Cross-technology route collaborative decision method and system for new energy vehicle power system

Abstract

The application provides a cross-technology route collaborative decision method and system for a new energy vehicle power system, the method comprising: acquiring multi-source heterogeneous data of a new energy vehicle, preprocessing the multi-source heterogeneous data to obtain initial processed data; performing cross-domain federal fusion on the initial processed data to obtain fused data; performing data quality evaluation and data adjustment on the fused data to obtain adjusted data; performing matrix mapping and mixed model construction on the adjusted data to obtain target data; and performing reinforcement learning decision based on the target data to output a decision scheme. The application can break through the multi-modal data barrier, solve the cross-system fusion problem of R&D data / competitive product data / user feedback, realize adaptive decision of a technology route, and realize a unified optimization framework of a pure electric / mixed drive / fuel cell system.

Description

Technical Field

[0001] This invention belongs to the technical field of data analysis and decision-making, specifically relating to a cross-technology route collaborative decision-making method and system for new energy vehicle power systems. Background Technology

[0002] In recent years, the technology of new energy vehicle power systems has been rapidly evolving towards multiple parallel routes (pure electric, hybrid, fuel cell) and full-domain data-driven development, but the industry still faces three core challenges that need to be overcome.

[0003] At the data fusion level, the efficient utilization of multi-source heterogeneous data remains a bottleneck. Currently, within the industry, experimental data generated during R&D, component information in product BOMs (Bill of Materials), and user feedback data are distributed across different systems with significant format differences, resulting in an overall utilization rate of less than 40% (latest research from IEEE VT 2023). More critically, a standardized system for heterogeneous data has not yet been established—for example, motor efficiency data is presented as a "percentage" in R&D systems, but in user feedback it is often described as "power response speed." This semantic and format discrepancy severely restricts the extraction of data value.

[0004] In terms of analytical frameworks, current technologies still operate in a fragmented manner when evaluating different powertrain routes. Pure electric routes focus on battery energy density and electric drive efficiency, hybrid routes focus on the coupling strategy between the engine and motor, and fuel cell routes focus on hydrogen consumption and stack life. Each of these is modeled independently. This fragmented framework leads to a persistently high error rate when comparing different routes, exceeding 25% (SAE 2024 industry report). For example, when evaluating "comprehensive energy consumption per 100km," pure electric vehicles use "kWh" as the unit, while hybrid vehicles use "L," making it difficult for traditional models to achieve accurate conversions and cross-route comparisons.

[0005] At the decision-making and response level, the lag in traditional models can no longer adapt to rapid market changes. Matching powertrain parameters requires 6-8 rounds of physical testing, each round involving complex processes such as vehicle construction and operational condition simulation. This results in a market response cycle for a new model from parameter design to finalization typically exceeding 14 months (data from the China Association of Automobile Manufacturers' 2023 White Paper). This slow pace not only makes it difficult to keep up with policy adjustments (such as the dual-credit policy and the reduction of subsidies for new energy vehicles), but also fails to quickly meet the dynamic changes in user demands for power performance, driving range, and other aspects.

[0006] In summary, existing technologies suffer from issues such as fragmented multi-source data, lack of cross-route adaptation, and delayed dynamic response. Summary of the Invention

[0007] To address the aforementioned technical problems, this invention provides a cross-technology route collaborative decision-making method and system for new energy vehicle power systems, which is used to solve the technical problems in the prior art.

[0008] In a first aspect, the present invention provides the following technical solution: a cross-technology route collaborative decision-making method for a new energy vehicle power system, comprising:

[0009] Acquire multi-source heterogeneous data of new energy vehicles, and preprocess the multi-source heterogeneous data to obtain initial processed data;

[0010] The initial processed data is then subjected to cross-domain federated fusion to obtain fused data;

[0011] The fused data is then subjected to data quality assessment and data adjustment to obtain adjusted data;

[0012] The adjusted data is subjected to matrix mapping and a hybrid model is constructed to obtain the target data;

[0013] Reinforcement learning decisions are made based on the target data to output a decision scheme.

[0014] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention first acquires multi-source heterogeneous data of new energy vehicles, preprocesses the multi-source heterogeneous data to obtain initial processed data; then, it performs cross-domain federated fusion on the initial processed data to obtain fused data; then, it performs data quality assessment and data adjustment on the fused data to obtain adjusted data; then, it performs matrix mapping and hybrid model construction on the adjusted data to obtain target data; finally, it performs reinforcement learning decision-making based on the target data to output a decision scheme. This invention can break down multimodal data barriers, solve the cross-system fusion problem of R&D data / competitor data / user feedback, and simultaneously achieve adaptive decision-making of technical routes, realizing a unified optimization framework for pure electric / hybrid / fuel cell systems.

[0015] Preferably, the step of performing cross-domain federated fusion of the initial processed data to obtain fused data includes:

[0016] The data type of the initial processing data is identified. If the initial processing data is image data, a first preset model is used to extract features from the initial processing image. If the initial processing data is text data, a second preset model is used to extract features from the initial processing data. If the initial processing data is numerical data, a third preset model is used to extract features from the initial processing data, so as to obtain extracted features.

[0017] The extracted features are homomorphically encrypted to obtain encrypted data. Then, a Laplace mechanism is used to perform differential privacy protection on the encrypted data to obtain processed features. The noise level of the differential privacy protection is... :

[0018] ;

[0019] In the formula, A collection of encrypted data, for The collection after deleting one record;

[0020] A target iterative algorithm is used to minimize the target distance and calculate the optimal transfer matrix. Based on the optimal transfer matrix, the processed features are mapped to a unified space to obtain fused data, wherein the target distance is... :

[0021] ;

[0022] In the formula, These are the first marginal distribution and the second marginal distribution, respectively. To remove the indeterminate boundary, Indicates the joint distribution Samples in the following processing features Seeking expectations, Indicates all that satisfy the marginal distribution as joint distribution A set of.

[0023] Preferably, the step of performing data quality assessment and data adjustment on the fused data to obtain adjusted data includes:

[0024] Calculate the timeliness score of the fused data :

[0025] ;

[0026] In the formula, Half-life, The age of the data to be merged;

[0027] Calculate the integrity score of the fused data :

[0028] ;

[0029] In the formula, These represent the number of missing fields, the number of validated fields, and the total number of fields in the merged data, respectively.

[0030] Data reliability is calculated based on the timeliness score and the completeness score. :

[0031] ;

[0032] In the formula, To determine the authority coefficient of the data sources;

[0033] Data with a credibility level lower than the first threshold is removed. When the credibility level of the fused data is not lower than the first threshold and the integrity score is lower than the second threshold, the corresponding fused data is cleaned and missing values ​​are filled. When the credibility level of the fused data is not lower than the first threshold and the integrity score is lower than the third threshold, the corresponding fused data is downgraded to historical reference data to obtain adjusted data.

[0034] Preferably, the step of performing matrix mapping and hybrid model construction on the adjusted data to obtain the target data includes:

[0035] The core parameters of each technical route are extracted from the adjusted data, and the core parameters are then processed to obtain dimensionless data. :

[0036] ;

[0037] In the formula, For the first Correction factors for each technical route;

[0038] Based on the dimensionless data Constructing a technology-independent transformation matrix :

[0039] ;

[0040] In the formula, For the first The change of a dimensionless parameter For the first The amount of change in the original data for each technical route;

[0041] Elements are extracted from the technology route-independent transformation matrix to obtain technical data. Based on the technical data, a physical model based on first principles and a data model based on deep learning algorithms are constructed.

[0042] The physical model and the data model are weighted and fused to obtain the target data.

[0043] Preferably, the step of performing reinforcement learning decision-making based on the target data to output a decision scheme includes:

[0044] Construct a continuous action space and a discrete action space based on the target data;

[0045] A reward function is constructed based on the reinforcement learning mechanism and the continuous action space and the discrete action space. :

[0046] ;

[0047] ;

[0048] ;

[0049] ;

[0050] In the formula, For efficiency weighting, As a cost weight, For patent weighting, These are efficiency gain, cost gain, and patent risk value, respectively. These are the efficiency values ​​for the new scheme and the original scheme, respectively. These are the cost values ​​for the new plan and the original plan, respectively. These represent the total number of patents related to the new solution and the number of patents that conflict with the new solution, respectively. The similarity score between the new solution and existing patents is calculated.

[0051] The PPO algorithm is used to update the policy and maximize the reward function to obtain the decision scheme.

[0052] Secondly, the present invention provides the following technical solution: a cross-technology route collaborative decision-making system for a new energy vehicle power system, the system comprising:

[0053] The preprocessing module is used to acquire multi-source heterogeneous data of new energy vehicles and preprocess the multi-source heterogeneous data to obtain initial processed data.

[0054] The fusion module is used to perform cross-domain federated fusion on the initial processed data to obtain fused data;

[0055] An adjustment module is used to perform data quality assessment and data adjustment on the fused data to obtain adjusted data;

[0056] A construction module is used to perform matrix mapping and hybrid model construction on the adjusted data to obtain the target data;

[0057] The decision module is used to perform reinforcement learning decisions based on the target data to output a decision scheme.

[0058] Preferably, the adjustment module includes:

[0059] The identification submodule is used to identify the data type of the initial processing data. If the initial processing data is image data, a first preset model is used to extract features from the initial processing image. If the initial processing data is text data, a second preset model is used to extract features from the initial processing data. If the initial processing data is numerical data, a third preset model is used to extract features from the initial processing data to obtain extracted features.

[0060] The encryption submodule is used to perform homomorphic encryption on the extracted features to obtain encrypted data, and then uses a Laplace mechanism to perform differential privacy protection on the encrypted data to obtain processed features, wherein the noise level of the differential privacy protection is... :

[0061] ;

[0062] In the formula, A collection of encrypted data, for The collection after deleting one record;

[0063] The mapping submodule is used to minimize the target distance using a target iterative algorithm and calculate the optimal transfer matrix. Based on the optimal transfer matrix, the processed features are mapped to a unified space to obtain fused data, wherein the target distance is... :

[0064] ;

[0065] In the formula, These are the first marginal distribution and the second marginal distribution, respectively. To remove the indeterminate boundary, Indicates the joint distribution Samples in the following processing features Seeking expectations, Indicates all that satisfy the marginal distribution as joint distribution A set of.

[0066] Preferably, the adjustment module includes:

[0067] The first calculation submodule is used to calculate the timeliness score of the fused data. :

[0068] ;

[0069] In the formula, Half-life, The age of the data to be merged;

[0070] The second calculation submodule is used to calculate the integrity score of the fused data. :

[0071] ;

[0072] In the formula, These represent the number of missing fields, the number of validated fields, and the total number of fields in the merged data, respectively.

[0073] The third calculation submodule is used to calculate the data credibility based on the timeliness score and the integrity score. :

[0074] ;

[0075] In the formula, To determine the authority coefficient of the data sources;

[0076] The adjustment submodule is used to remove fused data whose data credibility is less than the first threshold. When the data credibility of the fused data is not less than the first threshold and the integrity score is less than the second threshold, the corresponding fused data is cleaned and missing values ​​are filled. When the data credibility of the fused data is not less than the first threshold and the integrity score is less than the third threshold, the corresponding fused data is downgraded to historical reference data to obtain the adjusted data.

[0077] Thirdly, the present invention provides the following technical solution: a computer, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements the cross-technology route collaborative decision-making method for new energy vehicle power systems as described above.

[0078] Fourthly, the present invention provides the following technical solution: a storage medium storing a computer program, wherein when the computer program is executed by a processor, it implements the cross-technology route collaborative decision-making method for new energy vehicle power systems as described above. Attached Figure Description

[0079] To more clearly illustrate the technical solutions in the embodiments of the present invention, 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 some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0080] Figure 1 A flowchart of a cross-technology route collaborative decision-making method for new energy vehicle power systems provided in Embodiment 1 of the present invention;

[0081] Figure 2 This is a structural block diagram of a cross-technology route collaborative decision-making system for a new energy vehicle power system provided in Embodiment 2 of the present invention;

[0082] Figure 3 This is a schematic diagram of the hardware structure of a computer provided for another embodiment of the present invention.

[0083] The embodiments of the present invention will be further described below with reference to the accompanying drawings. Detailed Implementation

[0084] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain embodiments of the present invention, and should not be construed as limiting the present invention.

[0085] Example 1

[0086] In Embodiment 1 of the present invention, as Figure 1 As shown, a cross-technology route collaborative decision-making method for new energy vehicle power systems includes:

[0087] S1. Obtain multi-source heterogeneous data of new energy vehicles, and preprocess the multi-source heterogeneous data to obtain initial processed data;

[0088] Specifically, the multi-source heterogeneous data here includes vehicle terminal data, supply chain data, patent knowledge data, and user feedback data. Vehicle terminal data is collected in real time via the CAN bus protocol to collect vehicle operation data (motor speed, battery SOC value, etc.) at a sampling frequency of ≥100Hz. Supply chain data is obtained by connecting to the supplier's BOM system API to extract component costs, delivery cycles, and quality indicators. Patent knowledge data is obtained by accessing a global patent database and crawling technical specifications and claims. User feedback data is obtained by crawling structured comments from automotive forums and analyzing keyword frequency and sentiment. The data preprocessing here includes steps such as data missing identification, missing value imputation, and anomaly identification in existing technologies.

[0089] S2. Perform cross-domain federated fusion on the initial processed data to obtain fused data;

[0090] Step S2 includes:

[0091] S21. Identify the data type of the initial processing data. If the initial processing data is image data, then use a first preset model to extract features from the initial processing image. If the initial processing data is text data, then use a second preset model to extract features from the initial processing data. If the initial processing data is numerical data, then use a third preset model to extract features from the initial processing data, so as to obtain extracted features.

[0092] Specifically, for image data, the MobileNet-V3 model is used, with an input size of 224×224 and an output of 256-dimensional feature vectors. For text data, the Bi-LSTM model is used, with a word vector dimension of 300 and a hidden layer dimension of 512. For numerical data, an autoencoder is used, with an encoding dimension of 128.

[0093] S22. The extracted features are homomorphically encrypted to obtain encrypted data. The encrypted data is then subjected to differential privacy protection using a Laplace mechanism to obtain processed features. The noise level of the differential privacy protection is... :

[0094] ;

[0095] In the formula, A collection of encrypted data, for The collection after deleting one record;

[0096] Specifically, after all features are extracted, the extracted features are obtained and then encrypted using the Paillier homomorphic encryption algorithm with a key length of 2048 bits. Differential privacy protection uses the Laplace mechanism, and the noise level is adaptively adjusted according to the feature sensitivity. At the same time, the privacy pre-selection here is controlled between 0.3 and 0.7 to ensure compliance with GDPR and cybersecurity laws.

[0097] S23. A target iterative algorithm is used to minimize the target distance and calculate the optimal transmission matrix. Based on the optimal transmission matrix, the processed features are mapped to a unified space to obtain fused data, wherein the target distance is... :

[0098] ;

[0099] In the formula, These are the first marginal distribution and the second marginal distribution, respectively. To remove the indeterminate boundary, Indicates the joint distribution Samples in the following processing features Seeking expectations, Indicates all that satisfy the marginal distribution as joint distribution A set;

[0100] Specifically, the Sinkhorn iterative algorithm is then used to calculate the optimal transfer matrix and minimize the target distance. The number of iterations is 100-500, and the alignment error is controlled within 0.05. After verification, the data utilization rate is increased from 42% of the traditional method to 89%, the success rate of member inference attacks is reduced to 2.3%, the time taken for a single round of federated learning is reduced from 5.2 minutes to 1.8 minutes, and the feature extraction F1-score reaches 0.91.

[0101] S3. Perform data quality assessment and data adjustment on the fused data to obtain adjusted data;

[0102] Step S3 includes:

[0103] S31. Calculate the timeliness score of the fused data. :

[0104] ;

[0105] In the formula, Half-life, The age of the data to be merged.

[0106] S32. Calculate the integrity score of the fused data. :

[0107] ;

[0108] In the formula, These represent the number of missing fields, the number of verified fields, and the total number of fields in the merged data, respectively.

[0109] S33. Calculate data reliability based on the timeliness score and the integrity score. :

[0110] ;

[0111] In the formula, To determine the authority coefficient of the data sources;

[0112] Specifically, the authoritative source of data here is determined based on the type of data source, including data from national-level testing institutions. =1.0 (CNCA / CNAS certification number required); Automotive R&D data =1.2 (Requires ISO 26262 certification, ASIL level ≥ C); Supplier data =1.5 (must comply with IATF 16949 quality system, PPAP documentation complete); User feedback data =2.0 (requires triple verification: cross-platform verification, sentiment analysis, and expert review).

[0113] S34. Remove fused data whose data credibility is less than the first threshold. When the data credibility of the fused data is not less than the first threshold and the integrity score is less than the second threshold, perform data cleaning and missing value filling on the corresponding fused data. When the data credibility of the fused data is not less than the first threshold and the integrity score is less than the third threshold, the corresponding fused data is downgraded to historical reference data to obtain adjusted data.

[0114] Specifically, the first threshold, the second threshold, and the third threshold here are 0.3, 0.4, and 0.2, respectively.

[0115] S4. Perform matrix mapping and hybrid model construction on the adjusted data to obtain the target data;

[0116] Specifically, step S4 includes:

[0117] S41. Extract the core parameters of each technical route from the adjusted data, and perform dimensionless processing on the core parameters to obtain dimensionless data. :

[0118] ;

[0119] In the formula, For the first Correction factors for each technical route;

[0120] Specifically, the technical routes here refer to pure electric, hybrid, and fuel cell routes. The core parameters here include pure electric vehicles, such as electric drive efficiency, battery capacity, and energy consumption per 100 kilometers; hybrid vehicles, such as engine power, motor power, and fuel consumption while maintaining power; and fuel cells, such as stack power density, hydrogen consumption rate, and system efficiency.

[0121] S42, Based on the dimensionless data Constructing a technology-independent transformation matrix :

[0122] ;

[0123] In the formula, For the first The change of a dimensionless parameter For the first The amount of change in the original data for each technical route;

[0124] Specifically, by determining the technology route-independent transformation matrix, the performance parameters of different technology routes can be made comparable horizontally, providing a unified input space for multi-objective optimization and supporting the adaptive evaluation and selection of technology routes.

[0125] S43. Extract the elements from the technology route-independent transformation matrix to obtain technical data, and construct a physical model based on first principles and a data model based on deep learning algorithms based on the technical data.

[0126] Specifically, the first principles here include the electromagnetic equations of electric motors and thermodynamic formulas. The physical models constructed have good extrapolation and interpretability. The deep learning algorithms include LSTM / Transformer and other deep learning algorithms, which are good at capturing complex nonlinear relationships.

[0127] S44. The physical model and the data model are weighted and fused to obtain the target data;

[0128] Specifically, the weights of the two models are adjusted according to the working mode, including: High-precision mode (physical model weight 0.85, data model weight 0.15), triggered by tolerance requirement ≤0.5% and sensor confidence ≥95%, suitable for final solution verification and bench testing comparison. Rapid iteration mode (physical model weight 0.25, data model weight 0.75), triggered by response time <5 seconds and parameter space exploration, suitable for conceptual design stage and rapid selection of multiple solutions. Fault early warning mode (physical model weight 0.6, data model weight 0.4), triggered by sensor confidence <90% and residual >3σ, suitable for system health monitoring and early fault warning.

[0129] S5. Based on the target data, perform reinforcement learning decision-making to output a decision scheme;

[0130] Step S5 includes:

[0131] S51. Construct a continuous action space and a discrete action space based on the target data.

[0132] S52. Construct a reward function based on the reinforcement learning mechanism, the continuous action space, and the discrete action space. :

[0133] ;

[0134] ;

[0135] ;

[0136] ;

[0137] In the formula, For efficiency weighting, As a cost weight, For patent weighting, These are efficiency gain, cost gain, and patent risk value, respectively. These are the efficiency values ​​for the new scheme and the original scheme, respectively. These are the cost values ​​for the new plan and the original plan, respectively. These represent the total number of patents related to the new solution and the number of patents that conflict with the new solution, respectively. The similarity score between the new solution and the existing patent is calculated.

[0138] S53. Use the PPO algorithm to update the policy and maximize the reward function to obtain the decision scheme;

[0139] Specifically, the reinforcement learning optimization process employs a proximal policy optimization (PPO)-based algorithm to construct a four-stage decision-making closed loop. In the state awareness stage, the digital twin loads the current powertrain parameters, including motor parameters, battery parameters, and vehicle parameters. In the action decision stage, the agent outputs parameter adjustment schemes; the continuous action space may include motor position and reduction ratio, while the discrete action space includes configuration selection.

[0140] The reward calculation adopts a dynamic multi-objective reward function, which comprehensively considers efficiency gains, costs (including material costs, manufacturing costs and patent licensing costs) and patent risk values.

[0141] The dynamic weight adjustment mechanism adjusts according to changes in external signals: when policy standards become stricter, Increase by 0.1, Increase by 0.1; when cost pressures increase, Increase by 0.2; during technological breakthroughs, Reduce by 0.15; during patent litigation warnings, Increase by 0.3. The PPO algorithm is used in the policy update phase, with learning rates set to 0.0003 (actor) and 0.001 (critic), discount factor γ=0.99, and GAE parameter λ=0.95.

[0142] The cross-technology route collaborative decision-making method for new energy vehicle power systems provided in Embodiment 1 of this invention first acquires multi-source heterogeneous data of new energy vehicles, preprocesses the multi-source heterogeneous data to obtain initial processed data, then performs cross-domain federated fusion on the initial processed data to obtain fused data, then performs data quality assessment and data adjustment on the fused data to obtain adjusted data, then performs matrix mapping and hybrid model construction on the adjusted data to obtain target data, and finally performs reinforcement learning decision-making based on the target data to output a decision scheme. This invention can break down multimodal data barriers, solve the cross-system fusion problem of R&D data / competitor data / user feedback, and simultaneously achieve adaptive decision-making for technology routes, realizing a unified optimization framework for pure electric / hybrid / fuel cell systems.

[0143] Example 2

[0144] like Figure 2 As shown, in Embodiment 2 of the present invention, a cross-technology route collaborative decision-making system for a new energy vehicle powertrain is provided, the system comprising:

[0145] Preprocessing module 1 is used to acquire multi-source heterogeneous data of new energy vehicles and preprocess the multi-source heterogeneous data to obtain initial processed data.

[0146] Fusion module 2 is used to perform cross-domain federated fusion on the initial processed data to obtain fused data;

[0147] Adjustment module 3 is used to perform data quality assessment and data adjustment on the fused data to obtain adjusted data;

[0148] Module 4 is used to perform matrix mapping and hybrid model construction on the adjusted data to obtain the target data;

[0149] Decision module 5 is used to make reinforcement learning decisions based on the target data in order to output a decision scheme.

[0150] The adjustment module 2 includes:

[0151] The identification submodule is used to identify the data type of the initial processing data. If the initial processing data is image data, a first preset model is used to extract features from the initial processing image. If the initial processing data is text data, a second preset model is used to extract features from the initial processing data. If the initial processing data is numerical data, a third preset model is used to extract features from the initial processing data to obtain extracted features.

[0152] The encryption submodule is used to perform homomorphic encryption on the extracted features to obtain encrypted data, and then uses a Laplace mechanism to perform differential privacy protection on the encrypted data to obtain processed features, wherein the noise level of the differential privacy protection is... :

[0153] ;

[0154] In the formula, A collection of encrypted data, for The collection after deleting one record;

[0155] The mapping submodule is used to minimize the target distance using a target iterative algorithm and calculate the optimal transfer matrix. Based on the optimal transfer matrix, the processed features are mapped to a unified space to obtain fused data, wherein the target distance is... :

[0156] ;

[0157] In the formula, These are the first marginal distribution and the second marginal distribution, respectively. To remove the indeterminate boundary, Indicates the joint distribution Samples in the following processing features Seeking expectations, Indicates all that satisfy the marginal distribution as joint distribution A set of.

[0158] The adjustment module 3 includes:

[0159] The first calculation submodule is used to calculate the timeliness score of the fused data. :

[0160] ;

[0161] In the formula, Half-life, The age of the data to be merged;

[0162] The second calculation submodule is used to calculate the integrity score of the fused data. :

[0163] ;

[0164] In the formula, These represent the number of missing fields, the number of validated fields, and the total number of fields in the merged data, respectively.

[0165] The third calculation submodule is used to calculate the data credibility based on the timeliness score and the integrity score. :

[0166] ;

[0167] In the formula, To determine the authority coefficient of the data sources;

[0168] The adjustment submodule is used to remove fused data whose data credibility is less than the first threshold. When the data credibility of the fused data is not less than the first threshold and the integrity score is less than the second threshold, the corresponding fused data is cleaned and missing values ​​are filled. When the data credibility of the fused data is not less than the first threshold and the integrity score is less than the third threshold, the corresponding fused data is downgraded to historical reference data to obtain the adjusted data.

[0169] The construction module 4 includes:

[0170] The dimension processing submodule is used to extract the core parameters of each technical route from the adjustment data, and to perform dimensionless processing on the core parameters to obtain dimensionless data. :

[0171] ;

[0172] In the formula, For the first Correction factors for each technical route;

[0173] The matrix submodule is used to base the dimensionless data. Constructing a technology-independent transformation matrix :

[0174] ;

[0175] In the formula, For the first The change of a dimensionless parameter For the first The amount of change in the original data for each technical route;

[0176] The model submodule is used to extract elements from the technology route-independent transformation matrix to obtain technical data, and to construct a physical model based on first principles and a data model based on deep learning algorithms based on the technical data.

[0177] The weighting submodule is used to perform weighted fusion of the physical model and the data model to obtain the target data.

[0178] The decision module 5 includes:

[0179] The spatial submodule is used to construct a continuous action space and a discrete action space based on the target data;

[0180] The function submodule is used to construct a reward function based on the reinforcement learning mechanism, the continuous action space, and the discrete action space. :

[0181] ;

[0182] ;

[0183] ;

[0184] ;

[0185] In the formula, For efficiency weighting, As a cost weight, For patent weighting, These are efficiency gain, cost gain, and patent risk value, respectively. These are the efficiency values ​​for the new scheme and the original scheme, respectively. These are the cost values ​​for the new plan and the original plan, respectively. These represent the total number of patents related to the new solution and the number of patents that conflict with the new solution, respectively. The similarity score between the new solution and existing patents is calculated.

[0186] The reward submodule is used to update the policy using the PPO algorithm and maximize the reward function to obtain a decision scheme.

[0187] In other embodiments of the present invention, the present invention provides the following technical solution: a computer, including a memory 102, a processor 101, and a computer program stored in the memory 102 and executable on the processor 101, wherein the processor 101 executes the computer program to implement the cross-technology route collaborative decision-making method for new energy vehicle power systems as described above.

[0188] Specifically, the processor 101 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of the present invention.

[0189] The memory 102 may include a large-capacity memory for data or instructions. For example, and not limitingly, the memory 102 may include a hard disk drive (HDD), a floppy disk drive, a solid-state drive (SSD), flash memory, an optical disk drive, a magneto-optical disk drive, magnetic tape, or a Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, the memory 102 may include removable or non-removable (or fixed) media. Where appropriate, the memory 102 may be internal or external to a data processing device. In a particular embodiment, the memory 102 is non-volatile memory. In a particular embodiment, the memory 102 includes read-only memory (ROM) and random access memory (RAM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable read-only memory (PROM), an erasable read-only memory (EPROM), an electrically erasable read-only memory (EEPROM), an electrically alterable read-only memory (EAROM), or flash memory, or a combination of two or more of these. Where appropriate, the RAM can be Static Random-Access Memory (SRAM) or Dynamic Random-Access Memory (DRAM). DRAM can be Fast Page Mode Dynamic Random Access Memory (FPMDRAM), Extended Data Out Dynamic Random Access Memory (EDODRAM), Synchronous Dynamic Random-Access Memory (SDRAM), etc.

[0190] The memory 102 can be used to store or cache various data files that need to be processed and / or used for communication, as well as possible computer program instructions executed by the processor 101.

[0191] The processor 101 reads and executes the computer program instructions stored in the memory 102 to implement the cross-technology route collaborative decision-making method of the new energy vehicle power system.

[0192] In some embodiments, the computer may further include a communication interface 103 and a bus 100. For example, Figure 3 As shown, the processor 101, memory 102, and communication interface 103 are connected through bus 100 and complete communication with each other.

[0193] The communication interface 103 is used to enable communication between the various modules, devices, units, and / or equipment in the embodiments of the present invention. The communication interface 103 can also enable data communication with other components such as external devices, image / data acquisition devices, databases, external storage, and image / data processing workstations.

[0194] Bus 100 includes hardware, software, or both, that couples components of a computer device together. Bus 100 includes, but is not limited to, at least one of the following: data bus, address bus, control bus, expansion bus, and local bus. For example, and not as a limitation, bus 100 may include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an InfiniBand interconnect, a Low Pin Count (LPC) bus, a memory bus, a Micro Channel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local Bus (VLB) bus, or other suitable buses, or a combination of two or more of these. Where appropriate, bus 100 may include one or more buses. Although specific buses are described and illustrated in the embodiments of the present invention, the present invention is contemplated by any suitable bus or interconnect.

[0195] The computer can execute the cross-technology route collaborative decision-making method of the new energy vehicle power system of the present invention based on the cross-technology route collaborative decision-making system of the new energy vehicle power system, thereby realizing cross-technology route collaborative decision-making of the new energy vehicle power system.

[0196] In some further embodiments of the present invention, in conjunction with the above-described cross-technology route collaborative decision-making method for new energy vehicle power systems, the present invention provides the following technical solution: a storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the above-described cross-technology route collaborative decision-making method for new energy vehicle power systems.

[0197] Those skilled in the art will understand that the logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can mean any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0198] More specific examples of readable media (a non-exhaustive list) include: electrical connections (electronic devices) with one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0199] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0200] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0201] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A cross-technology route collaborative decision-making method for new energy vehicle power systems, characterized in that, include: Acquire multi-source heterogeneous data of new energy vehicles, and preprocess the multi-source heterogeneous data to obtain initial processed data; The initial processed data is then subjected to cross-domain federated fusion to obtain fused data; The fused data is then subjected to data quality assessment and data adjustment to obtain adjusted data; The adjusted data is subjected to matrix mapping and a hybrid model is constructed to obtain the target data; Reinforcement learning decision-making is performed based on the target data to output a decision scheme. In the state perception stage, the digital twin loads the current power system parameters, including motor parameters, battery parameters and vehicle parameters. In the action decision-making stage, the agent outputs a parameter adjustment scheme. The continuous action space includes motor position and reduction ratio, and the discrete action space includes configuration selection. The steps of performing reinforcement learning decision-making based on the target data to output a decision scheme include: Construct a continuous action space and a discrete action space based on the target data; A reward function is constructed based on the reinforcement learning mechanism and the continuous action space and the discrete action space. : ； ； ； ； In the formula, For efficiency weighting, As a cost weight, For patent weighting, These are efficiency gain, cost gain, and patent risk value, respectively. These are the efficiency values ​​for the new scheme and the original scheme, respectively. These are the cost values ​​for the new plan and the original plan, respectively. These represent the total number of patents related to the new solution and the number of patents that conflict with the new solution, respectively. The similarity score between the new solution and existing patents is calculated. The PPO algorithm is used to update the policy and maximize the reward function to obtain the decision scheme.

2. The cross-technology route collaborative decision-making method for new energy vehicle power systems according to claim 1, characterized in that, The step of performing cross-domain federated fusion of the initial processed data to obtain fused data includes: The data type of the initial processing data is identified. If the initial processing data is image data, a first preset model is used to extract features from the initial processing data. If the initial processing data is text data, a second preset model is used to extract features from the initial processing data. If the initial processing data is numerical data, a third preset model is used to extract features from the initial processing data, so as to obtain extracted features. The extracted features are homomorphically encrypted to obtain encrypted data. Then, a Laplace mechanism is used to perform differential privacy protection on the encrypted data to obtain processed features. The noise level of the differential privacy protection is... : ； In the formula, A collection of encrypted data, for The collection after deleting one record; A target iterative algorithm is used to minimize the target distance and calculate the optimal transfer matrix. Based on the optimal transfer matrix, the processed features are mapped to a unified space to obtain fused data, wherein the target distance is... : ； In the formula, These are the first marginal distribution and the second marginal distribution, respectively. To remove the indeterminate boundary, Indicates the joint distribution Samples in the following processing features Seeking expectations, Indicates all that satisfy the marginal distribution as joint distribution A set of.

3. The cross-technology route collaborative decision-making method for new energy vehicle power systems according to claim 1, characterized in that, The steps of performing data quality assessment and data adjustment on the fused data to obtain adjusted data include: Calculate the timeliness score of the fused data : ； In the formula, Half-life, The age of the data to be merged; Calculate the integrity score of the fused data : ； In the formula, These represent the number of missing fields, the number of validated fields, and the total number of fields in the merged data, respectively. Data reliability is calculated based on the timeliness score and the completeness score. : ； In the formula, To determine the authority coefficient of the data sources; Data with a credibility level lower than the first threshold is removed. When the credibility level of the fused data is not lower than the first threshold and the integrity score is lower than the second threshold, the corresponding fused data is cleaned and missing values ​​are filled. When the credibility level of the fused data is not lower than the first threshold and the integrity score is lower than the third threshold, the corresponding fused data is downgraded to historical reference data to obtain adjusted data.

4. The cross-technology route collaborative decision-making method for new energy vehicle power systems according to claim 1, characterized in that, The steps of performing matrix mapping and hybrid model construction on the adjusted data to obtain the target data include: The core parameters of each technical route are extracted from the adjusted data, and the core parameters are then processed to obtain dimensionless data. : ； In the formula, For the first Correction factors for each technical route; Based on the dimensionless data Constructing a technology-independent transformation matrix : ； In the formula, For the first The change of a dimensionless parameter For the first The amount of change in the original data for each technical route; Elements are extracted from the technology route-independent transformation matrix to obtain technical data. Based on the technical data, a physical model based on first principles and a data model based on deep learning algorithms are constructed. The physical model and the data model are weighted and fused to obtain the target data.

5. A cross-technology route collaborative decision-making system for a new energy vehicle powertrain, wherein the system employs the cross-technology route collaborative decision-making method for a new energy vehicle powertrain as described in claim 1, characterized in that, The system includes: The preprocessing module is used to acquire multi-source heterogeneous data of new energy vehicles and preprocess the multi-source heterogeneous data to obtain initial processed data. The fusion module is used to perform cross-domain federated fusion on the initial processed data to obtain fused data; An adjustment module is used to perform data quality assessment and data adjustment on the fused data to obtain adjusted data; A construction module is used to perform matrix mapping and hybrid model construction on the adjusted data to obtain the target data; The decision module is used to perform reinforcement learning decisions based on the target data to output a decision scheme.

6. The cross-technology route collaborative decision-making system for new energy vehicle power systems according to claim 5, characterized in that, The adjustment module includes: The identification submodule is used to identify the data type of the initial processing data. If the initial processing data is image data, a first preset model is used to extract features from the initial processing data. If the initial processing data is text data, a second preset model is used to extract features from the initial processing data. If the initial processing data is numerical data, a third preset model is used to extract features from the initial processing data to obtain extracted features. The encryption submodule is used to perform homomorphic encryption on the extracted features to obtain encrypted data, and then uses a Laplace mechanism to perform differential privacy protection on the encrypted data to obtain processed features, wherein the noise level of the differential privacy protection is... : ； In the formula, A collection of encrypted data, for The collection after deleting one record; The mapping submodule is used to minimize the target distance using a target iterative algorithm and calculate the optimal transfer matrix. Based on the optimal transfer matrix, the processed features are mapped to a unified space to obtain fused data, wherein the target distance is... : ； In the formula, These are the first marginal distribution and the second marginal distribution, respectively. To remove the indeterminate boundary, Indicates the joint distribution Samples in the following processing features Seeking expectations, Indicates all that satisfy the marginal distribution as joint distribution A set of.

7. The cross-technology route collaborative decision-making system for new energy vehicle power systems according to claim 5, characterized in that, The adjustment module includes: The first calculation submodule is used to calculate the timeliness score of the fused data. : ； In the formula, Half-life, The age of the data to be merged; The second calculation submodule is used to calculate the integrity score of the fused data. : ； In the formula, These represent the number of missing fields, the number of validated fields, and the total number of fields in the merged data, respectively. The third calculation submodule is used to calculate the data credibility based on the timeliness score and the integrity score. : ； In the formula, To determine the authority coefficient of the data sources; The adjustment submodule is used to remove fused data whose data credibility is less than the first threshold. When the data credibility of the fused data is not less than the first threshold and the integrity score is less than the second threshold, the corresponding fused data is cleaned and missing values ​​are filled. When the data credibility of the fused data is not less than the first threshold and the integrity score is less than the third threshold, the corresponding fused data is downgraded to historical reference data to obtain the adjusted data.

8. A computer comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the cross-technology route collaborative decision-making method for new energy vehicle power systems as described in any one of claims 1 to 4.

9. A storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the cross-technology route collaborative decision-making method for the new energy vehicle power system as described in any one of claims 1 to 4.

Images