A combined prediction method for the consumption of repair parts for communication equipment

Abstract

The present invention provides a combined prediction method for the consumption of repair spare parts of communication equipment provided by the present invention. By acquiring the historical consumption data of repair spare parts of communication equipment and determining the main influencing factors of the consumption of repair spare parts, the accuracy of the prediction model is improved; a prediction model with high prediction accuracy in terms of influencing factors and historical consumption data is selected, the prediction accuracy and prediction value at each moment are calculated, and each prediction accuracy and the corresponding prediction value are formed into multiple two-dimensional arrays, so as to construct an IOWA induced ordered series, and a corresponding weight is assigned according to the size of the prediction accuracy of a single prediction model at each moment in a sample interval; an error information matrix is ​​generated, the value of the error information matrix is ​​minimized, and a new combined prediction model is established with the minimum error square sum of the prediction result as a criterion; the method of "completely adopting" or "completely abandoning" a single prediction model is overcome, so that each moment in the service life cycle has a higher prediction accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of communication equipment maintenance and support, and in particular to a combined prediction method for the consumption of repair spare parts for communication equipment. Background Art

[0002] As the use intensity of communication equipment in various fields continues to increase, the failure rate of communication equipment continues to increase. It is particularly important to ensure the stable operation of communication equipment. Therefore, it is necessary to strengthen the maintenance and support of communication equipment to improve the integrity rate of communication equipment. Communication equipment repair spare parts are an important material basis for communication equipment maintenance and support. Scientifically and accurately predicting the consumption of communication equipment repair spare parts is of great significance to improving the economic benefits of communication equipment maintenance and support.

[0003] At present, the traditional communication equipment repair spare parts consumption prediction model mainly uses one of the reliability prediction model, time series prediction model, regression prediction model, neural network prediction model and grey prediction model. However, these models have many unavoidable substantial defects in practical application. Among them, (1) the reliability prediction model requires a large amount of accurate reliability data support, while for general communication equipment products, the reliability index needs to be obtained in a specific experimental environment, which is quite different from the actual use environment, resulting in inaccurate consumption prediction; (2) the time series prediction model requires a sufficient number of samples, while the sample data available for observation in a statistical period is relatively small; (3) the regression prediction model needs to collect data on all influencing factors, and usually selects a few major influencing factors for prediction, which will affect the accuracy of the prediction to a certain extent; (4) the neural network prediction model also relies on historical consumption data, while the historical consumption data collected by communication equipment repair spare parts, the training samples and test samples available for selection are very limited, resulting in insufficient training of the BP neural network, slow convergence of the neural network prediction model, and lack of timeliness; (5) the grey prediction model does not fully explore the influencing factors, and it is difficult to accurately measure them through simple quantitative relationships, resulting in errors in the prediction.

[0004] Therefore, different prediction models have different scopes of application and emphases. No matter which single prediction model is used, it is difficult to fully and accurately describe the relationship between the influencing factors and consumption of communication equipment repair parts. Summary of the invention

[0005] The present invention provides a combined prediction method for the consumption of repair parts for communication equipment, which combines different prediction models, comprehensively utilizes the information provided by various prediction models, and combines predictions in an appropriate manner to obtain the best prediction results, thereby achieving the purpose of improving the prediction accuracy of the consumption of repair parts for communication equipment.

[0006] The present invention provides a combined prediction method for the consumption of repair parts for communication equipment, which is characterized by comprising:

[0007] Obtain historical consumption data of repair parts for communication equipment and determine the main influencing factors of the consumption of the repair parts;

[0008] Based on the historical consumption data and the main influencing factors, multiple prediction models are determined, significance tests are performed on individual prediction models respectively, and prediction accuracies of the selected prediction models that pass the significance tests are calculated respectively;

[0009] For each of the selected prediction models, the prediction accuracy and the prediction value at each moment are calculated, and each prediction accuracy and the corresponding prediction value are formed into multiple two-dimensional arrays, and an IOWA induced ordered sequence is constructed based on the multiple two-dimensional arrays;

[0010] For each of the selected prediction models, the prediction error at each moment is calculated, and an error information matrix is ​​generated; at each moment, the sum of squares of the errors between the predicted value of each selected prediction model and the actual consumption value is calculated to minimize the value of the error information matrix, and the weighted coefficient of each selected prediction model in the combined prediction model is calculated to obtain the expression of the combined prediction model, and the consumption of repair parts for communication equipment is predicted through the expression of the combined prediction model.

[0011] According to a combined prediction method for the consumption of repair spare parts for communication equipment provided by the present invention, determining the main influencing factors of the consumption of repair spare parts includes the following steps:

[0012] Select a reference series for each influencing factor, calculate the gray absolute correlation and gray relative correlation relative to the reference series based on the observed value corresponding to each influencing factor, calculate the gray comprehensive correlation corresponding to each influencing factor, and sort the corresponding influencing factors from large to small based on the value of the gray comprehensive correlation.

[0013] According to a combined prediction method for the consumption of repair spare parts for communication equipment provided by the present invention, forming a plurality of the two-dimensional arrays comprises the following steps:

[0014] Get the prediction accuracy of the i-th selected prediction model at the t-th time, use the prediction accuracy as the induced value of the prediction value, and form m two-dimensional arrays with the prediction accuracy and corresponding prediction values ​​of the m selected prediction models at the t-th time, recorded as 1t , x 1t >, 2t , x 2t >, ...< a it , x it >, …, mt , x mt > ​​​

[0015] Among them, i=1, 2,..., m; t=1, 2,..., N; a it represents the prediction accuracy of the i-th selected prediction model at the t-th time, x it Represents the predicted value of the i-th selected prediction model at the t-th time.

[0016] According to a combined prediction method for the consumption of repair spare parts for communication equipment provided by the present invention, constructing an IOWA induced ordered sequence includes the following steps:

[0017] The multiple induced values ​​a it Sort by the largest to the smallest and by the induced value a it The predicted value x corresponding to the sorted pair it Perform ordered arrangement and combine weighted vectors to generate IOWA induced ordered sequences, including:

[0018]

[0019] Among them, the function f w is a 1 , a 2 , …, a m The generated m-dimensional induced ordered weighted average operator, i.e., IOWA operator; a-index(it) is the prediction accuracy a 1 , a 2 , …, a m The subscript of the ith largest number in the descending order, W = (w 1 , w 2 ,…,w m ) T is the weight vector of IOWA, satisfying

[0020] According to a combined prediction method for the consumption of repair spare parts for communication equipment provided by the present invention, generating an error information matrix comprises the following steps:

[0021] Among the m selected prediction models, the prediction error of the i-th selected prediction model at the tth moment is e it , the corresponding error vector is E i =[e i1 , e i2 ,…,e iN ] T , and the combined prediction error matrix e=[E 1 , E 2 ,…,E m ], then the error information matrix is:

[0022]

[0023] According to a combined prediction method for the consumption of repair spare parts for communication equipment provided by the present invention, the error information matrix E is minimized and the formula is applied:

[0024]

[0025]

[0026] Among them, E ij is the covariance between the i-th selected prediction model and the j-th selected prediction model, E ii is the variance of the i-th selected prediction model; k i is the weight of the i-th selected prediction model.

[0027] According to a combined prediction method for the consumption of repair spare parts for communication equipment provided by the present invention, the formula of the combined prediction model is:

[0028]

[0029] Further,

[0030] in, It represents the average value of the prediction accuracy of communication equipment repair spare parts consumption of the i-th selected prediction model starting from the Nth year in the past T years.

[0031] Optionally, according to a combined forecasting method for the consumption of repair spare parts for communication equipment provided by the present invention, the combined forecasting model of the induced ordered weighted average in the N+T year is:

[0032]

[0033] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of the combined prediction method for the consumption of repair spare parts for communication equipment as described in any one of the above are implemented.

[0034] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the above-mentioned combined prediction methods for the consumption of repair spare parts for communication equipment.

[0035] The present invention provides a combined prediction method for the consumption of repair parts of communication equipment. The method is beneficial to improving the accuracy of the prediction model by acquiring the historical consumption data of the repair parts of communication equipment and determining the main influencing factors of the consumption of the repair parts; based on the historical consumption data and the main influencing factors, a single prediction model with high prediction accuracy in terms of both the influencing factors and the historical consumption data is selected; for each selected prediction model, the prediction accuracy and prediction value at each moment are calculated, and each prediction accuracy and the corresponding prediction value are formed into a plurality of two-dimensional arrays, an IOWA induced ordered series is constructed based on the plurality of two-dimensional arrays, and corresponding weights are assigned according to the size of the prediction accuracy of the single prediction model at each moment in the sample interval; an error information matrix is ​​generated; the value of the error information matrix is ​​minimized, and finally, a new combined prediction model is established with the minimum error square sum of the prediction results as a criterion; the simple approach of "completely adopting" or "completely abandoning" a single prediction model is overcome, and the purpose of high prediction accuracy of the combined prediction model at each moment in the service life cycle is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0037] Figure 1 This is one of the flow charts of the combined prediction method for the consumption of repair spare parts for communication equipment provided by the present invention;

[0038] Figure 2 It is a schematic diagram comparing the prediction data of each model fitting and the actual consumption data of the combined prediction method for the consumption of repair spare parts of communication equipment provided by the present invention;

[0039] Figure 3 It is a schematic diagram comparing the combined model prediction data and the actual consumption data of the combined prediction method for the consumption of repair spare parts for communication equipment provided by the present invention;

[0040] Figure 4 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are 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.

[0042] In one embodiment, Figure 1 As shown, a combined prediction method for the consumption of repair parts for communication equipment includes the steps of:

[0043] Obtain historical consumption data of repair parts for communication equipment and determine the main influencing factors of the consumption of the repair parts;

[0044] Based on the historical consumption data and the main influencing factors, multiple prediction models are determined, significance tests are performed on individual prediction models respectively, and prediction accuracies of the selected prediction models that pass the significance tests are calculated respectively;

[0045] For each of the selected prediction models, the prediction accuracy and the prediction value at each moment are calculated, and each prediction accuracy and the corresponding prediction value are formed into multiple two-dimensional arrays, and an IOWA induced ordered sequence is constructed based on the multiple two-dimensional arrays;

[0046] For each of the selected prediction models, the prediction error at each moment is calculated, and an error information matrix is ​​generated; at each moment, the sum of squares of the errors between the predicted value of each selected prediction model and the actual consumption value is calculated to minimize the value of the error information matrix, and the weighted coefficient of each selected prediction model in the combined prediction model is calculated to obtain the expression of the combined prediction model, and the consumption of repair parts for communication equipment is predicted through the expression of the combined prediction model.

[0047] Further, determining the main influencing factors of the repair spare parts consumption includes the steps of:

[0048] Select a reference series for each influencing factor, calculate the gray absolute correlation and gray relative correlation relative to the reference series based on the observed value corresponding to each influencing factor, calculate the gray comprehensive correlation corresponding to each influencing factor, and sort the corresponding influencing factors from large to small based on the value of the gray comprehensive correlation.

[0049] Specifically, the main influencing factors of communication equipment repair spare parts consumption are determined by using the grey correlation analysis method, and the specific steps are as follows:

[0050] First, select a reference series and perform dimensionless processing on the observed data. Let X i is (i=1,2,…,n) system factor, x0 (k) is the reference sequence, and the remaining x i (k) is the comparison sequence, and the observation data at sequence number k is x i (k)(k=1,2,…,n), then X i = (x i (1), x i (2), …, x i (n)) is the behavioral sequence of factors;

[0051] Further, calculate the grey absolute correlation degree, assuming that the sequence X 0 With X i The lengths are the same, then we call ε 0i For X 0 With X i The grey absolute correlation degree.

[0052]

[0053]

[0054] Furthermore, the grey relative correlation degree is calculated.

[0055] remember For X 0 With X i relative correlation.

[0056] Further, the grey comprehensive correlation degree is calculated:

[0057] Let θ∈[0,1], then ρ 0i =θ·ε 0i +(1-θ)·r 0i For X 0 With X i Grey comprehensive correlation degree;

[0058] Optionally, θ=0.5 can generally be taken. If the relationship between the absolute quantities is relatively close, θ can be taken upward; conversely, if the rate of change has a greater impact on the result, θ can be taken downward.

[0059] As a preferred embodiment, the present invention determines three commonly used models, namely, the grey prediction model, the multivariate regression prediction model, and the BP neural network prediction model, as single prediction models for analysis according to the principle of determining the standard of a single model;

[0060] Specifically, the significance test of the selected prediction model includes:

[0061] For the grey GM (0, N) model, the grey GM (0, N) model verification generally adopts the residual test method to test the prediction result, which is only used as an example of the embodiment of the present invention and should not be regarded as a further limitation of the present invention;

[0062] For the multiple regression model, there are two types of this prediction model: linear and nonlinear. The multiple regression model used in the present invention is a multiple linear regression model. The test of this model includes the following methods:

[0063] ① Goodness of fit: Goodness of fit test statistic determination coefficient r 2 :

[0064]

[0065] r 2 The closer it is to 1, the better the model fit is.

[0066] ②Significance F test: Describe whether the linear relationship of the equation is significant overall. The statistics are as follows:

[0067]

[0068] It obeys the F distribution with degrees of freedom (k, nk-1). Given the significance level α, the value of the statistic F is obtained from the sample, and F>F α (k, nk-1) or F≤F α (k, nk-1) is used to determine whether the overall linear relationship of the original equation is significant.

[0069] ③ Hypothesis t test of partial regression coefficient: given a significance level α, if t i > t α / 2 (k, nk-1), then the corresponding independent variable x i has a significant impact on the dependent variable y; otherwise, it means that the independent variable x i It has no significant effect on the dependent variable y, so the variable is deleted from the equation.

[0070] For the BP neural network model, the communication equipment repair spare parts BP neural network must be constructed by historical consumption data training samples and verified by test samples before it can be used as a communication equipment repair spare parts consumption prediction model to achieve the relationship mapping between the actual historical consumption of communication equipment repair spare parts and the demand forecast consumption.

[0071] Optionally, error analysis of the BP neural network prediction model for communication equipment repair spare parts is performed by bringing the test sample data before the n+mth year into the constructed network model for calculation, predicting the consumption in the n+m+1th year, and finally comparing the predicted consumption with the actual consumption data. If the error accuracy is within the range, the communication equipment repair spare parts neural network prediction model is considered to have passed the test and can be used to predict the consumption of communication equipment repair spare parts. Otherwise, the BP neural network prediction model needs to be rebuilt. The specific process can be completed by simulation using the neural network toolbox of MATLAB7.0.

[0072] Specifically, the calculation of the prediction accuracy of a single selected prediction model includes the following steps:

[0073] Suppose there are m feasible single-item prediction methods, x t is the actual consumption value at time t, x it is the predicted value (or fitted value) of the i-th selected prediction model at time t, let

[0074]

[0075] i=1,2,…,m; t=1,2,…,N;

[0076] Further, forming a plurality of the two-dimensional arrays comprises the steps of:

[0077] Get the prediction accuracy of the i-th selected prediction model at the t-th time, use the prediction accuracy as the induced value of the prediction value, and form m two-dimensional arrays with the prediction accuracy and corresponding prediction values ​​of the m selected prediction models at the t-th time, recorded as 1t , x 1t >, 2t , x 2t >, ...< a it , x it >, …, mt , x mt >

[0078] Among them, i=1, 2,..., m; t=1, 2,..., N; a it represents the prediction accuracy of the i-th selected prediction model at the t-th time, x it Represents the predicted value of the i-th selected prediction model at the t-th time.

[0079] Furthermore, constructing the IOWA induced ordered sequence includes the following steps:

[0080] The multiple induced values ​​a it Sort by the largest to the smallest and by the induced value a it The predicted value x corresponding to the sorted pair it ​​​Perform ordered arrangement and combine weighted vectors to generate IOWA induced ordered sequences, including:

[0081]

[0082] Among them, the function f w is a 1 , a 2 , …, a m The generated m-dimensional induced ordered weighted average operator, i.e., IOWA operator; a-index(it) is the prediction accuracy a 1 , a 2 , …, a m The subscript of the ith largest number in the descending order, W = (w 1 , w 2 ,…,w m ) T is the weight vector of IOWA, satisfying

[0083] Further, generating the error information matrix includes the steps of:

[0084] Among the m selected prediction models, the prediction error of the i-th selected prediction model at the tth moment is e it , the corresponding error vector is E i =[e i1 , e i2 ,…,e iN ] T , and the combined prediction error matrix e=[E 1 , E 2 ,…,E m ], then the error information matrix is:

[0085]

[0086] Furthermore, let the error information matrix E value be minimized and apply the formula:

[0087]

[0088]

[0089] Among them, E ij is the covariance between the i-th selected prediction model and the j-th selected prediction model, E ii is the variance of the i-th selected prediction model; k i is the weight of the i-th selected prediction model;

[0090] Optionally, the model can be solved using the fmincon function in the MATLAB optimization toolbox, and the weight of each individual prediction model in the combined model can be calculated by inputting the command.

[0091] Furthermore, the formula of the combined prediction model is:

[0092]

[0093] Further,

[0094] in, It represents the average value of the prediction accuracy of communication equipment repair spare parts consumption of the i-th selected prediction model starting from the Nth year in the past T years.

[0095] Specifically, in one embodiment, according to the principle of forecast consistency, it can be inferred that the forecast accuracy of each single forecast model for the number of spare parts consumed for maintenance of communication equipment in the future N+T years is consistent with the average forecast accuracy in the past T years. For example, taking 2017 as the forecast starting point, the forecast accuracy of a single model in 2017 is the same as the forecast accuracy of the model in 2016, and 2018 is the same as the average forecast accuracy of 2015 and 2016 two years ago. Therefore, the combined forecast model of induced ordered weighted average in the N+T year is:

[0096]

[0097] Substitute the prediction accuracy of the single model for the predicted year into the above formula to calculate the corresponding annual combination prediction value.

[0098] In a specific embodiment, the present invention takes the original data of communication equipment repair parts consumption of a telecommunications company from 2006 to 2015 as an example, conducts an example analysis on the constructed communication equipment repair parts consumption combined prediction model, calculates the prediction results of the combined prediction model and the single prediction model respectively, and conducts a comparative analysis, including:

[0099] Preferably, in order to test the validity of the prediction effect, according to the prediction effect evaluation principle, the present invention selects 5 indicators for evaluation: sum of square error SSE, mean square error MSE, mean absolute error MAE, mean absolute percentage error MAPE, mean square percentage error MSPE. The prediction results of each model are calculated according to the calculation formula of each indicator. The calculation results are shown in Table 1. Obviously, the fitting accuracy of the combined prediction model of communication equipment repair spare parts consumption is significantly better than that of a single prediction model.

[0100] Table 1 Comparison of prediction performance of prediction models

[0101]

[0102]

[0103] like Figure 2 As shown, according to the comparative analysis of the fitting curves, the combined prediction model provided by the present invention has the highest fitting degree, followed by the neural network model, while the multivariate regression model has the poorest fitting degree, which verifies the weight coefficient K of the consumption of repair parts for communication equipment. 2 =0, K 3 = 0.99659 result;

[0104] It should be noted that this is because neural networks have a higher advantage in fitting nonlinear targets, while multiple regression analysis excludes the influence of some minor factors when building the model, resulting in low fitting accuracy. In addition, high fitting accuracy also verifies the accuracy of data collection and influence, but since the focus of the present invention is consumption prediction, the prediction function of the model is further tested below.

[0105] Furthermore, the prediction accuracy of multiple prediction models is compared, and the consumption of the next one or several years is predicted through the constructed prediction model, and then compared and analyzed with the actual consumption value. Figure 3 As shown in the figure, the combined prediction model is used to calculate: the predicted consumption of spare parts for this type of equipment in 2014 and 2015 is 388 and 408 pieces, while the actual consumption of spare parts for this type of equipment in 2014 and 2015 is 382 and 403 pieces, with an error rate of 1.6% and 1.2% respectively. It can be seen that the model prediction value is close to the actual consumption value, achieving the expected goal. The prediction results verify the scientificity and accuracy of the combined model.

[0106] It should be noted that although the regression prediction model omits the influence of some minor factors, resulting in lower fitting accuracy, it cannot be abandoned. Only by comprehensively considering the main influencing factors of communication equipment repair parts and the trend of historical consumption data changes can the accuracy of repair parts consumption be ensured.

[0107] Figure 4 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 4 As shown, the electronic device may include: a processor 410, a communications interface 420, a memory 430 and a communications bus 440, wherein the processor 410, the communications interface 420 and the memory 430 communicate with each other via the communications bus 440. The processor 410 may call the logic instructions in the memory 430 to execute the above-mentioned combined prediction method for the consumption of repair spare parts of communication equipment, including:

[0108] Obtain historical consumption data of repair parts for communication equipment and determine the main influencing factors of the consumption of the repair parts;

[0109] Based on the historical consumption data and the main influencing factors, multiple prediction models are determined, significance tests are performed on individual prediction models respectively, and prediction accuracies of the selected prediction models that pass the significance tests are calculated respectively;

[0110] For each of the selected prediction models, the prediction accuracy and the prediction value at each moment are calculated, and each prediction accuracy and the corresponding prediction value are formed into multiple two-dimensional arrays, and an IOWA induced ordered sequence is constructed based on the multiple two-dimensional arrays;

[0111] For each of the selected prediction models, the prediction error at each moment is calculated, and an error information matrix is ​​generated; at each moment, the sum of squares of the errors between the predicted value of each selected prediction model and the actual consumption value is calculated to minimize the value of the error information matrix, and the weighted coefficient of each selected prediction model in the combined prediction model is calculated to obtain the expression of the combined prediction model, and the consumption of repair parts for communication equipment is predicted through the expression of the combined prediction model.

[0112] In addition, the logic instructions in the above-mentioned memory 430 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0113] On the other hand, the present invention further provides a computer program product, the computer program product comprising a computer program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions, and when the program instructions are executed by a computer, the computer can execute the combined prediction method for the consumption of repair spare parts of communication equipment provided by the above methods, including:

[0114] Obtain historical consumption data of repair parts for communication equipment and determine the main influencing factors of the consumption of the repair parts;

[0115] Based on the historical consumption data and the main influencing factors, multiple prediction models are determined, significance tests are performed on individual prediction models respectively, and prediction accuracies of the selected prediction models that pass the significance tests are calculated respectively;

[0116] For each of the selected prediction models, the prediction accuracy and the prediction value at each moment are calculated, and each prediction accuracy and the corresponding prediction value are formed into multiple two-dimensional arrays, and an IOWA induced ordered sequence is constructed based on the multiple two-dimensional arrays;

[0117] For each of the selected prediction models, the prediction error at each moment is calculated, and an error information matrix is ​​generated; at each moment, the sum of squares of the errors between the predicted value of each selected prediction model and the actual consumption value is calculated to minimize the value of the error information matrix, and the weighted coefficient of each selected prediction model in the combined prediction model is calculated to obtain the expression of the combined prediction model, and the consumption of repair parts for communication equipment is predicted through the expression of the combined prediction model.

[0118] In another aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is implemented when the computer program is executed by a processor to perform the above-mentioned combined prediction method for the consumption of repair spare parts for communication equipment, including:

[0119] Obtain historical consumption data of repair parts for communication equipment and determine the main influencing factors of the consumption of the repair parts;

[0120] Based on the historical consumption data and the main influencing factors, multiple prediction models are determined, significance tests are performed on individual prediction models respectively, and prediction accuracies of the selected prediction models that pass the significance tests are calculated respectively;

[0121] For each of the selected prediction models, the prediction accuracy and the prediction value at each moment are calculated, and each prediction accuracy and the corresponding prediction value are formed into multiple two-dimensional arrays, and an IOWA induced ordered sequence is constructed based on the multiple two-dimensional arrays;

[0122] For each of the selected prediction models, the prediction error at each moment is calculated, and an error information matrix is ​​generated; at each moment, the sum of squares of the errors between the predicted value of each selected prediction model and the actual consumption value is calculated to minimize the value of the error information matrix, and the weighted coefficient of each selected prediction model in the combined prediction model is calculated to obtain the expression of the combined prediction model, and the consumption of repair parts for communication equipment is predicted through the expression of the combined prediction model.

[0123] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0124] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0125] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A combined prediction method for the consumption of repair parts for communication equipment, characterized in that: include: Obtain historical consumption data of repair parts for communication equipment and determine the main influencing factors of the consumption of the repair parts; Based on the historical consumption data and the main influencing factors, multiple prediction models are determined, significance tests are performed on individual prediction models respectively, and prediction accuracies of the selected prediction models that pass the significance tests are calculated respectively; For each of the selected prediction models, the prediction accuracy and the prediction value at each moment are calculated, and each prediction accuracy and the corresponding prediction value are formed into multiple two-dimensional arrays, and an IOWA induced ordered sequence is constructed based on the multiple two-dimensional arrays; Constructing a plurality of the two-dimensional arrays comprises the steps of: Get the The selected prediction model is The prediction accuracy at the moment is used as the induced value of the prediction value. time The prediction accuracy and corresponding prediction value of the selected prediction model are composed of A two-dimensional array, denoted by ;in, ; Indicates The selected prediction model is The prediction accuracy at the moment, Indicates The selected prediction model is The predicted value at the moment; The construction of IOWA induced ordered sequence includes the following steps: The multiple induction values Sort by largest to smallest and by induced value The predicted value corresponding to the sorted pair Perform ordered arrangement and combine weighted vectors to generate IOWA induced ordered sequences, including: ； Among them, the function yes The resulting Dimensional induced ordered weighted average operator, namely IOWA operator; is the prediction accuracy In order from largest to smallest The subscript of a large number, is the weight vector of IOWA, satisfying ; For each of the selected prediction models, the prediction error at each moment is calculated, and an error information matrix is ​​generated; at each moment, the sum of squares of the errors between the predicted value of each selected prediction model and the actual consumption value is calculated to minimize the value of the error information matrix, and the weighted coefficient of each selected prediction model in the combined prediction model is calculated to obtain the expression of the combined prediction model, and the consumption of repair parts for communication equipment is predicted through the expression of the combined prediction model.

2. A combined prediction method for the consumption of repair parts for communication equipment according to claim 1, characterized in that: Determining the main influencing factors of the repair spare parts consumption includes the steps of: Select a reference series for each influencing factor, calculate the gray absolute correlation and gray relative correlation relative to the reference series based on the observed value corresponding to each influencing factor, calculate the gray comprehensive correlation corresponding to each influencing factor, and sort the corresponding influencing factors from large to small based on the value of the gray comprehensive correlation.

3. A combined prediction method for the consumption of repair parts for communication equipment according to claim 1, characterized in that: Generating the error information matrix includes the following steps: exist Among the selected prediction models, The selected prediction model is The prediction error at time , the corresponding error vector is , and the combined prediction error matrix is , then the error information matrix is: 。 4. A combined prediction method for the consumption of repair parts for communication equipment according to claim 3, characterized in that: Let the error information matrix The value is the smallest, and the formula is applied: ； in, For the The selected prediction model and The covariance of the selected prediction model, For the The variance of the selected prediction model; For the The weights of the selected prediction model.

5. A combined prediction method for the consumption of repair parts for communication equipment according to claim 4, characterized in that: The formula of the combined prediction model is: Further, ; in, Indicates The selected prediction model is Starting from the year, in the past The average value of the forecast accuracy of communication equipment repair spare parts consumption during the year.

6. A combined prediction method for the consumption of repair parts for communication equipment according to claim 5, characterized in that: No. The combined forecasting model of the induced ordered weighted average of the year is: 。 7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the steps of the combined prediction method for the consumption of repair parts for communication equipment according to any one of claims 1 to 6 are implemented.

8. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the combined prediction method for the consumption of repair parts for communication equipment according to any one of claims 1 to 6 are implemented.

Images