance metrics to generate composite features. For example, create a "strength / density ratio" feature to evaluate the performance efficiency of materials. Statistical Features of Performance Metrics: Calculate the mean, standard deviation, minimum, maximum, etc. of performance metrics to form statistical features.

[0051] Complexity Features: Diversity Index: Evaluate the diversity of material performance by calculating the variance or standard deviation of multiple performance metrics. Feature Interaction: Capture the relationships between different performance metrics by creating interaction features (such as the product of strength and ductility). Dimensionality Reduction and Principal Component Analysis (PCA): Use PCA to reduce the dimensionality of high-dimensional performance data, extract the main components to reduce redundancy while retaining most of the information.

[0052] Conduct sentiment analysis on user comments, extract the proportions of positive and negative feedback, and the comprehensive sentiment score as features. Calculate the timestamps of the feedback, and extract the response time (such as the number of days from product launch to the first feedback) as a feature. Statistically count the number of feedback within a specific time period to evaluate product popularity. Statistically analyze user ratings (such as star ratings), and extract features such as the mean, distribution, and extreme values.

[0053] Generate a material performance complexity index based on the material performance complexity features in the extracted material performance data. The method for obtaining the material performance complexity index is as follows:

[0054] Obtain the performance data of all materials and construct a dataset. The dataset should contain multiple features, such as tensile strength, elastic modulus, and ductility. Before performing PCA, it is first necessary to standardize the data to eliminate the influence of different feature magnitudes. Calculate the covariance matrix for the standardized data. The expression is: ; where C is the covariance matrix, X′ is the standardized data matrix, n is the number of samples, T is the matrix transpose. Calculate the eigenvalues and eigenvectors of the covariance matrix to identify the main components: ; λ is the eigenvalue, I is the identity matrix. Select the main components according to the eigenvalue magnitudes. Select the eigenvalues with a cumulative variance proportion reaching 85% - 95% as the main components. Project the standardized data onto the selected main components to obtain the coordinates Z of each sample in the main component space. Calculate the material performance complexity index through the scores of the selected main components. The expression is: ; where p is the number of selected main components, is the eigenvalue of the k-th main component, and MP is the material performance complexity index.

[0055] The larger the material property complexity index, the more it usually indicates that the material has higher diversity and complexity in multiple performance characteristics, which provides a richer information basis for the product prediction model. When the performance characteristics of the material are more complex and diverse, the model can better learn the potential patterns of the material's performance under different conditions, thereby enhancing its prediction ability. A higher material property complexity index means that the material shows stronger adaptability when facing different environments and application scenarios, which provides more comprehensive training data for the model.

[0056] Secondly, materials with a higher complexity index often involve multiple interacting performance indicators, such as strength, toughness, and heat resistance, etc. This enables the model to identify the complex relationships between these characteristics during construction. This complex relationship can help the model capture the key performance of the material in actual use, thereby improving the prediction accuracy. By exploring these potential interactions, the model can more effectively handle non-linear and multi-dimensional data and reduce the risk of overfitting.

[0057] Finally, when the material property complexity index increases, the prediction model may also show higher reliability during the verification process. The high-complexity material data provides a wider range of test samples for the evaluation of the model, allowing the R & D team to make more accurate performance predictions in actual applications. This can not only improve the accuracy of product design but also provide guidance for subsequent material optimization, thereby ultimately enhancing the market competitiveness and user satisfaction of the product.

[0058] Generate a feedback frequency fluctuation index based on the market feedback timeliness characteristics in the extracted market feedback data. The method for obtaining the feedback frequency fluctuation index is as follows:

[0059] Obtain the market feedback data within the H time period, including the number of feedbacks received each day or week. This data can come from user reviews, product ratings, or other market surveys. Select a suitable time window W (such as 7 days, 30 days, or a specific sales cycle) to calculate the feedback frequency. Within this time window, count the number of feedbacks received daily or weekly. For each time unit (such as daily or weekly), count the number of feedbacks received to generate a feedback frequency sequence: ; Calculate the mean value of the feedback frequency within the W time window. The expression is: ; In the formula, MF is the mean value of the feedback frequency. Calculate the fluctuation value of the feedback frequency. The expression is: ; In the formula, is the fluctuation value of the feedback frequency. After performing a weighted average summation calculation on the fluctuation values of the feedback frequencies within each time window, the feedback frequency fluctuation index is obtained.

[0060] The larger the feedback frequency fluctuation index, it generally means that the volatility of the number of market feedbacks on the product is stronger, which may reflect the instability of the product's acceptance and attention in the market. When the volatility of the feedback frequency is large, the model may face more challenges because the unstable feedback data makes it difficult for the model to capture consistent trends and patterns during the learning stage. In this way, when the model makes future predictions, it may produce larger errors and reduce the prediction accuracy.

[0061] Secondly, highly volatile feedback may mean that market demands change rapidly, and consumers' evaluations of products are affected by various factors, such as seasonal changes, promotional activities, or competitors' market strategies. Such external influences will cause drastic fluctuations in the feedback data, making it impossible for the prediction model to effectively adapt to these changes. As a result, the model may not be able to accurately evaluate the product's performance under different market conditions, thus affecting R & D and market decisions.

[0062] Finally, the increase in the feedback frequency fluctuation index may also indicate significant user divergence and disagreement about the product. If negative and positive feedback coexist and change frequently, it will be difficult for the model to determine the true market performance of the product. Such uncertainty reduces the reliability of the model when evaluating the success or failure of the product, and may ultimately lead to wrong business decisions. Therefore, the increase in the feedback frequency fluctuation index generally means a decrease in the accuracy of the product prediction model, and further measures need to be taken to improve the data quality and the adaptability of the model.

[0063] S3: Based on the extracted material property complexity features and market feedback timeliness features, after comprehensively analyzing them using fuzzy logic, determine the accuracy of the product prediction model.

[0064] Take the material property complexity index MP and the feedback frequency fluctuation index KL as the input items of fuzzy logic, and take the accuracy level F of the product prediction model as the output item of fuzzy logic;

[0065] For the material property complexity index (MP), divide it into fuzzy sets, such as: Low, Medium, High. For the feedback frequency fluctuation index (KL), divide it into fuzzy sets, such as: Low Variation, Medium Variation, High Variation.

[0066] The accuracy level (F) of the product prediction model is divided into fuzzy sets, such as: Inaccurate, Moderate, Accurate.

[0067] Based on the relationships between input variables, formulate fuzzy rules to determine the values of output variables. For example, the following fuzzy rules can be used:

[0068] If MP is low and KL has low fluctuations, then F is accurate. If MP is low and KL has medium fluctuations, then F is possibly accurate. If MP is low and KL has high fluctuations, then F is inaccurate. If MP is medium and KL has low fluctuations, then F is accurate. If MP is medium and KL has medium fluctuations, then F is possibly accurate. If MP is medium and KL has high fluctuations, then F is inaccurate. If MP is high and KL has low fluctuations, then F is possibly accurate. If MP is high and KL has medium fluctuations, then F is inaccurate. If MP is high and KL has high fluctuations, then F is inaccurate.

[0069] Convert the actual Material Performance Complexity Index (MP) and Feedback Frequency Fluctuation Index (KL) into fuzzy values. Fuzzification can be achieved using Membership Functions. For example:

[0070] For MP, the following membership function can be defined:

[0071] Low: ; Medium: ; High: ;

[0072] Evaluate the fuzzy inputs according to the fuzzy rules. Use the "Max - Min" method to combine the input values with the fuzzy rules to determine the activation degree of each rule.

[0073] Aggregate the output fuzzy sets of all rules. Usually, the Max Aggregation method is used to form the output fuzzy set. The activation values for each output category (inaccurate, possibly accurate, accurate) can be obtained.

[0074] Convert the fuzzy output values into clear output values (Defuzzification). Common defuzzification methods include the Centroid Method.

[0075] The final result obtained is the accuracy level (F) of the product prediction model, which can be used to support product development and market decisions.

[0076] For example: Material Performance Complexity Index (MP) = 70, Feedback Frequency Fluctuation Index (KL) = 20; Fuzzify the input data: Calculate the membership degrees of MP and KL through the set membership functions. For example: MP = 70 > High (0.8), Medium (0.2), Low (0); KL = 20 > Low Fluctuation (0.9), Medium Fluctuation (0.1), High Fluctuation (0);

[0077] For example, activate Rule 1: If MP is Low and KL is Low Fluctuation, then F is Accurate. The activation degree of this rule is 0. Rule 5 is activated: If MP is Medium and KL is Low Fluctuation, then F is Probably Accurate. The activation degree of this rule is 0.2 * 0.9 = 0.18. Rule 8 is activated: If MP is High and KL is Medium Fluctuation, then F is Inaccurate. The activation degree of this rule is 0.8 * 0.1 = 0.08.

[0078] Sum up the membership degrees of each output category to get Probably Accurate (0.18) and Inaccurate (0.08).

[0079] Calculate the centroid value of the output to obtain the final output result.

[0080] Assume that the final defuzzification result is "Probably Accurate", then the accuracy level of the model is Probably Accurate, which means that the performance of the prediction model is acceptable, but there is still room for improvement.

[0081] S4: According to the prediction results, classify the prediction results of the product prediction model into accurate prediction results, probably accurate prediction results, and inaccurate prediction results, and perform corresponding processing.

[0082] According to the prediction results, classify the prediction results of the product prediction model into accurate prediction results, probably accurate prediction results, and inaccurate prediction results, and perform corresponding processing, specifically as follows:

[0083] For accurate prediction results, such prediction results indicate that the prediction of the model is very close to the actual market performance, and the product performance and market feedback are relatively ideal. For the prediction results marked as accurate, it is recommended to give priority to their use in production and market promotion. Such products can be marketed more widely to utilize their advantages. Conduct in-depth analysis of these accurate prediction results to identify the success factors, such as material selection, design features, market positioning, etc., to form best practices and provide references for the development of future products. Although these results indicate that the model is accurate, continuous monitoring of market feedback is still required to ensure that the product maintains its performance under different market conditions.

[0084] For inaccurate prediction results, such results indicate significant differences between the model's predictions and actual market performance, suggesting possible quality issues or market rejection of the product. Conduct in-depth analysis of these results to identify the causes of inaccurate predictions. It may be necessary to trace back to data collection, model design, or parameter settings for problem troubleshooting. Re-evaluate the products identified as inaccurate, including material properties, design characteristics, and market feedback. Decide whether product improvement or redesign is needed based on the analysis results. If, after detailed analysis, it is confirmed that the product has serious quality problems or is not accepted by the market, consider temporarily halting production and making comprehensive improvements. Establish a sound feedback mechanism to feed inaccurate prediction results back to the R & D team, prompting the team to conduct in-depth discussions and formulate solutions to the problems.

[0085] For potentially accurate prediction results, such results indicate a certain degree of uncertainty between the model's predictions and actual performance. Dynamically adjust the product according to the changes in the model's prediction accuracy to flexibly address potential problems.

[0086] Calculate the anomaly degree value of the model prediction accuracy, with the expression: : In the formula, TP is the true positive example (correctly predicted as accurate), TN is the true negative example (correctly predicted as inaccurate), FP is the false positive example (incorrectly predicted as accurate), FN is the false negative example (incorrectly predicted as inaccurate), and A is the anomaly degree value; Calculate the adjustment coefficient, with the expression: ; In the formula, AF is the adjustment coefficient, K is a constant representing the flexibility or sensitivity of the adjustment, is the target prediction accuracy rate, and the adjusted material property is: ; M is the current material property (such as strength, elasticity, etc.), is the adjustment range, representing the amount of adjustment each time.

[0087] In this embodiment, during the R & D process of plastic products, first comprehensively collect R & D data from different data sources, including material property data and market feedback data, and preprocess this data to ensure data quality and consistency. Then, through feature extraction techniques, extract material property complexity features from the material property data and market feedback real-time features from the market feedback data respectively. Based on these extracted features, use fuzzy logic to comprehensively analyze the data to determine the accuracy of the product prediction model. Finally, according to the model's prediction results, classify them into accurate prediction results, potentially accurate prediction results, and inaccurate prediction results, and take corresponding treatment measures according to different categories to optimize product performance and market performance.

[0088] Example 2. A data processing system for plastic product R & D described in this example includes a data acquisition module, a feature extraction module, an accuracy analysis module, and a processing module;

[0089] Data acquisition module: Collect plastic product R & D data from different data sources. The plastic product R & D data includes material property data and market feedback data, and preprocess the acquired plastic product R & D data;

[0090] Feature extraction module: Extract features from the preprocessed plastic product R & D data, and respectively extract the material property complexity feature in the material property data and the market feedback real-time feature in the market feedback data;

[0091] Accuracy analysis module: Based on the extracted material property complexity feature and market feedback real-time feature, determine the accuracy of the product prediction model through comprehensive analysis using fuzzy logic;

[0092] Processing module: According to the prediction result, divide the prediction result of the product prediction model into accurate prediction result, possible accurate prediction result, and inaccurate prediction result, and perform corresponding processing.

[0093] The above formulas are all dimensionless and take their numerical values for calculation. The formula is a formula obtained by software simulation of a large amount of collected data to approximate the real situation as closely as possible. The preset parameters in the formula are set by those skilled in the art according to the actual situation.

[0094] It should be understood that the term "and / or" in this article is merely a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. Here, A and B can be singular or plural. In addition, the character " / " in this article generally represents an "or" relationship between the associated objects before and after, but it may also represent an "and / or" relationship. Specifically, it can be understood by referring to the context before and after.

[0095] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed in this article can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0096] As described above, it is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application.

Claims

1. A method for processing data for plastic product research and development, characterized in that: The steps include: S1: collecting plastic product R&D data from different data sources, wherein the plastic product R&D data includes material performance data and market feedback data, and preprocessing the acquired plastic product R&D data; S2: Extract features from the pre-processed plastic product R&D data, and extract the material performance complexity features from the material performance data and the market feedback real-time features from the market feedback data; The material performance complexity index is generated according to the material performance complexity characteristics in the extracted material performance data. The method for obtaining the material performance complexity index is: obtain the performance data of all materials, build a data set, standardize the data, and calculate the covariance matrix of the standardized data. The expression is: ; Where C is the covariance matrix, X′ is the standardized data matrix, n is the number of samples, T is the matrix transpose, and the eigenvalues ​​and eigenvectors of the covariance matrix are calculated to identify the principal components: ; λ is the eigenvalue, I is the unit matrix, and the principal component is selected according to the size of the eigenvalue. The eigenvalue with a cumulative variance ratio of 85%-95% is selected as the principal component. The standardized data is projected onto the selected principal component to obtain the coordinate Z of each sample in the principal component space. The material performance complexity index is calculated by the selected principal component score, and the expression is: ; Where p is the number of principal components selected, is the eigenvalue of the kth principal component, MP material performance complexity index; S3: Based on the extracted material performance complexity characteristics and market feedback real-time characteristics, the accuracy of the product prediction model is determined by comprehensive analysis using fuzzy logic; S4: According to the prediction results, the prediction results of the product prediction model are divided into accurate prediction results, possible accurate prediction results and inaccurate prediction results, and corresponding processing is performed.

2. A method for processing data for plastic product research and development according to claim 1, characterized in that: In S2, the feedback frequency fluctuation index is generated according to the real-time characteristics of the market feedback in the extracted market feedback data. The method for obtaining the feedback frequency fluctuation index is: Get the market feedback data within the H time period, select the time window W, count the number of feedbacks within this time window, count the number of feedbacks received for each time unit, and generate a feedback frequency sequence: ; Calculate the mean of the feedback frequency in the W time window, the expression is: ; In the formula, MF is the mean value of the feedback frequency. The fluctuation value of the feedback frequency is calculated as follows: ; In the formula, is the fluctuation value of the feedback frequency. The feedback frequency fluctuation index is obtained by performing weighted average summation calculation on the fluctuation values ​​of the feedback frequency in each time window.

3. A method for processing data for plastic product R&D according to claim 2, characterized in that: In S3, based on the extracted material performance complexity characteristics and market feedback real-time characteristics, the accuracy of the product prediction model is determined by using fuzzy logic for comprehensive analysis, specifically: The material performance complexity index MP and the feedback frequency fluctuation index KL are used as the input items of fuzzy logic, and the accuracy level F of the product prediction model is used as the output item of fuzzy logic; For the material property complexity index, feedback frequency fluctuation index and the accuracy level of the product prediction model, they are divided into fuzzy sets; According to the relationship between input variables, fuzzy rules are formulated to determine the value of output variables; Convert the actual material performance complexity index and feedback frequency fluctuation index into fuzzy values; Evaluate the fuzzy input according to the fuzzy rules, aggregate the output fuzzy sets of all rules, and usually use the maximum aggregation method to form the output fuzzy set; The fuzzy output values ​​are converted into crisp output values, and the final result is the accuracy level of the product prediction model.

4. A method for processing data for plastic product research and development according to claim 3, characterized in that: In S4, for the possible accuracy prediction results, the product is dynamically adjusted according to the changes in the model prediction accuracy, specifically: Calculate the abnormal degree value of the model prediction accuracy, the expression is: : In the formula, TP is a true positive example, TN is a true negative example, FP is a false positive example, FN is a false negative example, and A is the abnormality value; the adjustment coefficient is calculated as follows: ; In the formula, AF is the adjustment coefficient, K is a constant, Adjusted material properties for target prediction accuracy for: ; M is the current material property, It is the adjustment range, indicating the amount of each adjustment.

5. A data processing system for plastic product research and development, used to implement a data processing method for plastic product research and development as claimed in any one of claims 1 to 4, characterized in that: Data acquisition module, feature extraction module, accuracy analysis module and processing module; Data acquisition module: collects plastic product R&D data from different data sources, including material performance data and market feedback data, and pre-processes the acquired plastic product R&D data; Feature extraction module: extracts features from the pre-processed plastic product R&D data, and extracts the material performance complexity features from the material performance data and the market feedback real-time features from the market feedback data; Accuracy analysis module: Based on the extracted material performance complexity characteristics and market feedback real-time characteristics, the accuracy of the product prediction model is determined by comprehensive analysis using fuzzy logic; Processing module: According to the prediction results, the prediction results of the product prediction model are divided into accurate prediction results, possible accurate prediction results and inaccurate prediction results, and corresponding processing is performed.

Images