Supply chain service quality evaluation system and method based on big data

Abstract

The invention relates to the technical field of quality evaluation, and discloses a supply chain service quality evaluation system and method based on big data. The method comprises the following steps: firstly, acquiring network questionnaire sample data, and performing reliability and validity test on the network questionnaire sample data; secondly, training an LSTM neural network to obtain an LSTM neural network model, and marking label values for the network questionnaire sample data; estimating the supply chain service quality evaluation data based on a Gaussian mixture model, deleting abnormal data, obtaining an objective evaluation value through a principal component analysis method, and setting a threshold value to judge the objective evaluation value to obtain processed supply chain service quality evaluation data; and finally, constructing a cloud model by using the processed supply chain service quality evaluation data, and completing supply chain service quality evaluation according to the supply chain service quality condition. Supply chain service quality evaluation is realized by processing supply chain service quality evaluation data, and the result is accurate and objective.

Description

Technical Field

[0001] The present invention relates to the technical field of quality assessment, and in particular to a supply chain service quality assessment system and method based on big data. Background Art

[0002] Chinese patent CN111563662A discloses a service quality evaluation system and method based on time-window deep reinforcement learning. The system specifically includes a data collection module for collecting various data of the service quality evaluation object, including multi-dimensional data of the service itself, service provider-related data, and consumer evaluation data; a model adjustment module for building a service quality evaluation model and designing a reinforcement learning model adjustment action set; a parallel learning module for parallel learning of multiple service providers; a reward feedback module for designing and summarizing the delayed rewards of the evaluation model; a Q table update module for maximizing the reward obtained by adjusting the reward action according to the algorithm update formula; a cycle iteration module for controlling the cycle iteration of the service quality evaluation model; a prediction learning module for predicting the reward of the adjustment action under different service quality evaluation models; and a time window adjustment module for predicting the model evaluation effect based on the time window of the service quality evaluation model under different circumstances. The system evaluates the service quality by building a service quality evaluation system, and the service quality evaluation data is not objective.

[0003] Traditional service quality evaluation uses subjective service scoring, which may result in large differences between different evaluators and is random, resulting in high evaluation costs. Alternatively, it uses traditional network objective evaluation, but this method is prone to cheating or malicious behavior and has poor evaluation results. At the same time, it does not use high-tech technologies such as machine learning and neural networks, and has disadvantages such as slow data processing speed and the presence of abnormal data. Summary of the invention

[0004] In response to the problems in the related technology, the present invention provides a supply chain service quality evaluation system and method based on big data to overcome the above-mentioned technical problems existing in the existing related technology.

[0005] In order to solve the above technical problems, the present invention is achieved through the following technical solutions:

[0006] The present invention is a supply chain service quality evaluation method based on big data, comprising the following steps:

[0007] S1. According to the supply chain service quality evaluation standard, obtain the online survey questionnaire sample data, conduct reliability and validity tests on the online survey questionnaire sample data, and obtain the processed online survey questionnaire sample data;

[0008] S2. Train the LSTM neural network to obtain an LSTM neural network model, and mark label values ​​according to the LSTM neural network model, and generate supply chain service quality evaluation data in combination with the processed online questionnaire sample data;

[0009] S3. Estimate the supply chain service quality evaluation data based on a Gaussian mixture model, delete abnormal supply chain service quality evaluation data, and obtain processed supply chain service quality evaluation data; then use the principal component analysis method to extract the principal component scores of the processed supply chain service quality evaluation data, calculate the objective evaluation value based on the principal component scores, and obtain the processed supply chain service quality evaluation data by determining the objective evaluation value;

[0010] S4. Use the processed supply chain service quality evaluation data to generate an indicator evaluation matrix, and use the entropy weight method to determine the entropy weight of the indicators in the indicator evaluation matrix. Build a supply chain service quality cloud based on the entropy weight of the indicators, judge the supply chain service quality by calculating the similarity, and complete the supply chain service quality evaluation.

[0011] The invention obtains sample data of online survey questions, makes reliability judgment on the sample data of online survey questionnaires by calculating Cronbach's α coefficient, and realizes validity judgment by calculating the correlation of sample data of online survey questionnaires in combination with KMO test method and Bartlett sphericity test method; through reliability and validity test, reliability and credibility analysis of sample data of online survey questionnaires can be realized; secondly, LSTM neural network is trained to obtain LSTM neural network model, and supply chain service quality evaluation data is obtained by marking label values ​​of sample data of online survey questionnaires; the neural network can mark new data, avoiding the tedious process of manual marking, which is convenient for the next step. step; then estimate the supply chain service quality evaluation data based on the Gaussian mixture model, delete the abnormal data, extract the principal component score through the principal component analysis method, calculate the objective evaluation value, and judge the objective evaluation value by setting a threshold to obtain the processed supply chain service quality evaluation data; this processing method can eliminate the influence of cheating data and improve the authenticity of the data; finally, use the entropy weight method to calculate the entropy weight of the indicator, build a supply chain service quality cloud, use the similarity algorithm to calculate the similarity, obtain the supply chain service quality situation by matching the similarity interval, and complete the supply chain service quality evaluation; introduce the cloud model in the evaluation process, which effectively solves the problem of data ambiguity and improves the evaluation accuracy.

[0012] Preferably, the S1 comprises the following steps:

[0013] S11. Formulate an online questionnaire according to the supply chain service quality evaluation standard, and use the online questionnaire method to obtain the online questionnaire. The online questionnaire includes a1 first-level indicators, a2 second-level indicators and an answer part in the supply chain service quality evaluation system, wherein the answer part includes five levels, namely very satisfied, satisfied, general, dissatisfied and very dissatisfied, and the corresponding scores are 5 points, 4 points, 3 points, 2 points and 1 point respectively. The a1 first-level indicator and a2 second-level indicators constitute the indicator, and the online questionnaire sample data is obtained; the number of questions in the online questionnaire is set to b, and the variance of the score of the i1th online questionnaire question is The variance of the total score of the online questionnaire is The Cronbach's α coefficient calculation formula is as follows:

[0014]

[0015] Calculate and obtain the Cronbach's α coefficient value, set the reliability threshold as ω, and when the Cronbach's α coefficient value is greater than the reliability threshold, the online survey questionnaire sample data is valid, otherwise the online survey questionnaire sample data is suspected to be invalid, and the reliability test of the online survey questionnaire sample data is completed;

[0016] S12. Select variables from the online questionnaire sample data, record them as the first variable and the second variable, calculate the simple correlation coefficient and partial correlation coefficient between the first variable and the second variable, set the sum of squares of the simple correlation coefficients of the first variable and the second variable to be a′, and the sum of squares of the partial correlation coefficients of the first variable and the second variable to be a″, then the KMO (sampling suitability) test calculation formula is as follows:

[0017]

[0018] Among them, β1 represents the KMO test value;

[0019] A correlation coefficient matrix is ​​established for the online questionnaire sample data, and the correlation coefficient matrix is ​​a unit matrix. The determinant of the correlation coefficient matrix is ​​calculated, and the Bartlett sphericity test calculation formula is as follows:

[0020] β2=det(A);

[0021] Among them, β2 represents the Bartlett sphericity test value, A represents the correlation coefficient matrix, and det(A) represents the value of the determinant of the correlation coefficient matrix;

[0022] S13. According to the KMO test value and the Bartlett sphericity test value, the chi-square distribution table is queried to obtain the corresponding accompanying probability value, and the correlation of the online survey questionnaire sample data at the significance level is obtained; the test threshold is set to ξ, when the KMO test value is greater than the test threshold and the online survey questionnaire sample data is significantly correlated, the online survey questionnaire sample data is valid; otherwise, the online survey questionnaire sample data is invalid, and a new round of online survey questionnaire is started; the validity test of the online survey questionnaire sample data is completed to obtain the processed online survey questionnaire sample data.

[0023] The invention obtains sample data of online survey questionnaires, makes reliability judgment on the sample data of online survey questionnaires by calculating Cronbach's α coefficient, and uses KMO test method and Bartlett sphericity test method to calculate the correlation of sample data of online survey questionnaires to achieve validity judgment; through reliability and validity tests, data reliability and credibility can be improved.

[0024] Preferably, S2 comprises the following steps:

[0025] S21, setting the LSTM (Long Short-Term Memory) neural network to include an input gate, a forget gate and an output gate, taking the current state of the sigmoid layer in the forget gate as output, and the previous layer of the current state as input, updating the current cell state, and obtaining update information; creating a vector in the tanh layer in the forget gate and adding it to the current cell state, calculating the sum of the product of the old cell state and the update information and the vector, and updating the cell state again; outputting the product of the outputs of the sigmoid layer and the tanh layer from the output gate, and obtaining the output result of the LSTM neural network;

[0026] Acquire the service quality evaluation data to be processed from the network to obtain the service quality evaluation data set to be processed, and score the service quality evaluation data to be processed in the service quality evaluation data set to be processed according to the five levels, and use the score as the label value of the service quality evaluation data to be processed; then divide the service quality evaluation data set to be processed into an evaluation training set and an evaluation verification set, input the evaluation training set into the LSTM neural network, and iterate continuously until the LSTM neural network converges to obtain a trained LSTM neural network; then input the evaluation verification set into the trained LSTM neural network, set the error threshold to ψ, and according to the label value, when the output result error is less than the error threshold, obtain the LSTM neural network model, otherwise adjust the neuron weights and biases until the output result error is less than the error threshold;

[0027] S22. Input the processed online questionnaire sample data into the LSTM neural network model, and the LSTM neural network model outputs the label value corresponding to the processed online questionnaire sample data; then conduct multiple online questionnaires based on the supply chain service quality evaluation standard to obtain the online questionnaire sample data to be processed, and input the online questionnaire sample data to be processed into the LSTM neural network model to obtain the label value corresponding to the online questionnaire sample data to be processed; then combine the processed online questionnaire sample data and the online questionnaire sample data to be processed and record them as the supply chain service quality evaluation data.

[0028] The invention obtains an LSTM neural network model by training the LSTM neural network, which can realize labeling of sample data of online questionnaires; the neural network can directly label new data with high accuracy, avoiding the tedious process of manual labeling and facilitating the next step of processing.

[0029] Preferably, S3 comprises the following steps:

[0030] S31, normalizing the supply chain service quality evaluation data to obtain normalized supply chain service quality evaluation data to form a normalized supply chain service quality evaluation data set, and the labels corresponding to the normalized supply chain service quality evaluation data set are recorded as label set B = {b1, b2, b3, ..., b m}, where b m represents the label value corresponding to the mth normalized supply chain service quality evaluation data; setting a Gaussian mixture model, inputting the normalized supply chain service quality evaluation data set into the Gaussian mixture model, estimating the Gaussian mixture model based on the expectation maximization algorithm, deleting abnormal normalized supply chain service quality evaluation data, and obtaining processed supply chain service quality evaluation data. The specific steps are as follows:

[0031] S311, initialize the parameters of the Gaussian mixture model, set the potential state, calculate the probability of observing the normalized supply chain service quality assessment data in the normalized supply chain service quality assessment data set under the potential state, record it as the likelihood function, and then calculate the posterior probability of the normalized supply chain service quality assessment data in the normalized supply chain service quality assessment data set under the potential state; use the likelihood function and the posterior probability to estimate the parameters of the Gaussian mixture model to obtain the parameters of the Gaussian mixture model for the second iteration;

[0032] S312, repeating S311 for the second iteration of the Gaussian mixture model parameters, continuously iterating, setting the current number of iterations to 2, the maximum number of iterations to C, and when the current number of iterations reaches the maximum number of iterations, stopping the iteration to obtain the final Gaussian mixture model parameters, the final Gaussian mixture model parameters including the Gaussian model distribution cluster distribution mean and the Gaussian model mixing probability;

[0033] S313, using the Gaussian model distribution cluster distribution mean to divide the mixed categories of the normalized supply chain service quality assessment data set, selecting the mixed categories with Gaussian model mixing probability less than 50% in the mixed categories, recording them as abnormal mixed categories, deleting the normalized supply chain service quality assessment data in the abnormal mixed categories and the labels corresponding to the normalized supply chain service quality assessment data, to form processed supply chain service quality assessment data;

[0034] S32, extracting the principal component scores of the processed supply chain service quality evaluation data based on the principal component analysis method to obtain a principal component score set, the specific steps are as follows:

[0035] S321. The processed supply chain service quality evaluation data constitutes a processed supply chain service quality evaluation data set B1 = {b1′, b2′, b3′, ..., b′ m′}, where b′ m′ represents the m′th processed supply chain service quality evaluation data, and the processed supply chain service quality evaluation data contains n′ indicators, and the initial matrix A1 is generated as follows:

[0036]

[0037] Among them, b′ m′n′ represents the sample data corresponding to the n′th indicator of the m′th processed supply chain service quality evaluation data;

[0038] S322, calculating the mean value and variance of the sample data in the initial matrix, and using the mean value and variance to standardize the initial matrix, the calculation formula is as follows:

[0039]

[0040] in, Indicates b′ m′n′ The corresponding standardized data, represents the average value of the sample data in the initial matrix, Represents the variance of the sample data in the initial matrix;

[0041] The standardized data is used to form a standardized matrix A2, and a correlation coefficient matrix A3 is generated from the standardized matrix. The calculation formula is as follows:

[0042]

[0043] Among them, A2 T represents the transpose of a normalized matrix;

[0044] Calculate the eigenvalues ​​and eigenvectors of the correlation coefficient matrix, determine the variance contribution rate according to the eigenvalues ​​and eigenvectors of the correlation coefficient matrix, and select the principal component by taking the eigenvalue greater than 1 as the standard; calculate the principal component score according to the variance contribution rate to form a principal component score set;

[0045] S33. Map the principal component score set to the same scale as the processed supply chain service quality assessment data, and then calculate the arithmetic mean of the principal component score set as the objective assessment value of the supply chain service quality. Set the upper limit of the assessment threshold to ζ1 and the lower limit of the assessment threshold to ζ2. When the objective assessment value is less than or equal to the upper limit of the assessment threshold and the objective assessment value is greater than or equal to the lower limit of the assessment threshold, the processed supply chain service quality assessment data is obtained; otherwise, repeat S31 and S32 until the objective assessment value is between the upper limit of the assessment threshold and the lower limit of the assessment threshold.

[0046] The invention estimates the supply chain service quality evaluation data based on a Gaussian mixture model, deletes abnormal data, and eliminates the impact of cheating data; and extracts principal component scores through principal component analysis, calculates objective evaluation values, and judges objective evaluation values ​​by setting thresholds, thereby maintaining the original characteristics of the data and effectively improving the authenticity of the data.

[0047] Preferably, S4 comprises the following steps:

[0048] S41, the processed supply chain service quality evaluation data constitute a processed supply chain service quality evaluation data set B2 = {b1", b2", b3", ..., b' m ″ ′}, where b′ m ″ ′ represents the m″th processed supply chain service quality evaluation data, and the processed supply chain service quality evaluation data contains n′ indicators, and the indicator evaluation matrix A4 is generated as follows:

[0049]

[0050] Among them, b′ m ″ ′n′ represents the n′th indicator of the m″th processed supply chain service quality evaluation data;

[0051] Setting m″n″ represents the entropy of the n″th indicator of the m″th processed supply chain service quality evaluation data, b′m ″ ′n″ represents the value of the nth indicator of the mth processed supply chain service quality evaluation data, then the entropy calculation formula of the nth indicator is as follows:

[0052]

[0053] Among them, D n″ represents the entropy of the n″th indicator, and δ represents the coefficient;

[0054] S42. Obtain the entropy weight of the n″th indicator according to the entropy of the n″th indicator, and obtain the entropy weights of all indicators in turn; set the supply chain service quality standard cloud, use the indicator as a feature, and combine the entropy weights of all indicators to establish the supply chain service quality cloud; set the indicator coefficient of the eth indicator to be χ e , then the similarity calculation formula is as follows:

[0055]

[0056] Among them, f e represents the similarity coefficient of the e-th index, G e It represents the similarity of the e-th indicator between the supply chain service quality cloud and the supply chain service quality standard cloud;

[0057] The similarities of all indicators of the supply chain service quality cloud and the supply chain service quality standard cloud are obtained in turn, and the service quality interval is set. The similarity interval [0, η1] is recorded as extremely low service quality, the similarity interval (η1, η2] is recorded as low service quality, the similarity interval (η2, η3] is recorded as general service quality, the similarity interval (η3, η4] is recorded as high service quality, and the similarity interval (η4, 1] is recorded as extremely high service quality. According to the service quality interval, the similarities of all indicators of the supply chain service quality cloud and the supply chain service quality standard cloud are matched to obtain the supply chain service quality situation and complete the supply chain service quality evaluation.

[0058] This invention calculates the entropy weight of the indicator by using the entropy weight method, builds a supply chain service quality cloud, calculates the similarity by using the similarity algorithm, obtains the supply chain service quality situation by matching the similarity interval, and completes the supply chain service quality evaluation; the introduction of the cloud model effectively improves the evaluation accuracy.

[0059] The present invention also discloses a system for a supply chain service quality evaluation method based on big data, which specifically includes: an online questionnaire sample data verification module, a neural network model marking module, a supply chain service quality evaluation data processing module and a supply chain service quality cloud construction module;

[0060] The network survey questionnaire sample data verification module is used to perform reliability and validity verification on the network survey questionnaire sample data;

[0061] The neural network model labeling module is used to label the processed online questionnaire sample data based on the LSTM neural network model;

[0062] The supply chain service quality evaluation data processing module is used to delete abnormal supply chain service quality evaluation data and make judgments based on objective evaluation values;

[0063] The supply chain service quality cloud construction module is used to construct a supply chain service quality cloud according to the entropy weight of the indicators in the indicator evaluation matrix to judge the supply chain service quality situation.

[0064] The present invention has the following beneficial effects:

[0065] 1. The invention can improve the reliability and credibility of online survey questionnaire sample data by obtaining online survey questionnaire sample data and making reliability and validity judgments on the online survey questionnaire sample data.

[0066] 2. The invention obtains an LSTM neural network model by training the LSTM neural network, which can realize labeling of sample data of online questionnaires and directly label new data with high accuracy, thus avoiding the tedious process of manual labeling.

[0067] 3. The invention estimates the supply chain service quality evaluation data based on the Gaussian mixture model, deletes abnormal data, and significantly eliminates the impact of cheating data.

[0068] 4. The invention uses principal component analysis to extract principal component scores, calculates objective evaluation values, and judges objective evaluation values ​​by setting thresholds, thereby maintaining the original characteristics of the data and effectively improving the authenticity of the data.

[0069] 5. This invention uses the entropy weight method to calculate the entropy weight of the indicator, constructs a supply chain service quality cloud, and introduces a cloud model, which solves the ambiguity problem in the supply chain service quality evaluation process and effectively improves the evaluation accuracy.

[0070] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] In order to more clearly illustrate the technical solutions of the embodiments of the invention, the drawings required for describing the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For ordinary technicians in this field, they can also obtain drawings based on these drawings without paying creative work.

[0072] Figure 1A schematic diagram of the process of supply chain service quality evaluation by a supply chain service quality evaluation system based on big data provided by the present invention. DETAILED DESCRIPTION

[0073] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0074] In the description of the present invention, it is necessary to understand that the terms "opening", "upper", "lower", "top", "middle", "inside" and the like indicating orientation or positional relationship are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the components or elements referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as limiting the invention.

[0075] Example 1

[0076] See also Figure 1 The present invention is a supply chain service quality evaluation method based on big data, comprising the following steps:

[0077] S1. According to the supply chain service quality evaluation standard, obtain the online survey questionnaire sample data, conduct reliability and validity tests on the online survey questionnaire sample data, and obtain the processed online survey questionnaire sample data;

[0078] The S1 comprises the following steps:

[0079] S11. Formulate an online questionnaire according to the supply chain service quality evaluation standard, and use the online questionnaire method to obtain the online questionnaire. The online questionnaire includes a1 first-level indicators, a2 second-level indicators and an answer part in the supply chain service quality evaluation system, wherein the answer part includes five levels, namely very satisfied, satisfied, general, dissatisfied and very dissatisfied, and the corresponding scores are 5 points, 4 points, 3 points, 2 points and 1 point respectively. The a1 first-level indicator and a2 second-level indicators constitute the indicator, and the online questionnaire sample data is obtained; the number of questions in the online questionnaire is set to b, and the variance of the score of the i1th online questionnaire question is The variance of the total score of the online questionnaire is The Cronbach's α coefficient calculation formula is as follows:

[0080]

[0081] Calculate and obtain the Cronbach's α coefficient value, set the reliability threshold as ω, and when the Cronbach's α coefficient value is greater than the reliability threshold, the online survey questionnaire sample data is valid, otherwise the online survey questionnaire sample data is suspected to be invalid, and the reliability test of the online survey questionnaire sample data is completed;

[0082] S12. Select variables from the online questionnaire sample data, record them as the first variable and the second variable, calculate the simple correlation coefficient and partial correlation coefficient between the first variable and the second variable, set the sum of squares of the simple correlation coefficients of the first variable and the second variable to be a′, and the sum of squares of the partial correlation coefficients of the first variable and the second variable to be a″, then the KMO test calculation formula is as follows:

[0083]

[0084] Among them, β1 represents the KMO test value;

[0085] A correlation coefficient matrix is ​​established for the online questionnaire sample data, and the correlation coefficient matrix is ​​a unit matrix. The determinant of the correlation coefficient matrix is ​​calculated, and the Bartlett sphericity test calculation formula is as follows:

[0086] β2=det(A);

[0087] Among them, β2 represents the Bartlett sphericity test value, A represents the correlation coefficient matrix, and det(A) represents the value of the determinant of the correlation coefficient matrix;

[0088] S13. According to the KMO test value and the Bartlett sphericity test value, the chi-square distribution table is queried to obtain the corresponding accompanying probability value, and the correlation of the online survey questionnaire sample data at the significance level is obtained; the test threshold is set to ξ, and when the KMO test value is greater than the test threshold and the online survey questionnaire sample data is significantly correlated, the online survey questionnaire sample data is valid; otherwise, the online survey questionnaire sample data is invalid, and a new round of online survey questionnaire is started; the validity test of the online survey questionnaire sample data is completed to obtain the processed online survey questionnaire sample data;

[0089] S2. Train the LSTM neural network to obtain an LSTM neural network model, and mark label values ​​according to the LSTM neural network model, and generate supply chain service quality evaluation data in combination with the processed online questionnaire sample data;

[0090] The S2 comprises the following steps:

[0091] S21, setting the LSTM neural network to include an input gate, a forget gate and an output gate, taking the current state of the sigmoid layer in the forget gate as output, and the previous layer of the current state as input, updating the current cell state, and obtaining update information; creating a vector in the tanh layer in the forget gate and adding it to the current cell state, calculating the sum of the product of the old cell state and the update information and the vector, and updating the cell state again; outputting the product of the outputs of the sigmoid layer and the tanh layer from the output gate, and obtaining the output result of the LSTM neural network;

[0092] Acquire the service quality assessment data to be processed from the network to obtain the service quality assessment data set to be processed, and score the service quality assessment data to be processed in the service quality assessment data set to be processed according to the five levels, and use the score as the label value of the service quality assessment data to be processed; then divide the service quality assessment data set to be processed into an evaluation training set and an evaluation verification set, input the evaluation training set into the LSTM neural network, and iterate continuously until the LSTM neural network converges to obtain a trained LSTM neural network; then input the evaluation verification set into the trained LSTM neural network, set the error threshold to ψ, and when the output result error is less than the error threshold, obtain the LSTM neural network model, otherwise adjust the neuron weights and biases until the output result error is less than the error threshold;

[0093] S22, input the processed online questionnaire sample data into the LSTM neural network model, and the LSTM neural network model outputs the label value corresponding to the processed online questionnaire sample data; then conduct multiple online questionnaires of the supply chain service quality evaluation standard to obtain the online questionnaire sample data to be processed, input the online questionnaire sample data to be processed into the LSTM neural network model, and obtain the label value corresponding to the online questionnaire sample data to be processed; then combine the processed online questionnaire sample data and the online questionnaire sample data to be processed, and record them as supply chain service quality evaluation data;

[0094] S3. Estimate the supply chain service quality evaluation data based on a Gaussian mixture model, delete abnormal supply chain service quality evaluation data, and obtain processed supply chain service quality evaluation data; then use the principal component analysis method to extract the principal component scores of the processed supply chain service quality evaluation data, calculate the objective evaluation value based on the principal component scores, and obtain the processed supply chain service quality evaluation data by determining the objective evaluation value;

[0095] The S3 comprises the following steps:

[0096] S31, normalizing the supply chain service quality evaluation data to obtain normalized supply chain service quality evaluation data to form a normalized supply chain service quality evaluation data set, and the labels corresponding to the normalized supply chain service quality evaluation data set are recorded as label set B = {b1, b2, b3, ..., b m}, where b m represents the label value corresponding to the mth normalized supply chain service quality evaluation data; setting a Gaussian mixture model, inputting the normalized supply chain service quality evaluation data set into the Gaussian mixture model, estimating the Gaussian mixture model based on the expectation maximization algorithm, deleting abnormal normalized supply chain service quality evaluation data, and obtaining processed supply chain service quality evaluation data. The specific steps are as follows:

[0097] S311, initialize the parameters of the Gaussian mixture model, set the potential state, calculate the probability of observing the normalized supply chain service quality assessment data in the normalized supply chain service quality assessment data set under the potential state, record it as the likelihood function, and then calculate the posterior probability of the normalized supply chain service quality assessment data in the normalized supply chain service quality assessment data set under the potential state; use the likelihood function and the posterior probability to estimate the parameters of the Gaussian mixture model to obtain the parameters of the Gaussian mixture model for the second iteration;

[0098] S312, repeating S311 for the second iteration of the Gaussian mixture model parameters, continuously iterating, setting the current number of iterations to 2, the maximum number of iterations to C, and when the current number of iterations reaches the maximum number of iterations, stopping the iteration to obtain the final Gaussian mixture model parameters, the final Gaussian mixture model parameters including the Gaussian model distribution cluster distribution mean and the Gaussian model mixing probability;

[0099] S313, using the Gaussian model distribution cluster distribution mean to divide the mixed categories of the normalized supply chain service quality assessment data set, selecting the mixed categories with Gaussian model mixing probability less than 50% in the mixed categories, recording them as abnormal mixed categories, deleting the normalized supply chain service quality assessment data in the abnormal mixed categories and the labels corresponding to the normalized supply chain service quality assessment data, to form processed supply chain service quality assessment data;

[0100] S32, extracting the principal component scores of the processed supply chain service quality evaluation data based on the principal component analysis method to obtain a principal component score set, the specific steps are as follows:

[0101] S321. The processed supply chain service quality evaluation data constitutes a processed supply chain service quality evaluation data set B1 = {b1′, b2′, b3′, ..., b′ m′}, where b′ m′represents the m′th processed supply chain service quality evaluation data, and the processed supply chain service quality evaluation data contains n′ indicators, and the initial matrix A1 is generated as follows:

[0102]

[0103] Among them, b′ m′n′ represents the sample data corresponding to the n′th indicator of the m′th processed supply chain service quality evaluation data;

[0104] S322, calculating the mean value and variance of the sample data in the initial matrix, and using the mean value and variance to standardize the initial matrix, the calculation formula is as follows:

[0105]

[0106] in, Indicates b′ m′n′ The corresponding standardized data, represents the average value of the sample data in the initial matrix, Represents the variance of the sample data in the initial matrix;

[0107] The standardized data is used to form a standardized matrix A2, and a correlation coefficient matrix A3 is generated from the standardized matrix. The calculation formula is as follows:

[0108]

[0109] Among them, A2 T represents the transpose of a normalized matrix;

[0110] Calculate the eigenvalues ​​and eigenvectors of the correlation coefficient matrix, determine the variance contribution rate according to the eigenvalues ​​and eigenvectors of the correlation coefficient matrix, and select the principal component by taking the eigenvalue greater than 1 as the standard; calculate the principal component score according to the variance contribution rate to form a principal component score set;

[0111] S33, mapping the principal component score set to the same scale as the processed supply chain service quality evaluation data, and then calculating the arithmetic mean of the principal component score set as the objective evaluation value of the supply chain service quality, setting the upper limit of the evaluation threshold to ζ1, and the lower limit of the evaluation threshold to ζ2, when the objective evaluation value is less than or equal to the upper limit of the evaluation threshold and the objective evaluation value is greater than or equal to the lower limit of the evaluation threshold, the processed supply chain service quality evaluation data is obtained; otherwise, repeat S31 and S32 until the objective evaluation value is between the upper limit of the evaluation threshold and the lower limit of the evaluation threshold;

[0112] S4. Use the processed supply chain service quality evaluation data to generate an indicator evaluation matrix, and use the entropy weight method to determine the entropy weight of the indicator in the indicator evaluation matrix. According to the entropy weight of the indicator, a supply chain service quality cloud is constructed. The supply chain service quality is judged by calculating the similarity, and the supply chain service quality evaluation is completed.

[0113] The S4 comprises the following steps:

[0114] S41, the processed supply chain service quality evaluation data constitute a processed supply chain service quality evaluation data set B2 = {b1", b2", b3", ..., b' m ″ ′}, where b′ m ″ ′ represents the m″th processed supply chain service quality evaluation data, and the processed supply chain service quality evaluation data contains n′ indicators, and the indicator evaluation matrix A4 is generated as follows:

[0115]

[0116] Among them, b′ m ″ ′n′ represents the n′th indicator of the m″th processed supply chain service quality evaluation data;

[0117] Setting m″n″ represents the entropy of the n″th indicator of the m″th processed supply chain service quality evaluation data, b′ m ″ ′n″ represents the value of the nth indicator of the mth processed supply chain service quality evaluation data, then the entropy calculation formula of the nth indicator is as follows:

[0118]

[0119] Among them, D n″ represents the entropy of the n″th indicator, and δ represents the coefficient;

[0120] S42. Obtain the entropy weight of the n″th indicator according to the entropy of the n″th indicator, and obtain the entropy weights of all indicators in turn; set the supply chain service quality standard cloud, use the indicator as a feature, and combine the entropy weights of all indicators to establish the supply chain service quality cloud; set the indicator coefficient of the eth indicator to be χ e , then the similarity calculation formula is as follows:

[0121]

[0122] Among them, f e represents the similarity coefficient of the e-th index, G eIt represents the similarity of the e-th indicator between the supply chain service quality cloud and the supply chain service quality standard cloud;

[0123] The similarities of all indicators of the supply chain service quality cloud and the supply chain service quality standard cloud are obtained in turn, and the service quality interval is set. The similarity interval [0, η1] is recorded as extremely low service quality, the similarity interval (η1, η2] is recorded as low service quality, the similarity interval (η2, η3] is recorded as general service quality, the similarity interval (η3, η4] is recorded as high service quality, and the similarity interval (η4, 1] is recorded as extremely high service quality. According to the service quality interval, the similarities of all indicators of the supply chain service quality cloud and the supply chain service quality standard cloud are matched to obtain the supply chain service quality situation and complete the supply chain service quality evaluation.

[0124] Example 2

[0125] The present invention also discloses a system for a supply chain service quality evaluation method based on big data, which specifically includes: an online questionnaire sample data verification module, a neural network model marking module, a supply chain service quality evaluation data processing module and a supply chain service quality cloud construction module;

[0126] The network survey questionnaire sample data verification module is used to perform reliability and validity verification on the network survey questionnaire sample data;

[0127] The neural network model labeling module is used to label the processed online questionnaire sample data based on the LSTM neural network model;

[0128] The supply chain service quality evaluation data processing module is used to delete abnormal supply chain service quality evaluation data and make judgments based on objective evaluation values;

[0129] The supply chain service quality cloud construction module is used to construct a supply chain service quality cloud according to the entropy weight of the indicators in the indicator evaluation matrix to judge the supply chain service quality situation.

[0130] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0131] The preferred embodiments of the invention disclosed above are only used to help explain the invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the invention, so that those skilled in the art can understand and use the invention well.

Claims

1. A supply chain service quality evaluation method based on big data, characterized in that: The steps include: S1. According to the supply chain service quality evaluation standard, obtain the online survey questionnaire sample data, conduct reliability and validity tests on the online survey questionnaire sample data, and obtain the processed online survey questionnaire sample data; S2. Train the LSTM neural network to obtain an LSTM neural network model, and mark label values ​​according to the LSTM neural network model, and generate supply chain service quality evaluation data in combination with the processed online questionnaire sample data; S3. Estimate the supply chain service quality evaluation data based on a Gaussian mixture model, delete abnormal supply chain service quality evaluation data, and obtain processed supply chain service quality evaluation data; then use the principal component analysis method to extract the principal component scores of the processed supply chain service quality evaluation data, calculate the objective evaluation value based on the principal component scores, and obtain the processed supply chain service quality evaluation data by determining the objective evaluation value; S4. Use the processed supply chain service quality evaluation data to generate an indicator evaluation matrix, and use the entropy weight method to determine the entropy weight of the indicators in the indicator evaluation matrix. Build a supply chain service quality cloud based on the entropy weight of the indicators, judge the supply chain service quality by calculating the similarity, and complete the supply chain service quality evaluation.

2. According to the big data-based supply chain service quality evaluation method of claim 1, it is characterized in that: The S1 comprises the following steps: S11. Formulate an online survey questionnaire according to the supply chain service quality evaluation standard, obtain sample data of the online survey questionnaire, calculate the Cronbach's α coefficient value, and complete the reliability test of the sample data of the online survey questionnaire by comparing the Cronbach's α coefficient value with the reliability threshold; S12, based on the KMO test method and Bartlett sphericity test method, the KMO test value and the Bartlett sphericity test value are calculated; S13. According to the KMO test value and the Bartlett sphericity test value, the correlation of the online survey questionnaire sample data at the significance level is obtained, and by judging the correlation, the validity test of the online survey questionnaire sample data is completed to obtain the processed online survey questionnaire sample data.

3. According to the big data-based supply chain service quality evaluation method of claim 2, it is characterized in that: The S2 comprises the following steps: S21, acquiring the service quality assessment data to be processed from the network to obtain a service quality assessment data set to be processed, and scoring the service quality assessment data to be processed in the service quality assessment data set to be processed, and using the score as a label value of the service quality assessment data to be processed; Then, the service quality evaluation data set to be processed is divided into an evaluation training set and an evaluation verification set, and the evaluation training set is input into the LSTM neural network, and it is continuously iterated until the LSTM neural network converges to obtain a trained LSTM neural network; then, the evaluation verification set is input into the trained LSTM neural network, and the error threshold is set to ψ. When the output result error is less than the error threshold, the LSTM neural network model is obtained, otherwise, the neuron weights and biases are adjusted until the output result error is less than the error threshold; S22. Input the processed online questionnaire sample data into the LSTM neural network model, and the LSTM neural network model outputs the label value corresponding to the processed online questionnaire sample data; then conduct multiple online questionnaires based on the supply chain service quality evaluation standard to obtain the online questionnaire sample data to be processed, and input the online questionnaire sample data to be processed into the LSTM neural network model to obtain the label value corresponding to the online questionnaire sample data to be processed; then combine the processed online questionnaire sample data and the online questionnaire sample data to be processed and record them as the supply chain service quality evaluation data.

4. According to the big data-based supply chain service quality evaluation method of claim 3, it is characterized in that: The S3 comprises the following steps: S31, estimating the supply chain service quality evaluation data based on a Gaussian mixture model, deleting abnormal supply chain service quality evaluation data, and obtaining processed supply chain service quality evaluation data; S32, extracting the principal component scores of the processed supply chain service quality evaluation data based on the principal component analysis method to obtain a principal component score set; S33. Map the principal component score set to the same scale as the processed supply chain service quality evaluation data, and then calculate the arithmetic mean of the principal component score set as the objective evaluation value of the supply chain service quality, and obtain the processed supply chain service quality evaluation data by judging the relationship between the objective evaluation value and the threshold.

5. The supply chain service quality evaluation method based on big data according to claim 4 is characterized in that: The S31 comprises the following steps: S311, initializing the Gaussian mixture model parameters, calculating the likelihood function and the posterior probability; using the likelihood function and the posterior probability to estimate the Gaussian mixture model parameters, and obtaining the Gaussian mixture model parameters for the second iteration; S312, repeating S311 for the second iteration of the Gaussian mixture model parameters, continuously iterating, setting the current number of iterations to 2, the maximum number of iterations to C, and when the current number of iterations reaches the maximum number of iterations, stopping the iteration to obtain the final Gaussian mixture model parameters, the final Gaussian mixture model parameters including the Gaussian model distribution cluster distribution mean and the Gaussian model mixing probability; S313. Use the Gaussian model distribution cluster distribution mean to divide the supply chain service quality evaluation data into abnormal mixed categories, delete the supply chain service quality evaluation data in the abnormal mixed categories, and obtain processed supply chain service quality evaluation data.

6. According to the big data-based supply chain service quality evaluation method of claim 4, it is characterized in that: The S32 comprises the following steps: S321. Generate an initial matrix using principal component analysis based on the processed supply chain service quality evaluation data; S322, normalizing the initial matrix to obtain a correlation coefficient matrix, calculating the principal component scores according to the variance contribution rate of the correlation coefficient matrix, and forming a principal component score set.

7. The supply chain service quality evaluation method based on big data according to claim 6 is characterized in that: The S4 comprises the following steps: S41, generating an indicator evaluation matrix with the processed supply chain service quality evaluation data, and calculating the entropy weight of the indicators in the indicator evaluation matrix; S42. Set the supply chain service quality standard cloud, and combine the entropy weights of the indicators in the indicator evaluation matrix to establish the supply chain service quality cloud, calculate the similarity between the supply chain service quality cloud and the supply chain service quality standard cloud indicators, judge the supply chain service quality situation by the similarity, and complete the supply chain service quality evaluation.

8. The supply chain service quality evaluation method based on big data according to claim 7 is characterized in that: The method of judging the supply chain service quality by similarity comprises the following steps: Set the service quality interval, record the similarity interval [0, η1] as extremely low service quality, the similarity interval (η1, η2] as low service quality, the similarity interval (η2, η3] as average service quality, the similarity interval (η3, η4] as high service quality, and the similarity interval (η4, 1] as extremely high service quality, and match the similarity of the supply chain service quality cloud and the supply chain service quality standard cloud indicators according to the service quality interval.

9. A system for implementing the supply chain service quality evaluation method based on big data as described in any one of claims 1 to 8, characterized in that: Specifically include: Online questionnaire sample data verification module, neural network model labeling module, supply chain service quality assessment data processing module and supply chain service quality cloud construction module; The network survey questionnaire sample data verification module is used to perform reliability and validity verification on the network survey questionnaire sample data; The neural network model labeling module is used to label the processed online questionnaire sample data based on the LSTM neural network model; The supply chain service quality evaluation data processing module is used to delete abnormal supply chain service quality evaluation data and make judgments based on objective evaluation values; The supply chain service quality cloud construction module is used to construct a supply chain service quality cloud according to the entropy weight of the indicators in the indicator evaluation matrix to judge the supply chain service quality situation.

Images