A new energy vehicle grading recommendation method based on incomplete preferences of decision makers
By constructing adaptive classification boundaries and intelligent data completion methods, the problems of incomplete preferences of decision-makers and incomplete data in the classification recommendation of new energy vehicles are solved, and more accurate and reliable classification recommendations are achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HEBEI UNIV OF TECH
- Filing Date
- 2026-03-04
- Publication Date
- 2026-06-05
Smart Images
Figure CN122153531A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent decision-making and data processing technology, and in particular to a method for classifying and recommending new energy vehicles based on the incomplete preferences of decision-makers. Background Technology
[0002] The new energy vehicle industry has entered a period of rapid development driven by both policy and market forces. Faced with an explosive increase in the number of models on the market and an increasingly complex system of technical parameters, consumers encounter significant information overload and difficulty in making choices when purchasing a vehicle. Essentially, the classification and recommendation of new energy vehicles is a typical multi-criteria classification problem, that is, classifying candidate models into predefined ordered categories based on a series of conflicting evaluation criteria.
[0003] Existing technologies for handling such problems primarily employ ranking methods based on utility functions or classification methods based on preference relationships. While these methods are relatively mature in theoretical research, they still reveal significant limitations in theoretical assumptions and insufficient technological adaptability when applied to the complex real-world scenarios of new energy vehicle recommendations. Existing methods typically rely entirely on expert subjective experience when determining category boundaries, ignoring the statistical characteristics of objective data and lacking adaptability to changes in market data, easily leading to distorted ranking results. Most algorithms assume that decision-makers can provide complete preference information (such as a complete pairwise comparison judgment matrix); however, in actual decision-making, due to the highly specialized nature of indicators, decision-makers often only provide partial preference relationships, and existing technologies lack effective mechanisms for completion and reasoning. Evaluation data for new energy vehicles often contains non-random missing information and comes from heterogeneous sources (such as technical parameters and word-of-mouth reviews). Existing methods often use simple statistics (such as the mean) to fill in the gaps, ignoring the high-dimensional correlations between attributes and the differences in the completeness of information across different samples, resulting in low reliability of the completion results and affecting recommendation accuracy.
[0004] Therefore, proposing a graded recommendation method for new energy vehicles based on the incomplete preferences of decision-makers to address the difficulties of existing technologies is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention
[0005] In view of this, the present invention provides a new energy vehicle classification recommendation method based on the incomplete preferences of decision-makers, which can achieve accurate classification recommendation of new energy vehicles by constructing adaptive classification boundaries, reconstructing incomplete preferences, and intelligently filling in missing data.
[0006] To achieve the above objectives, the present invention adopts the following technical solution: A new energy vehicle tiered recommendation method based on decision-makers' imperfect preferences includes: S1. For each evaluation criterion of new energy vehicles, construct knowledge- and data-driven classification boundaries for new energy vehicles; S2. Determine the weight of each evaluation criterion and the local priority value of the new energy vehicle scheme under the condition of imperfect preference of decision-makers; S3. Identify the attribute polarity of each evaluation criterion for new energy vehicles and construct a standardized evaluation information matrix for new energy vehicles; S4. Fill in the missing values in the standardized evaluation information matrix of new energy vehicles to obtain a complete evaluation information matrix; S5. Aggregate the weight of each evaluation criterion with the complete evaluation information matrix to calculate the global priority value of each new energy vehicle and its classification boundary. S6. Compare the global priority values of each new energy vehicle and its classification boundary, and classify the new energy vehicles into the corresponding recommendation level.
[0007] Optionally, in S1 of the above method, a knowledge- and data-driven classification boundary for new energy vehicles is constructed, specifically as follows: S101. Obtain initial classification boundary data, targeting... Evaluation criteria and There are ordered categories, among which , K Obtain them separately: Subjective classification boundaries : Experts provide advice based on experience for the first j The first criterion, the first The category and the first The dividing points of each category, among which K ; Objective classification boundaries : Determined based on quantiles of new energy vehicle parameter data for the th j The first criterion, the first The category and the first The dividing points of each category, among which K ; S102, Statistical Evaluation Criteria j The data completeness is obtained by calculating the coverage ratio of the effective data, and the weights of the data feature boundaries are calculated using a non-linearly growing S-shaped function.
[0008] in, , is a sensitivity coefficient used to adjust the rate at which data completeness affects the weights. Data completeness is calculated as the percentage of the total number of solutions with evaluation data. S103. Perform linear weighted fusion of the subjective and objective classification boundaries to obtain the final comprehensive classification boundary value: .
[0009] Optionally, in the above method, in S2, the weight and local priority value of each evaluation criterion under the decision-maker's imperfect preferences are determined as follows: S201. Obtaining guidelines provided by decision-makers i and j Incomplete pairwise comparison judgment matrix ,in For paired comparison preferences, if the preference is unknown, it is denoted as... If it is known, denoted as This is used to construct a convex optimization model:
[0010] Solve the model to obtain the initial criterion weights. ; S202. Calculate all unknown elements based on the absolute values before and after the change in criterion weights. Sensitivity to weights is assessed, and the top D elements with the highest sensitivity are selected as core missing values, scaled on the AHP scale. The system iterates through the missing values, randomly fills in the remaining non-core missing values, performs consistency checks, and retains all values that meet the consistency ratio. The filled matrix is used as the candidate judgment matrix set. ; S203. Based on the average local consistency deviation of all closed triangular loops of the judgment matrix, the deviation between the largest eigenvalue of the judgment matrix and the dimension of the judgment matrix, and the deviation between the existing judgment matrix and the ideal judgment matrix, calculate the weighted average of the three and select the judgment matrix corresponding to the minimum value as the optimal judgment matrix. S204. Using the AHP eigenvector method, calculate the eigenvectors of the optimal judgment matrix to obtain the criterion weights. ; S205. The comparison matrix of new energy vehicle schemes and classification boundaries is completed using the Top D filling method in step S202, and the linear interpolation method is used to calculate the new energy vehicle schemes. Local priority value .
[0011] Optionally, in the above method, S3 involves identifying the attribute polarity of each evaluation criterion for new energy vehicles and constructing a standardized evaluation information matrix for new energy vehicles, specifically as follows: Attribute polarity includes: benefit-oriented and cost-oriented; The range transformation method is used to map the original data to a unified interval, and a standardized evaluation information matrix is constructed.
[0012] Optionally, in step S4 of the above method, missing values in the standardized evaluation information matrix for new energy vehicles are filled in to obtain a complete evaluation information matrix. Specifically: S401, Solutions for Cases with Missing Values and plan Calculation scheme and Similarity:
[0013] in, For the plan and The number of known criteria The number of criteria for which data is incomplete. and Representation scheme and In the guidelines The standardized value It is the penalty intensity coefficient, representing the degree of influence of the number of missing items; S402, Set similarity threshold For the missing parameters of the target new energy vehicle model, a set of models with similarity higher than the threshold of the missing parameter is searched in the database. The weighted average value of the corresponding parameter values of the models in the set is used as the imputation value to generate a complete new energy vehicle evaluation information matrix.
[0014] Optionally, in S5 of the above method, the weight of each evaluation criterion is aggregated with the complete evaluation information matrix to calculate the global priority value of each new energy vehicle and its classification boundary, specifically as follows: A linear weighted comprehensive evaluation model is adopted, which aggregates the weight of each evaluation criterion with the local evaluation value of the complete evaluation information matrix to calculate the global priority value of each new energy vehicle. The same logic is used to calculate the global priority value of each classification boundary.
[0015] Optionally, in the above method, in S6, the global priority values of each new energy vehicle and its classification boundary are compared to determine the corresponding recommendation level for each new energy vehicle. Specifically: The global priority value of the new energy vehicle to be evaluated is compared with the global value of each category boundary. Based on the range in which the value falls, the vehicle model is mapped to the corresponding recommendation level category, and finally a graded recommendation list is output.
[0016] As can be seen from the above technical solutions, compared with the prior art, this invention provides a new energy vehicle classification recommendation method based on the incomplete preferences of decision-makers, which has the following beneficial effects: This invention dynamically integrates the subjective and objective boundaries of experts through data completeness, making the classification standard both interpretable by expert knowledge and adaptable to the distribution of market data, thus improving the robustness and accuracy of the classification results; through Top D sensitivity identification and multi-dimensional consistency optimization technology, it can accurately restore the potential, logically consistent complete preference structure from the limited preference information provided by decision-makers, solving the problem of sparsity of preference information caused by the ambiguity of decision-makers' cognition; by introducing a penalty mechanism related to the number of missing data to calculate similarity, it ensures that the filling process relies more on high-quality "nearest neighbor" samples with complete information and similar values, effectively reducing the filling bias caused by missing or incomplete data, and improving the reliability of subsequent recommendation calculations; This invention provides a complete technical solution from boundary construction, preference processing, data filling to final classification recommendation, systematically solving the core problem of incomplete information in the new energy vehicle recommendation scenario, and has important practical application value. Attached Figure Description
[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.
[0018] Figure 1 The flowchart of a new energy vehicle classification recommendation method based on the incomplete preferences of decision-makers provided by the present invention. Detailed Implementation
[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0020] Reference Figure 1 As shown, this invention discloses a new energy vehicle classification recommendation method based on decision-makers' imperfect preferences, including: S1. For each evaluation criterion of new energy vehicles, construct knowledge- and data-driven classification boundaries for new energy vehicles; S2. Determine the weight of each evaluation criterion and the local priority value of the new energy vehicle scheme under the imperfect preference of the decision-maker; S3. Identify the attribute polarity of each evaluation criterion for new energy vehicles and construct a standardized evaluation information matrix for new energy vehicles; S4. Fill in the missing values in the standardized evaluation information matrix of new energy vehicles to obtain a complete evaluation information matrix; S5. Aggregate the weight of each evaluation criterion with the complete evaluation information matrix to calculate the global priority value of each new energy vehicle and its classification boundary. S6. Compare the global priority values of each new energy vehicle and its classification boundary, and classify the new energy vehicles into the corresponding recommendation level.
[0021] Furthermore, in S1, a knowledge- and data-driven classification boundary for new energy vehicles is constructed to overcome the arbitrariness of subjective boundaries imposed by a single expert and the mechanical nature of objective boundaries imposed by a single data source. An adaptive weighting mechanism is used to integrate both, establishing a robust grading standard, specifically: S101. Obtain initial classification boundary data, targeting... Evaluation criteria and There are ordered categories, among which , K Obtain them separately: Subjective classification boundaries : Experts provide advice based on experience for the first j The first criterion, the first The category and the first The dividing points of each category, among which K ; Objective classification boundaries : Determined based on quantiles of new energy vehicle parameter data for the th j The first criterion, the first The category and the first The dividing points of each category, among which K ; S102. Define the data completeness by statistically analyzing the coverage ratio of effective data under each evaluation criterion, addressing the bias caused by data sparsity, and calculate the data feature boundary weights using a non-linearly growing S-shaped function:
[0022] in, , is a sensitivity coefficient used to adjust the rate at which data completeness affects the weights. For data integrity; S103. Perform linear weighted fusion of the subjective and objective classification boundaries to obtain the final comprehensive classification boundary value, which will serve as the benchmark for subsequent classification determination. .
[0023] Furthermore, in S2, the weight and local priority value of each evaluation criterion under the decision-maker's imperfect preferences are determined, specifically as follows: S201. Obtaining guidelines provided by decision-makers i and j Incomplete pairwise comparison judgment matrix ,in For paired comparison preferences, if the preference is unknown, it is denoted as... If it is known, denoted as Based on this, a convex optimization model is constructed with the objective of minimizing the sum of squared logarithmic biases of known pairwise comparison preferences as follows:
[0024] Solve the model to obtain the initial criterion weights. ; S202. Calculate all unknown elements based on the absolute values before and after the change in criterion weights. Sensitivity to weights is assessed, and the top D elements with the highest sensitivity are selected as core missing values, scaled on the AHP scale. The system iterates through the missing values, randomly fills in the remaining non-core missing values, performs consistency checks, and retains all values that meet the consistency ratio. The filled matrix is used as the candidate judgment matrix set. ; S203. Based on the average local consistency deviation of all closed triangular loops of the judgment matrix, the deviation between the largest eigenvalue of the judgment matrix and the dimension of the judgment matrix, and the deviation between the existing judgment matrix and the ideal judgment matrix, calculate the weighted average of the three and select the judgment matrix corresponding to the minimum value as the optimal judgment matrix. S204. Using the AHP eigenvector method, calculate the eigenvectors of the optimal judgment matrix to obtain the criterion weights. ; S205. The comparison matrix of new energy vehicle schemes and classification boundaries is completed using the Top D filling method in step S202, and the linear interpolation method is used to calculate the new energy vehicle schemes. Local priority value .
[0025] Furthermore, in S3, the attribute polarity of each evaluation criterion for new energy vehicles is identified, and a standardized evaluation information matrix for new energy vehicles is constructed, specifically as follows: Attribute polarity includes: benefit-oriented and cost-oriented; Identify the attribute polarity (benefit-oriented or cost-oriented) of each evaluation criterion, use the range transformation method to map the original data to a unified interval, eliminate the difference in dimensions, and construct a standardized evaluation information matrix.
[0026] Furthermore, in S4, missing values in the standardized evaluation information matrix for new energy vehicles are filled in to obtain a complete evaluation information matrix, specifically as follows: S401. When using the "nearest neighbor" concept to fill in missing values, to avoid being misled by samples with insufficient information, this invention improves the similarity calculation logic for schemes containing missing values. and plan Calculation scheme and Similarity:
[0027] in, For the plan and The number of known criteria The number of criteria for which data is incomplete. and Representation scheme and In the guidelines The standardized value It is the penalty intensity coefficient, representing the degree of influence of the number of missing items; S402, Set similarity threshold For the missing parameters of the target new energy vehicle model, a set of models with similarity higher than the threshold of the missing parameter is searched in the database. The weighted average value of the corresponding parameter values of the models in the set is used as the imputation value to generate a complete new energy vehicle evaluation information matrix.
[0028] Furthermore, in S5, the weight of each evaluation criterion is aggregated with the complete evaluation information matrix to calculate the global priority value for each new energy vehicle and its classification boundary, specifically: A linear weighted comprehensive evaluation model is adopted, which aggregates the weight of each evaluation criterion with the local evaluation value of the complete evaluation information matrix to calculate the global priority value of each new energy vehicle. The same logic is used to calculate the global priority value of each classification boundary.
[0029] Furthermore, in S6, the global priority values of various new energy vehicles and their classification boundaries are compared to categorize new energy vehicles into corresponding recommendation levels, specifically: The global priority value of the new energy vehicle to be evaluated is compared with the global value of each category boundary. Based on the range in which the value falls, the vehicle model is mapped to the corresponding recommendation level category, and finally a graded recommendation list is output.
[0030] In one specific embodiment, a new energy vehicle tiered recommendation method based on decision-makers' imperfect preferences is described: Constructing adaptive classification boundaries: Using factors such as driving range, price, energy consumption per 100 kilometers, and safety rating as evaluation criteria, firstly, experts in the field were invited to provide subjective boundary values (subjective classification boundaries) between each pair of levels for the three categories of "strongly recommended," "generally recommended," and "not recommended." At the same time, historical data of mainstream new energy vehicles in the market were collected, and specific quantiles (such as tertiles) of the data distribution under each criterion were calculated as objective classification boundaries.
[0031] The coverage of valid data under each criterion is defined as the data completeness of that criterion. Set sensitivity coefficient ε >0; According to the formula: Calculate the data feature weights for each criterion. For criteria with high data completeness, A value close to 0.5 indicates that the classification boundary is determined by both subjective and objective boundaries; for the criterion of data sparsity, A value close to 0 indicates that the classification boundary is mainly determined by subjective boundaries.
[0032] According to the formula: The final classification boundary after fusion is calculated.
[0033] Determine the weights of the evaluation criteria under incomplete preferences: For the four evaluation criteria of "driving range", "price", "energy consumption per 100 kilometers" and "safety rating", the decision-makers only provided some pairwise comparison results of importance, forming an incomplete judgment matrix A. For example, it is known that "driving range is slightly more important than price (scale 3)" and "safety is significantly more important than energy consumption (scale 5)", but the comparison relationship between "driving range and energy consumption" and "price and safety" is unknown. First, with the objective of minimizing the sum of squared logarithmic deviations of the known elements, a convex optimization model is constructed, and the initial weight vector is obtained by solving it. .
[0034] Then, for each unknown element Assuming it is on the AHP scale The values are taken sequentially, and the weight vector is recalculated. W' Calculate the sensitivity of the missing element. S = ||W' - W 0 || Select the one with the highest sensitivity D (For example D =2) missing elements; Regarding this DEach core missing element is combined across all its possible scale values. For each combination, the remaining non-core missing elements are randomly filled (randomly selected from their possible scale values) to generate a complete candidate decision matrix. A 1 ,examine A 1 Consistency ratio CR All rights reserved CR A matrix with a value less than 0.1.
[0035] For all retained candidate matrices, calculate their propagation conflict degree, principal eigenvalue deviation degree, and structural deviation degree, and calculate the weighted comprehensive index of the three. Select the candidate matrix with the smallest comprehensive index as the optimal judgment matrix.
[0036] Finally, the eigenvector method is used to calculate the final criterion weights for the optimal judgment matrix. .
[0037] Intelligent missing evaluation data: The dataset of new energy vehicles to be evaluated has missing parameters. First, the data is standardized by range. For a particular vehicle model lacking "energy consumption per 100 kilometers" data... A It is necessary to calculate its comparison with other models in the dataset. B The similarity. Assume A and B With an initial similarity of 0.8 for "electricity consumption per 100 kilometers", the two vehicles share three attributes with each other. =3), there is 1 common missing attribute ( =1), penalty intensity coefficient γ=0.05; then the final similarity:
[0038] Set similarity threshold Find all related to car models A Models with a similarity greater than 0.8 are grouped into a nearest neighbor set, and the arithmetic mean of the "energy consumption per 100 kilometers" values of all models in this set is used to fill in the missing models. A Missing values; Repeat this process for all missing attributes to obtain the complete evaluation information matrix.
[0039] Tiered Recommendations: Using the criterion weights obtained in the second step W The complete evaluation information matrix obtained in the third step is linearly weighted and aggregated to calculate the global priority score for each car. .
[0040] Using the same weights WThe final classification boundary matrix obtained in the first step (each row represents the value of a classification boundary on each criterion) is weighted and aggregated to calculate the priority score of the "strong recommendation - general recommendation" boundary. Priority scores relative to the "generally recommended - not recommended" boundary .
[0041] Compare and , : like If so, it will be classified as "strongly recommended"; like If so, it will be classified as "generally recommended"; like If so, it will be classified as "not recommended"; The final output is a recommended list of new energy vehicles with rating labels.
[0042] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, for system or system embodiments, since they are basically similar to method embodiments, the description is relatively simple, and relevant parts can be referred to the descriptions in the method embodiments. The systems and system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without creative effort.
[0043] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A method for tiered recommendation of new energy vehicles based on decision-makers' incomplete preferences, characterized in that, include: S1. For each evaluation criterion of new energy vehicles, construct knowledge- and data-driven classification boundaries for new energy vehicles; S2. Determine the weight of each evaluation criterion and the local priority value of the new energy vehicle scheme under the condition of imperfect preference of decision-makers; S3. Identify the attribute polarity of each evaluation criterion for new energy vehicles and construct a standardized evaluation information matrix for new energy vehicles; S4. Fill in the missing values in the standardized evaluation information matrix of new energy vehicles to obtain a complete evaluation information matrix; S5. Aggregate the weight of each evaluation criterion with the complete evaluation information matrix to calculate the global priority value of each new energy vehicle and its classification boundary. S6. Compare the global priority values of each new energy vehicle and its classification boundary, and classify the new energy vehicles into the corresponding recommendation level.
2. The method for classifying and recommending new energy vehicles based on the incomplete preferences of decision-makers, as described in claim 1, is characterized in that... In S1, the classification boundaries of new energy vehicles are constructed using knowledge and data-driven methods, specifically as follows: S101. Obtain initial classification boundary data, targeting... Evaluation criteria and There are ordered categories, among which , K Obtain them separately: Subjective classification boundaries : Experts provide advice based on experience for the first j The first criterion, the first The category and the first The dividing points of each category, among which K ; Objective classification boundaries : Determined based on quantiles of new energy vehicle parameter data for the th j The first criterion, the first The category and the first The dividing points of each category, among which K ; S102, Statistical Evaluation Criteria j The data completeness is obtained by calculating the coverage ratio of the effective data, and the weights of the data feature boundaries are calculated using a non-linearly growing S-shaped function. in, , is a sensitivity coefficient used to adjust the rate at which data completeness affects the weights. Data completeness is calculated as the percentage of the total number of solutions with evaluation data. S103. Perform linear weighted fusion of the subjective and objective classification boundaries to obtain the final comprehensive classification boundary value: 。 3. The new energy vehicle classification recommendation method based on decision-maker's incomplete preferences as described in claim 2, characterized in that, In S2, the weight and local priority value of each evaluation criterion under the imperfect preferences of the decision-maker are determined as follows: S201. Obtaining criteria provided by decision-makers i and j Incomplete pairwise comparison judgment matrix ,in For paired comparison preferences, if the preference is unknown, it is denoted as... ; If it is known, denoted as This is used to construct a convex optimization model: Solve the model to obtain the initial criterion weights. ; S202. Calculate all unknown elements based on the absolute values before and after the change in criterion weights. Sensitivity to weights is assessed, and the top D elements with the highest sensitivity are selected as core missing values, scaled on the AHP scale. The system iterates through the missing values, randomly fills in the remaining non-core missing values, performs consistency checks, and retains all values that meet the consistency ratio. The filled matrix is used as the candidate judgment matrix set. ; S203. Based on the average local consistency deviation of all closed triangular loops of the judgment matrix, the deviation between the largest eigenvalue of the judgment matrix and the dimension of the judgment matrix, and the deviation between the existing judgment matrix and the ideal judgment matrix, calculate the weighted average of the three and select the judgment matrix corresponding to the minimum value as the optimal judgment matrix. S204. Using the AHP eigenvector method, calculate the eigenvectors of the optimal judgment matrix to obtain the criterion weights. ; S205. The comparison matrix of new energy vehicle schemes and classification boundaries is completed using the Top D filling method in step S202, and the linear interpolation method is used to calculate the new energy vehicle schemes. Local priority value .
4. The new energy vehicle classification recommendation method based on decision-maker's imperfect preferences as described in claim 3, characterized in that, In S3, the attribute polarity of each evaluation criterion for new energy vehicles is identified, and a standardized evaluation information matrix for new energy vehicles is constructed, specifically as follows: Attribute polarity includes: benefit-oriented and cost-oriented; The range transformation method is used to map the original data to a unified interval, and a standardized evaluation information matrix is constructed.
5. The new energy vehicle classification recommendation method based on decision-maker's incomplete preferences as described in claim 4, characterized in that, In S4, missing values in the standardized evaluation information matrix for new energy vehicles are filled in to obtain a complete evaluation information matrix, specifically as follows: S401, Solutions for Cases with Missing Values and plan Calculation scheme and Similarity: in, For the plan and The number of known criteria The number of criteria for which data is incomplete. and Representation scheme and In the guidelines The standardized value It is the penalty intensity coefficient, representing the degree of influence of the number of missing items; S402, Set similarity threshold For the missing parameters of the target new energy vehicle model, a set of models with similarity higher than the threshold of the missing parameter is searched in the database. The weighted average value of the corresponding parameter values of the models in the set is used as the imputation value to generate a complete new energy vehicle evaluation information matrix.
6. The new energy vehicle classification recommendation method based on decision-maker's incomplete preferences as described in claim 5, characterized in that, In S5, the weight of each evaluation criterion is aggregated with the complete evaluation information matrix to calculate the global priority value of each new energy vehicle and its classification boundary, specifically: A linear weighted comprehensive evaluation model is adopted, which aggregates the weight of each evaluation criterion with the local evaluation value of the complete evaluation information matrix to calculate the global priority value of each new energy vehicle. The same logic is used to calculate the global priority value of each classification boundary.
7. The method for classifying and recommending new energy vehicles based on the incomplete preferences of decision-makers according to claim 6, characterized in that, In S6, the global priority values of various new energy vehicles and their classification boundaries are compared to determine the corresponding recommendation level for each new energy vehicle. Specifically: The global priority value of the new energy vehicle to be evaluated is compared with the global value of each category boundary. Based on the range in which the value falls, the vehicle model is mapped to the corresponding recommendation level category, and finally a graded recommendation list is output.