Vehicle preventive maintenance control method and device based on full life cycle

Abstract

The application provides a vehicle preventive maintenance control method and device based on a whole life cycle, which can be used in the field of artificial intelligence technology, and the method comprises the following steps: identifying the current life cycle of a vehicle based on multi-modal archive data of the vehicle; generating an optimal maintenance scheme according to the current life cycle and the multi-modal archive data through a pre-constructed maintenance strategy model, monitoring maintenance behavior data of the vehicle, and generating a knowledge report for guiding vehicle selection; through multi-modal fusion of an LSTM-Transformer architecture and an attention mechanism, the prediction accuracy and feature representation ability of the residual life of key components are significantly improved; based on the iterative driving model of the maintenance behavior data in a closed loop, the adaptive fast convergence of the maintenance strategy model is realized; meanwhile, combined with whole life cycle cost analysis and a rule engine, the calculation complexity of maintenance decisions under complex working conditions is reduced, and a high-dimensional knowledge report for guiding selection is generated.

Description

Technical Field

[0001] This invention relates to the field of vehicle maintenance control technology, particularly to the field of artificial intelligence technology, and especially to a method and device for vehicle preventive maintenance control based on the entire life cycle. Background Technology

[0002] Vehicle lifecycle management covers the entire process from planning and selection to disposal and scrapping, with the core objectives of maximizing asset value, minimizing operating costs, and controlling safety risks. Preventive maintenance, as a core component of lifecycle management, directly determines the operational safety and lifespan of a vehicle. However, current vehicle maintenance mainly relies on traditional passive maintenance or fixed-cycle maintenance models, with maintenance decisions based on a single factor, making it difficult to achieve the core objective of "prevention over repair."

[0003] Several vehicle maintenance and condition prediction solutions have been disclosed in the prior art. For example, prior art 1 (publication number: CN121412957A) proposes an artificial intelligence system and method for preventive maintenance of automobiles, which uses an improved adaptive temporal dynamic degradation prediction algorithm to predict the degradation of key components and utilizes a self-supervised anomaly detection algorithm for early warning. This solution only utilizes onboard sensors and static historical data, resulting in limited data sources; its maintenance suggestion generation module is only based on the component degradation prediction results, which means that the maintenance strategy cannot be dynamically adjusted according to changes in vehicle condition. As another example, prior art 2 (publication number: CN121436402A) discloses a real-time condition prediction management system for buses, which realizes operational status monitoring through functions such as CAN bus data acquisition, cloud-based predictive analysis, real-time early warning, mechanical maintenance, driver behavior assessment, and one-vehicle-one-file management. The current system relies on preset fixed thresholds (such as battery temperature threshold and rapid acceleration frequency threshold) for fault prediction and early warning. These thresholds cannot be adaptively adjusted based on the actual operating status of components. Its maintenance reminders use a dual-trigger mechanism of mileage and time thresholds, which constitutes fixed-cycle maintenance and does not consider the impact of different usage conditions on component wear and tear. Furthermore, the system lacks a closed-loop feedback mechanism for maintenance effectiveness, making it impossible to continuously improve the predictive model using post-maintenance data. Therefore, a more comprehensive vehicle preventative maintenance management solution is urgently needed.

[0004] This section is intended to provide background or context for the embodiments of this application set forth in the claims. The description herein is not an admission that it is prior art simply because it is included in this section. Summary of the Invention

[0005] One objective of this invention is to provide a vehicle preventative maintenance control method based on the entire lifecycle. Through multimodal fusion of an LSTM-Transformer architecture and an attention mechanism, it significantly improves the prediction accuracy and feature representation capability of the remaining lifespan of key components. Data-driven model iteration based on closed-loop monitoring of maintenance behavior enables adaptive and rapid convergence of the maintenance strategy model. Simultaneously, by combining lifecycle cost analysis and a rule engine, it reduces the computational complexity of maintenance decisions under complex operating conditions and generates a high-dimensional knowledge report that can guide vehicle selection. Another objective of this invention is to provide a vehicle preventative maintenance control device based on the entire lifecycle. A further objective of this invention is to provide a computer-readable medium. A final objective of this invention is to provide a computer device.

[0006] To achieve the above objectives, this invention discloses a vehicle preventive maintenance control method based on the entire life cycle, comprising: Based on the collected multimodal vehicle profile data, the current lifecycle of the vehicle is identified; By using a pre-built maintenance strategy model, the optimal maintenance plan is generated based on the current life cycle and multimodal profile data, and the vehicle's maintenance behavior data is monitored. The maintenance strategy model is iteratively optimized based on maintenance behavior data, and a knowledge report is generated to guide vehicle selection.

[0007] Preferably, the multimodal profile data includes mileage and years of use; Based on the collected multimodal vehicle profile data, the current lifecycle of the vehicle is identified, including: If the mileage is greater than or equal to the preset first mileage threshold, or if the service life is greater than or equal to the preset first year threshold, the current life cycle of the vehicle will be determined as the deterioration and scrapping period. If the mileage is greater than or equal to the preset second mileage threshold, or if the service life is greater than or equal to the preset second year threshold, the current life cycle of the vehicle will be determined as the period of accelerated wear and tear. If the mileage is greater than or equal to the preset third mileage threshold, or if the service life is greater than or equal to the preset third year threshold, the current life cycle of the vehicle will be determined as the stable operation period. If the mileage is less than the third mileage threshold and the service life is less than the third year threshold, the current life cycle of the vehicle is determined as the break-in period.

[0008] Preferably, an optimal maintenance plan is generated based on the current lifecycle and multimodal profile data using a pre-built maintenance strategy model, including: By using a pre-built lifespan prediction model, the lifespan information of key components is predicted based on the current lifespan and multimodal archive data. The lifespan prediction model is trained based on a long short-term memory network and a Transformer architecture. Based on the maintenance strategy model, the optimal maintenance plan is generated according to life information and multimodal archive data.

[0009] Preferably, the lifespan information of critical components is predicted using a pre-built lifespan prediction model based on current lifespan and multimodal profile data, including: A multimodal fusion algorithm based on an attention mechanism is used to fuse multimodal archival data to obtain a fused feature vector. By using a lifespan prediction model, the lifespan of key components is predicted based on the fused feature vector and the current lifespan, generating lifespan information for the key components.

[0010] Preferably, the multimodal archive data includes operational status data, driving behavior data, and environmental data; Based on the maintenance strategy model, and using lifespan information and multimodal profile data, an optimal maintenance plan is generated, including: Based on the rules engine, a list of maintenance requirements is generated according to lifespan information, operating status data, driving behavior data, and environmental data. By using a pre-built full life cycle cost analysis model, cost optimization is performed based on the maintenance needs list to generate the optimal maintenance list; Based on personalized maintenance standards and basic maintenance standards, the system dynamically adjusts according to the optimal maintenance list to generate the optimal maintenance plan. Personalized maintenance standards include vehicle scenarios and user preferences.

[0011] Preferably, the maintenance strategy model is iteratively optimized based on maintenance behavior data, and a knowledge report is generated to guide vehicle selection, including: Update the vehicle's multimodal profile data based on maintenance behavior data; Based on the updated multimodal profile data, the rule engine is optimized to obtain the updated rule engine; Statistical analysis is performed on the multimodal archive data, optimal maintenance plans, and maintenance behavior data of vehicles in the decline and scrapping period throughout their entire life cycle to generate knowledge reports. The knowledge reports include selection suggestions based on vehicle scenarios and user preferences.

[0012] This invention also discloses a vehicle preventive maintenance control device based on the entire life cycle, comprising: The lifecycle identification unit is used to identify the current lifecycle of a vehicle based on the collected multimodal vehicle profile data. The maintenance plan generation unit is used to generate the optimal maintenance plan based on the current life cycle and multimodal profile data through a pre-built maintenance strategy model, and to monitor the vehicle's maintenance behavior data. The maintenance strategy optimization unit is used to iteratively optimize the maintenance strategy model based on maintenance behavior data and generate a knowledge report to guide vehicle selection.

[0013] The present invention also discloses a computer-readable medium having a computer program stored thereon, which, when executed by a processor, implements the method described above.

[0014] The present invention also discloses a computer device, including a memory and a processor, wherein the memory is used to store information including program instructions, and the processor is used to control the execution of the program instructions, wherein the processor executes the program to implement the method described above.

[0015] The present invention also discloses a computer program product, including a computer program / instruction, which, when executed by a processor, implements the method described above.

[0016] This invention identifies the current lifecycle of a vehicle based on collected multimodal vehicle profile data. Through a pre-built maintenance strategy model, it generates an optimal maintenance plan based on the current lifecycle and multimodal profile data, and monitors vehicle maintenance behavior data. The maintenance strategy model is iteratively optimized based on the maintenance behavior data, and a knowledge report is generated to guide vehicle selection. Through multimodal fusion using an LSTM-Transformer architecture and attention mechanism, the prediction accuracy and feature representation capability of the remaining lifespan of key components are significantly improved. The closed-loop monitoring-driven maintenance behavior data-driven model iteration achieves adaptive and rapid convergence of the maintenance strategy model. Simultaneously, by combining full lifecycle cost analysis and a rule engine, the computational complexity of maintenance decisions under complex operating conditions is reduced, and a high-dimensional knowledge report that can guide vehicle selection is generated. 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 some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A schematic diagram of the architecture of a vehicle preventive maintenance control system based on the entire life cycle, provided for an embodiment of the present invention; Figure 2 A flowchart illustrating a vehicle preventive maintenance control method based on the entire life cycle, provided as an embodiment of the present invention; Figure 3 A flowchart of another vehicle preventive maintenance control method based on the entire life cycle provided in an embodiment of the present invention; Figure 4 A schematic diagram of a vehicle preventive maintenance control device based on the entire life cycle, provided for an embodiment of the present invention; Figure 5 This is a schematic diagram of the structure of a computer device provided in an embodiment of 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] It should be noted that the vehicle preventive maintenance control method and device based on the whole life cycle disclosed in this application can be used in the field of artificial intelligence technology, or in any field other than artificial intelligence technology. The application field of the vehicle preventive maintenance control method and device based on the whole life cycle disclosed in this application is not limited.

[0021] To facilitate understanding of the technical solution provided in this application, the relevant content of the technical solution will be explained below. The entire lifecycle of a vehicle typically encompasses multiple stages, including planning and selection, procurement and access, operation and use, maintenance and disposal, and scrapping. The wear and tear patterns of vehicle components and maintenance needs differ significantly at different stages. Preventative maintenance is a maintenance model that predicts potential faults based on the actual condition of the vehicle and intervenes in advance. Its core lies in dynamically assessing the vehicle's health status based on multi-source heterogeneous data (such as mileage, service life, operating status, driving behavior, and environmental data) and taking targeted measures before faults occur. The maintenance strategy model involved in this application can integrate the above-mentioned multi-modal data to achieve intelligent prediction of the lifespan of key components and generate an optimal maintenance plan that matches the vehicle's current lifecycle and user scenario.

[0022] This invention provides a vehicle preventative maintenance control method based on the entire lifecycle, covering the entire lifecycle of vehicle planning and selection, procurement access, operation and use, maintenance, and disposal. It integrates multi-source lifecycle data and uses intelligent algorithms to predict the lifespan of core components and accurately forecast maintenance needs. It then formulates personalized and dynamic maintenance plans and constructs a closed-loop management system encompassing "data collection, analysis and prediction, plan formulation, execution supervision, effect verification, iterative optimization, and disposal evaluation." This achieves "maintenance instead of repair," reduces vehicle failure rates, extends vehicle lifespan, optimizes lifecycle maintenance costs, improves the overall operation and maintenance management level of vehicles throughout their lifecycle, and is adaptable to the full-stage maintenance needs of various types of vehicles.

[0023] Figure 1 A schematic diagram of the architecture of a vehicle preventive maintenance control system based on the entire life cycle is provided for an embodiment of the present invention, as shown below. Figure 1 As shown, the system includes a full lifecycle data acquisition layer, a data preprocessing and fusion layer, an intelligent analysis and prediction layer, a maintenance plan generation layer, an execution monitoring layer, an effect verification layer, an iterative optimization layer, and a disposal evaluation layer.

[0024] The full lifecycle data acquisition layer serves as the data entry point, covering all stages of vehicle planning and selection, procurement access, operation and use, maintenance and disposal. It collects multi-source information such as basic vehicle data, operating status data, driving behavior data, environmental data, maintenance and fault data, and disposal data, and uploads them in real time through encrypted communication via the data transmission unit.

[0025] The data transmission unit adopts a multi-communication architecture with 5G as the main source and 4G as backup. Combined with the AES encryption algorithm, it transmits the collected multi-source data to the data preprocessing and fusion layer in real time with a transmission delay of ≤50ms, ensuring the real-time performance, security and integrity of the data and preventing data leakage and tampering.

[0026] The data preprocessing and fusion layer is used to clean, complete, denoise, and standardize multi-source data to eliminate data silos; a multimodal fusion algorithm based on attention mechanism is used to generate structured fusion feature vectors, providing high-quality input for subsequent analysis.

[0027] The intelligent analysis and prediction layer divides the entire life cycle of a vehicle (break-in period, stable operation period, accelerated wear period, and deterioration and scrapping period) based on fused data; it uses a life prediction model based on Long Short-Term Memory Network (LSTM) and Transformer architecture to predict the remaining life and aging trend of core components; predicts maintenance needs; and establishes a full life cycle cost analysis model.

[0028] The maintenance plan generation layer is used to develop personalized and dynamic preventive maintenance plans based on forecast results, vehicle characteristics, and user needs. It clarifies maintenance items, priorities, execution standards, and costs, and supports dynamic adjustments based on real-time status changes.

[0029] The execution and supervision layer is used to send maintenance reminders through multiple channels, track the progress of maintenance plans, remotely supervise the maintenance process (such as uploading photos and inspection reports), and synchronize the execution data to the vehicle's full life cycle file in real time.

[0030] The effectiveness verification layer is used to compare the condition of components before and after maintenance, statistically analyze changes in failure rate, evaluate the cost-effectiveness of maintenance, collect user feedback, and quantitatively verify the maintenance effect.

[0031] The iterative optimization layer is used to fine-tune the life prediction model, optimize maintenance strategies and plans (such as updating every 24 hours) based on the effect verification results and full life cycle data, so as to achieve continuous self-iteration of the system.

[0032] The disposal assessment layer is used to conduct condition assessment and residual value analysis on vehicles in the decline and scrapping period, generate selection recommendation reports, and feed back the disposal stage data to the planning and selection stage, forming a closed loop of the entire life cycle of "selection-operation-maintenance-disposal-selection".

[0033] In the technical solution provided by this invention, the current life cycle of a vehicle is identified based on the collected multimodal vehicle profile data. An optimal maintenance plan is generated based on the current life cycle and multimodal profile data using a pre-built maintenance strategy model, and the vehicle's maintenance behavior data is monitored. The maintenance strategy model is iteratively optimized based on the maintenance behavior data, and a knowledge report is generated to guide vehicle selection. Through multimodal fusion of the LSTM-Transformer architecture and attention mechanism, the prediction accuracy and feature representation capability of the remaining lifespan of key components are significantly improved. The model iteration driven by closed-loop monitoring of maintenance behavior data achieves adaptive and rapid convergence of the maintenance strategy model. Simultaneously, by combining full life cycle cost analysis and a rule engine, the computational complexity of maintenance decisions under complex operating conditions is reduced, and a high-dimensional knowledge report that can guide vehicle selection is generated.

[0034] It is worth noting that, Figure 1 The vehicle preventative maintenance control system architecture shown is also applicable to the entire lifecycle of vehicles. Figure 2 or Figure 3 The vehicle preventive maintenance control method based on the entire life cycle will not be elaborated here.

[0035] The following example uses a vehicle preventive maintenance control device based on the entire lifecycle as the executing entity to illustrate the implementation process of the vehicle preventive maintenance control method based on the entire lifecycle provided in this embodiment of the invention. It is understood that the executing entity of the vehicle preventive maintenance control method based on the entire lifecycle provided in this embodiment of the invention includes, but is not limited to, a vehicle preventive maintenance control device based on the entire lifecycle.

[0036] Figure 2 A flowchart of a vehicle preventive maintenance control method based on the entire life cycle, provided as an embodiment of the present invention, is shown below. Figure 2 As shown, the method includes: Step 101: Based on the collected multimodal vehicle profile data, identify the current lifecycle of the vehicle.

[0037] In this embodiment of the invention, multimodal archive data refers to a collection of data from multiple sources and with multiple structures generated throughout the entire life cycle of a vehicle, including but not limited to vehicle basic data, operating status data, driving behavior data, environmental data, maintenance data, fault data, and handling data.

[0038] Vehicle basic data clarifies the inherent attributes and initial state of a vehicle, providing a basis for setting maintenance benchmarks throughout its entire lifecycle. Vehicle basic data includes, but is not limited to: basic data from the vehicle planning and selection and procurement access stages, including vehicle model, factory parameters (engine model, battery capacity, chassis specifications, etc.), purchase information (purchase time, price, warranty period), vehicle configuration (safety features, powertrain type), and pre-delivery inspection (PDI) data.

[0039] Real-time operational status data of the vehicle during operation is collected through the On-Board Diagnostics (OBD) interface, GPS / BeiDou positioning module, and various on-board sensors (temperature, humidity, tire pressure, vibration sensors, etc.). As an optional solution, the data collection frequency is ≥1 time / minute, and the positioning accuracy is ≤1 meter. Operational status data includes, but is not limited to: engine speed, vehicle speed, fuel / electricity consumption, tire pressure, number of braking actions, fault codes, state of charge (SOC) of the battery in new energy vehicles, and component operating parameters (such as transmission oil temperature and braking system pressure).

[0040] By using onboard sensors and driver assistance systems, driving behavior data is collected to analyze its impact on vehicle component wear and tear, providing support for the development of personalized maintenance plans. Driving behavior data includes, but is not limited to: rapid acceleration, sudden braking, idling time, high-speed driving time, and overloading conditions.

[0041] By using onboard environmental sensors, connecting to meteorological departments, and accessing road condition databases, environmental data is collected from vehicles to analyze the impact of environmental factors on the aging of vehicle components. Examples include high humidity accelerating vehicle corrosion and rough road conditions accelerating chassis wear. Environmental data includes, but is not limited to: ambient temperature, humidity, rainfall, air quality, road conditions (urban roads / mountain roads / highways), and corrosive environments.

[0042] Maintenance data is a complete set of data for the vehicle maintenance phase, including but not limited to: maintenance time, maintenance items, replaced parts (brand, model, replacement cycle), maintenance costs, maintenance institutions, and maintenance personnel.

[0043] Fault data refers to fault data throughout the entire life cycle of a vehicle, including but not limited to: fault type, fault location, fault cause, repair plan, repair cost, and repair effect.

[0044] This application provides a basis for predicting component lifespan and optimizing maintenance strategies by establishing complete maintenance and failure records.

[0045] Disposal data refers to data from the vehicle's decline and scrapping stage, providing feedback for the evaluation of the effectiveness of full life-cycle maintenance and subsequent vehicle selection. Disposal data includes vehicle age, mileage, residual value assessment results, degree of component aging, disposal method (used car auction, targeted transfer, scrapping and dismantling), and disposal costs.

[0046] In this embodiment of the invention, the entire life cycle includes: break-in period, stable operation period, accelerated wear period, and deterioration and scrapping period.

[0047] Specifically, the collected mileage is compared with multiple preset mileage thresholds, or the service life is compared with multiple service life thresholds. Based on the comparison results, the vehicle is determined to be in one of the following stages: break-in period, stable operation period, accelerated wear period, or deterioration and scrapping period. For example, when the mileage is below the first threshold and the service life is below the first service life threshold, it is determined to be in the break-in period; when the mileage or service life reaches the threshold corresponding to the deterioration and scrapping period, it is determined to be in the deterioration and scrapping period.

[0048] This invention uses a quantitative threshold comparison mechanism to quickly and accurately identify the current life cycle stage of a vehicle without the need for human experience, providing a stage-adaptive basis for subsequent differentiated maintenance decisions.

[0049] Step 102: Using a pre-built maintenance strategy model, generate the optimal maintenance plan based on the current life cycle and multimodal profile data, and monitor the vehicle's maintenance behavior data.

[0050] In this embodiment of the invention, the maintenance strategy model integrates a life prediction model, a total life cost (TCO) analysis model, and a rule engine, which can output personalized maintenance plans based on the input multimodal archive data and life cycle stages.

[0051] The optimal maintenance plan includes, but is not limited to, an executable plan containing information such as maintenance items, maintenance time / mileage milestones, execution priority, estimated costs, and recommended maintenance institutions. This plan optimizes the total life cycle cost while meeting the vehicle's safe operation requirements.

[0052] Maintenance behavior data records the actual data of the entire maintenance plan execution process, including but not limited to: maintenance start and end time, actual items performed, replaced parts, costs incurred, executing agency, and photos or inspection reports uploaded during the maintenance process.

[0053] Specifically, the current lifecycle and multimodal profile data are input into a pre-trained maintenance strategy model; the remaining lifecycle of key components such as the engine, battery, and braking system is predicted using a lifecycle prediction module (e.g., based on an LSTM+Transformer architecture); a preliminary maintenance requirement list is generated through a rule engine by combining operating status, driving behavior, environmental data, and user preferences; the alternative solutions are optimized for cost through a full lifecycle cost analysis model, and the optimal maintenance solution is output; maintenance reminders are sent to users or maintenance personnel through in-vehicle terminals, mobile apps, and other channels, and the start time, completion time, actual items, costs, and on-site evidence (such as photos and inspection reports) are recorded in real time during the maintenance process to form complete maintenance behavior data.

[0054] This invention realizes an automated decision-making chain from status perception to maintenance plan generation. The generated plan takes into account the actual wear and tear of components, user preferences, and total life cycle cost. At the same time, by monitoring maintenance behavior data in real time, it ensures the traceability and transparency of maintenance plans.

[0055] Step 103: Iteratively optimize the maintenance strategy model based on maintenance behavior data, and generate a knowledge report to guide vehicle selection.

[0056] In this embodiment of the invention, the knowledge report is a summary document generated by statistical analysis of the full life cycle data of a vehicle in its decline and scrapping period. The content covers the failure patterns of components, maintenance cost-effectiveness, residual value change trends, and specific suggestions for future vehicle selection under different usage scenarios and user preferences.

[0057] Specifically, the monitored maintenance behavior data is written back to the vehicle's multimodal profile database in real time or periodically, updating the vehicle's maintenance records, component replacement records, and fault records. Based on the updated data, the rule engine parameters (such as maintenance cycle and lifespan prediction threshold) within the maintenance strategy model are automatically fine-tuned and optimized, enabling the model to continuously adapt to individualized changes in vehicle usage. Simultaneously, for vehicles that have entered their deterioration and scrapping period, all multimodal profile data, optimal maintenance plans, and actual maintenance behavior data throughout their entire lifecycle are aggregated for multi-dimensional statistical analysis (such as component failure rates under different driving habits and cost-residual value curves for different maintenance strategies), ultimately generating a knowledge report. This report, in the form of visual charts and textual suggestions, provides selection recommendations for different vehicle scenarios (such as urban logistics and mountain transportation) and user preferences (such as cost priority and safety priority).

[0058] This invention utilizes a closed-loop iterative mechanism, enabling the maintenance strategy model to continuously evolve with actual vehicle usage, maintaining the accuracy of long-term decision-making. Simultaneously, the generated knowledge report feeds back practical data to the vehicle selection stage, forming a data closed loop throughout the entire lifecycle and providing quantitative evidence for reducing the total lifecycle cost of future vehicles.

[0059] In the technical solution provided by this invention, the current life cycle of a vehicle is identified based on the collected multimodal vehicle profile data. An optimal maintenance plan is generated based on the current life cycle and multimodal profile data using a pre-built maintenance strategy model, and the vehicle's maintenance behavior data is monitored. The maintenance strategy model is iteratively optimized based on the maintenance behavior data, and a knowledge report is generated to guide vehicle selection. Through multimodal fusion of the LSTM-Transformer architecture and attention mechanism, the prediction accuracy and feature representation capability of the remaining lifespan of key components are significantly improved. The model iteration driven by closed-loop monitoring of maintenance behavior data achieves adaptive and rapid convergence of the maintenance strategy model. Simultaneously, by combining full life cycle cost analysis and a rule engine, the computational complexity of maintenance decisions under complex operating conditions is reduced, and a high-dimensional knowledge report that can guide vehicle selection is generated.

[0060] In one embodiment of the present invention, it is applied to a city logistics fleet (comprising 500 fuel-powered logistics vehicles and 200 new energy logistics vehicles). All parameters and processes are tailored to the actual needs of the logistics fleet's high-frequency use and multi-scenario operation. The vehicles in this logistics fleet are mainly used for urban and intercity logistics transportation, with high usage intensity and complex road conditions (urban roads, highways, and rugged suburban roads). Existing maintenance models suffer from over-maintenance, frequent breakdowns, and high costs. The present invention system achieves preventive maintenance throughout the entire life cycle, reducing the failure rate, optimizing maintenance costs, and improving fleet operation efficiency.

[0061] In terms of hardware environment: each vehicle is equipped with an OBD terminal, a GPS / BeiDou dual-mode positioning module, and multi-dimensional sensors (temperature, humidity, tire pressure, vibration, and engine speed sensors). New energy vehicles are additionally equipped with battery status sensors, totaling 700 sets of on-board data acquisition equipment; the backend is deployed with high-performance servers and 15 edge computing nodes (deployed at fleet depots for local data preprocessing); terminal equipment includes 10 fleet management terminals, 700 driver mobile apps, and 5 maintenance agency docking terminals; the communication network uses 50 5G base stations and 20 4G backup devices to ensure real-time data transmission.

[0062] Software environment: It adopts an enterprise-level operating system, and the core algorithm models (including the life prediction model based on LSTM and Transformer architecture and the multimodal fusion algorithm based on attention mechanism) have been fine-tuned and optimized with full life cycle data of more than 1 million vehicles; the data storage adopts a hybrid storage architecture, the programming language adopts a high-performance compiled language, the data processing framework supports distributed computing, the maintenance management platform is developed using mainstream front-end frameworks, and it connects to the meteorological department interface, road condition database and the systems of 5 cooperative maintenance institutions.

[0063] The core parameters are configured as follows: data acquisition frequency is 1 time / minute, positioning accuracy is no more than 1 meter, transmission delay is no more than 50 milliseconds, data completion rate is no less than 98%, and invalid data rejection rate is no more than 2%; in terms of prediction indicators, the accuracy rate of core component life prediction is no less than 97%, the accuracy of maintenance demand prediction is no less than 96%, and the failure rate is reduced by no less than 80%; the maintenance cycle is set differently according to the life cycle stage (3000 km / 3 months during the break-in period, 10000 km / 6 months for fuel vehicles and 15000 km / 6 months for new energy vehicles during the stable operation period, and wear and tear...). Maintenance reminders are sent 3 days in advance (acceleration period 5000 km / 3 months); the model is optimized every 24 hours, data is statistically analyzed weekly, the system is upgraded monthly, and the maintenance strategy is optimized monthly; the expected results are a reduction in maintenance costs of no less than 30%, an extension of vehicle lifespan of no less than 20%, and a reduction in total lifespan cost of no less than 25%; the component lifespan threshold triggering conditions include: tire tread depth no greater than 1.6 mm, brake fluid usage time no less than 2 years, timing belt usage mileage no less than 80,000 km, and new energy vehicle battery SOC degradation no less than 20% when maintenance reminders are triggered.

[0064] Figure 3 A flowchart of another vehicle preventive maintenance control method based on the entire life cycle provided in this embodiment of the invention is shown below. Figure 3 As shown, the method includes: Step 201: Based on the collected multimodal vehicle profile data, identify the current lifecycle of the vehicle.

[0065] In this invention, each step is performed by a vehicle preventative maintenance control device based on the entire life cycle.

[0066] In this embodiment of the invention, step 201 specifically includes: Step 2011: Determine whether the mileage is greater than or equal to the preset first mileage threshold, or whether the service life is greater than or equal to the preset first year threshold. If either is true, proceed to step 2012; if neither is true, proceed to step 2013.

[0067] In this embodiment of the invention, if the mileage is greater than or equal to the first mileage threshold, or the service life is greater than or equal to the preset first year threshold, it indicates that the vehicle has approached or reached the end of its service life, and step 2012 is executed; if the mileage is less than the first mileage threshold, and the service life is less than the first year threshold, it indicates that the vehicle has not reached the end of its service life, and step 2013 is executed.

[0068] It is worth noting that the first mileage threshold and the first year threshold can be set according to actual needs, and this embodiment of the invention does not limit them. As an optional solution, the first mileage threshold is 200,000 kilometers, and the first year threshold is 8 years.

[0069] Step 2012: Determine the current life cycle of the vehicle as its deterioration and scrapping period, and continue to step 202.

[0070] In this embodiment of the invention, if a vehicle is nearing or has reached the end of its service life, the current life cycle of the vehicle is defined as the deterioration and scrapping period. During this stage, the focus is on evaluating the condition of the vehicle's core components and the vehicle's residual value.

[0071] This step can automatically identify vehicles that are nearing or have reached the end of their service life, and then formulate corresponding disposal or maintenance strategies in subsequent steps based on the characteristics of this stage (such as focusing on assessing the remaining value and residual value of parts), so as to avoid unnecessary over-maintenance of vehicles that are about to be scrapped.

[0072] Step 2013: Determine whether the mileage is greater than or equal to the preset second mileage threshold, or whether the service life is greater than or equal to the preset second year threshold. If either is true, proceed to step 2014; if neither is true, proceed to step 2015.

[0073] In this embodiment of the invention, if the mileage is greater than or equal to the second mileage threshold, or the service life is greater than or equal to the second service life threshold, it indicates that the wear rate of vehicle components has begun to accelerate, and step 2014 is continued; if the mileage is less than the preset second mileage threshold, and the service life is less than the preset second service life threshold, it indicates that the vehicle has not entered the stage of rapid wear, and step 2015 is continued.

[0074] It is worth noting that the second mileage threshold and the second year threshold can be set according to actual needs, and this embodiment of the invention does not limit this. As an optional solution, the second mileage threshold is 100,000 kilometers, and the second year threshold is 5 years.

[0075] Step 2014: Determine the current life cycle of the vehicle as the period of accelerated wear and tear, and continue to step 202.

[0076] In this embodiment of the invention, if the wear rate of vehicle components begins to accelerate, the current life cycle of the vehicle is defined as the wear acceleration period. During this stage, the focus is on maintaining components that are prone to rapid aging, such as the chassis and braking system.

[0077] This step can identify when a vehicle is entering a stage of accelerated component wear, thus allowing for a focus on increasing the maintenance frequency of easily aging components such as the chassis and braking system in subsequent steps to prevent malfunctions caused by insufficient maintenance.

[0078] Step 2015: Determine whether the mileage is greater than or equal to the preset third mileage threshold, or whether the service life is greater than or equal to the preset third year threshold. If either is true, proceed to step 2016; if neither is true, proceed to step 2017.

[0079] In this embodiment of the invention, if the mileage is greater than or equal to the third mileage threshold, or the service life is greater than or equal to the preset third year threshold, it indicates that the vehicle is in a stable operating phase, and step 2016 is executed; if the mileage is less than the preset third mileage threshold, and the service life is less than the preset third year threshold, it indicates that the vehicle has not entered a stable operating phase, and step 2017 is executed.

[0080] It is worth noting that the third mileage threshold and the third year threshold can be set according to actual needs, and this embodiment of the invention does not limit them. As an optional solution, the third mileage threshold is 10,000 kilometers, and the third year threshold is 1 year.

[0081] Step 2016: Determine the current life cycle of the vehicle as the stable operation period, and continue to execute step 202.

[0082] In this embodiment of the invention, if the vehicle is in a stable operating phase, the current life cycle of the vehicle is defined as the stable operating period. During this phase, the focus is on maintaining routine components and adopting standard maintenance cycles.

[0083] This step identifies when the vehicle is in its most stable operating phase, allowing for the use of standard maintenance cycles and procedures in subsequent steps, thus avoiding resource waste caused by over-maintenance.

[0084] Step 2017: Determine the current life cycle of the vehicle as the break-in period, and continue to step 202.

[0085] In this embodiment of the invention, if the mileage is less than the third mileage threshold and the service life is less than the third service life threshold, the current life cycle of the vehicle is determined as the break-in period. During this period, the system focuses on maintaining key components such as the engine and transmission, and the maintenance cycle is relatively short.

[0086] This step can accurately identify new cars or vehicles that have just been put into use, thereby enabling the development of more frequent maintenance plans for key components such as the engine and transmission in subsequent steps, reducing the risk of early failures.

[0087] In one embodiment of the present invention (applied to a fleet of 700 logistics vehicles), the criteria for dividing each stage are as follows: Break-in period: Mileage ∈ [0, 10,000 km) and service life ∈ [0, 1 year). A total of 120 vehicles. Key maintenance: Engine and transmission.

[0088] Stable operating period: Mileage ∈ [10,000 km, 100,000 km) or service life ∈ [1 year, 5 years). Total 450 vehicles. Key maintenance: routine parts.

[0089] Accelerated wear and tear period: Mileage ∈ [100,000 km, 200,000 km) or service life ∈ [5 years, 8 years). A total of 110 vehicles are included. Key maintenance areas: chassis and braking system.

[0090] Deterioration and scrapping period: Mileage ≥ 200,000 km or service life ≥ 8 years. Total 20 vehicles. Key maintenance: Condition of core components and vehicle residual value.

[0091] Through the above steps, this invention can automatically output the current life cycle stage of a vehicle (break-in period, stable operation period, accelerated wear period, or deterioration and scrapping period) based on mileage and years of use, providing basic input for generating differentiated maintenance plans.

[0092] Step 202: Using a pre-built life prediction model, predict the life information of key components based on the current life cycle and multimodal archive data.

[0093] In this embodiment of the invention, the lifespan prediction model is trained based on a long short-term memory network and a Transformer architecture. The lifespan prediction model is fine-tuned and optimized using massive amounts of vehicle lifecycle data (covering aging data of different vehicle models, different usage scenarios, and different components) to predict the remaining lifespan and aging trend of key components based on the input fused feature vector and the current lifecycle of the vehicle.

[0094] In this embodiment of the invention, step 202 specifically includes: Step 2021: Use an attention-based multimodal fusion algorithm to fuse multimodal archive data and obtain a fused feature vector.

[0095] In this embodiment of the invention, multimodal archive data undergoes data preprocessing, which includes data cleaning, data noise reduction, and data standardization. Specifically, data cleaning includes: removing abnormal data triggered by sensor malfunctions (such as abnormal fluctuations in engine speed to 5000 rpm) and redundant data (such as repeatedly collected identical operating parameters); for missing data, linear interpolation is used to complete it, achieving a completion rate of 98.6%.

[0096] Data denoising includes: using Gaussian filtering algorithms to remove noise from operational status and environmental data. For unstructured data, such as fault images and maintenance record text, a structured transformation operation is performed to extract key information such as fault type and maintenance items. For image data such as tire wear images, enhancement and normalization processing are performed.

[0097] Data standardization includes: unifying the format, units of measurement, and coding rules of all data. Mileage is standardized to "kilometers," fuel consumption to "liters per 100 kilometers," and electricity consumption to "kilowatt-hours per 100 kilometers." Fault types and maintenance items use standardized codes; for example, engine fault code is 001, and oil change code is 010.

[0098] In this embodiment of the invention, based on the timestamp (e.g., 14:30:00 on August 10, 2024) and the vehicle's unique number (e.g., 001), the vehicle's operating status data (e.g., engine speed 1800 rpm, vehicle speed 60 km / h), driving behavior data (e.g., 4 times of rapid acceleration / hour, 6 times of emergency braking / hour), environmental data (e.g., ambient temperature 25 degrees Celsius, road condition is urban road), and maintenance and fault data (e.g., oil change on August 10, 2024) within the same time or time window are associated one by one to form a multi-source data record after association.

[0099] A multimodal fusion algorithm based on an attention mechanism is invoked to deeply fuse the aforementioned correlated and preprocessed multi-source data records (including vehicle basic data, operating status data, driving behavior data, environmental data, maintenance data, fault data, and handling data). The attention mechanism is used to automatically calculate the correlation weights between different modal data, generating a multi-dimensional (e.g., 768-dimensional) fusion feature vector to mine the correlations between data, such as the correlation between the number of emergency brakings and the acceleration of brake system wear, and the correlation between urban road driving and fuel consumption.

[0100] Furthermore, a real-time data update mechanism is established. Operational status data is updated every minute, environmental data is updated every hour, and maintenance data, fault data, and handling data are synchronized to the system in real time after they are generated, ensuring the timeliness of the fused feature vectors.

[0101] Through the data preprocessing and fusion operations described above, the system can convert multi-source, heterogeneous vehicle lifecycle data into structured fusion feature vectors, effectively integrating the correlation information between different modal data, providing high-quality, high-dimensional input features for subsequent lifecycle prediction models, thereby improving prediction accuracy.

[0102] Step 2022: Using the life prediction model, based on the fused feature vector and the current life cycle, predict the life of key components and generate life information for the key components.

[0103] In this embodiment of the invention, key components refer to core components in a vehicle that require focused monitoring and maintenance, including but not limited to the engine, transmission, tires, braking system, timing system, and batteries of new energy vehicles. Lifespan information refers to the prediction results output by the lifespan prediction model, which includes at least the remaining lifespan value of key components (in kilometers or months / years), a description of aging trends (such as normal wear, excessive wear, or abnormal wear), and whether they are approaching preset lifespan thresholds (such as tire tread depth less than 1.6 mm, brake fluid used for more than 2 years, timing belt used for more than 80,000 kilometers, and new energy vehicle battery SOC degradation exceeding 20%).

[0104] Specifically, the current lifespan of the vehicle is concatenated with the fused feature vector to form a complete input tensor. This input tensor is then fed into the lifespan prediction model. The internal LSTM layer is responsible for capturing the time-series dependencies in the fused feature vector, while the Transformer layer is responsible for weighted aggregation of global features, outputting the prediction results for each key component, i.e., lifespan information.

[0105] Taking the fuel-powered logistics vehicle numbered 001 in this embodiment of the invention as an example, this vehicle is currently in a stable operating period, with a mileage of 60,000 kilometers and a service life of 1.5 years. After inputting the fused feature vector and the current lifespan into the lifespan prediction model, the lifespan prediction model outputs the following lifespan information: Engine remaining life: 60,000 kilometers; Transmission remaining life: 80,000 kilometers; Tire life remaining: 20,000 kilometers (current tire tread depth is 2.2 mm, model predicts that the tread depth will reach the preset threshold of 1.6 mm after 20,000 kilometers). Braking system remaining life: 15,000 kilometers.

[0106] Meanwhile, the life prediction model outputs wear type identification results: the tires are in normal wear, while the braking system shows excessive wear due to frequent emergency braking.

[0107] The accuracy rate of the above prediction results is 97.2%. The life prediction model can also identify potential fault hazards in advance, such as triggering a battery maintenance warning when the SOC of a new energy vehicle battery reaches or exceeds the preset battery energy consumption threshold (20%).

[0108] This invention utilizes a life prediction model based on LSTM and Transformer architecture, combined with multimodal fusion feature vectors and the current life cycle, to accurately predict the remaining life, aging trend, and wear type (normal wear, excessive wear, abnormal wear) of key components. The prediction accuracy reaches over 97%, achieving quantitative prediction of component life and providing a quantitative basis for generating differentiated and precise maintenance plans.

[0109] Step 203: Based on the maintenance strategy model, generate the optimal maintenance plan according to the life information and multimodal archive data.

[0110] In this embodiment of the invention, step 203 specifically includes: Step 2031: Based on the rule engine, generate a maintenance requirement list according to lifespan information, operating status data, driving behavior data, and environmental data.

[0111] In this embodiment of the invention, the rule engine internally stores multiple predefined maintenance rules, each rule comprising a condition section and a conclusion section. The condition section is used to determine whether the input parameters meet specific thresholds or states (e.g., brake system remaining life ≤ 15,000 km; or emergency braking frequency ≥ 5 times / hour; or ambient humidity ≥ 80% and mileage exceeding 10,000 km). The conclusion section corresponds to one or more maintenance requirement items. The maintenance requirement list is a data structure containing several maintenance requirement items, each requirement item including at least the maintenance item (e.g., changing engine oil, checking brake pads), triggering conditions (e.g., remaining life threshold or abnormal operating status), and a suggested execution time or mileage window.

[0112] Specifically, the lifespan information of key components (including the remaining lifespan and wear type of each key component), operational status data (such as engine speed, vehicle speed, fuel consumption, and fault codes) from multimodal archive data, driving behavior data (such as the number of rapid accelerations, the number of emergency brakings, and idling time), and environmental data (such as temperature, humidity, and road condition type) are input parameters to a pre-configured rule engine. The rule engine iterates through all rules and matches each input parameter one by one. When the conditions of a rule are met, the rule engine adds the corresponding maintenance requirement item to the maintenance requirement list. If multiple rules are met simultaneously, all corresponding maintenance requirement items are added. Finally, the rule engine outputs a complete maintenance requirement list, where each requirement item can be accompanied by an initial priority identifier.

[0113] In one embodiment of the present invention, a fuel-powered logistics vehicle numbered 001 is taken as an example. This vehicle is in a stable operating period, with a mileage of 60,000 kilometers and a service life of 1.5 years. Its lifespan information is as follows: engine remaining lifespan 60,000 kilometers, transmission remaining lifespan 80,000 kilometers, tire remaining lifespan 20,000 kilometers (current tread depth 2.2 mm), and braking system remaining lifespan 15,000 kilometers, with the braking system showing signs of excessive wear. Operating status data shows a fuel consumption of 8.5 liters per 100 kilometers and no fault codes. Driving behavior data shows 4 instances of rapid acceleration per hour and 6 instances of emergency braking per hour. Environmental data is urban road conditions, with a temperature of 25 degrees Celsius and humidity of 60%.

[0114] After receiving the above input, the rule engine performs rule matching: Rule 1: If the remaining life of the braking system is ≤15,000 km, a "Brake System Maintenance" requirement will be generated. If the condition is met, a "Check Brake Pads, Replace Brake Pads" requirement will be added.

[0115] Rule 2: If the number of emergency braking incidents is ≥5 times / hour, a "Brake System Enhanced Inspection" request will be generated. If the condition is met, a "Brake System Deep Inspection" request item will be added (which can be combined with the above request).

[0116] Rule 3: If the remaining tire life is ≤20,000 km, generate a "Tire Inspection" request. If the condition is met, add a "Check tire tread depth and tire pressure" request item.

[0117] Rule 4: If the mileage reaches 60,000 kilometers and there are no recent maintenance records, a "routine maintenance" request will be generated. If the conditions are met, a "change engine oil, oil filter, and air filter" request item will be added.

[0118] Rule 5: If the ambient humidity remains above 70% and the mileage exceeds 50,000 kilometers, a "chassis rust prevention inspection" request will be generated. In this example, the humidity is 60%, which does not meet the condition.

[0119] The rules engine outputs a list of maintenance requirements, including the following items: brake system maintenance (replacing brake pads), tire inspection, and routine maintenance (changing engine oil, oil filter, and air filter).

[0120] Step 2032: Optimize costs based on the maintenance needs list using a pre-built TCO analysis model to generate the optimal maintenance list.

[0121] In this embodiment of the invention, the TCO analysis model is used to calculate the total lifecycle cost under different maintenance strategies based on a given list of maintenance needs. The model's inputs include vehicle purchase cost, historical maintenance costs, repair costs, operating costs, and residual value prediction data, and its output is an estimated total cost for each candidate maintenance plan. The optimal maintenance list is a subset or adjusted list of needs output from the maintenance need list after optimization by the TCO analysis model. The maintenance needs in this list minimize the total lifecycle cost while meeting the vehicle's safe operation requirements.

[0122] Specifically, the maintenance needs list is taken as input, and the output is a TCO analysis model. This model internally maintains the vehicle's purchase cost, historical maintenance and repair cost records, operating cost data, and a residual value function based on component remaining life prediction. For each item or combination of options in the maintenance needs list, the TCO analysis model performs the following calculations: estimating the direct costs (material costs, labor costs) required to perform the maintenance item; estimating the effect of maintenance on extending component life based on changes in component remaining life output from the life prediction model; calculating the expected residual value of the vehicle at the disposal and scrapping stage under different maintenance strategies, combined with the vehicle residual value model; summarizing the purchase cost, maintenance costs at each stage, repair costs, and operating costs, and subtracting the expected residual value, to obtain the total lifecycle cost.

[0123] The TCO analysis model iterates through all possible combinations of maintenance needs (e.g., performing only routine maintenance, performing routine maintenance plus brake maintenance, performing all needs, etc.), calculates the total cost for each combination, and selects the combination with the lowest total cost as the optimal maintenance list. If the total costs of two options are close, the option with higher safety is prioritized.

[0124] In one embodiment of the present invention, the maintenance requirement list includes three items: routine maintenance (changing engine oil, oil filter, and air filter), tire inspection, and brake system maintenance (replacing brake pads). The TCO model optimizes the cost of vehicle 001 (fuel logistics vehicle). The TCO analysis model obtains the purchase cost of vehicle 001 (assumed to be 200,000 yuan), historical maintenance cost records, and operating cost data; and compares two maintenance cycle schemes: Scheme A uses a 10,000-kilometer maintenance cycle, and Scheme B uses an 8,000-kilometer maintenance cycle.

[0125] TCO analysis model calculation: Option A (10,000 km cycle): The estimated number of maintenance services over the entire lifespan is 10, with an average cost of 800 yuan per service and a total maintenance cost of 8,000 yuan; the component lifespan meets expectations, and the residual value is estimated at 40,000 yuan.

[0126] Option B (8000 km cycle): The estimated number of maintenance times is 12.5, with an average cost of 800 yuan per maintenance, and a total maintenance cost of 10,000 yuan; the lifespan of parts is slightly extended, but the residual value only increases by 2,000 yuan, resulting in a higher total cost.

[0127] The TCO analysis model considers whether all three maintenance requirements are performed simultaneously. Calculations determined that performing all three maintenance requirements with a 10,000 km maintenance cycle (with the braking system maintained separately at 5,000 km cycles) is the optimal combination. This combination reduces the total lifecycle cost by 28% compared to traditional fixed-cycle maintenance.

[0128] The TCO analysis model outputs an optimal maintenance list, including: routine maintenance (every 10,000 km), brake system maintenance (every 5,000 km), and tire inspection (every 10,000 km).

[0129] Step 2033: Based on the personalized maintenance standards and the basic maintenance standards, dynamically adjust according to the optimal maintenance list to generate the optimal maintenance plan. The personalized maintenance standards include vehicle scenarios and user preferences.

[0130] In this embodiment of the invention, personalized maintenance standards are a set of maintenance rule parameters defined based on vehicle scenarios and user preferences, used to dynamically adjust basic maintenance standards. Vehicle scenarios include vehicle type (gasoline vehicle / new energy vehicle), usage scenario (ride-hailing vehicle / logistics vehicle / personal vehicle), usage intensity, driving style, and usage environment. User preferences include maintenance budget (cost priority / performance priority), required maintenance time, and acceptance of DIY projects.

[0131] Basic maintenance standards are a set of general maintenance rules developed with reference to industry standards, manufacturer maintenance standards, and regulatory requirements. They define the basic execution cycle, execution standards, and minimum safety requirements for various maintenance items.

[0132] The optimal maintenance plan is the final executable maintenance plan, which is a structured data object that includes, but is not limited to: a list of maintenance items, the specific content of each item, the execution standards, the required materials, the estimated cost, the execution cycle, recommended maintenance agencies, the maintenance priority ranking, and a DIY indicator.

[0133] Specifically, the system prioritizes items in the optimal maintenance list based on the urgency of maintenance needs, component importance (safety components have the highest priority), and cost-effectiveness. For example, core safety components such as the braking system and tires are set as high priority, while lights and interior components are set as low priority. The system dynamically adjusts the execution time or mileage of the maintenance plan by combining real-time synchronized vehicle operating status, component aging, and environmental changes. For example, if the vehicle frequently enters rough road conditions or high-humidity environments during subsequent driving, the system automatically advances chassis maintenance items and adds related items. The specific content of maintenance items is adjusted according to vehicle type and usage scenario. For example, battery testing and balancing maintenance are added for new energy vehicles; maintenance cycles are shortened for frequently used ride-hailing vehicles; and chassis maintenance frequency is increased for vehicles driven in mountainous areas. Based on user preferences, the system distinguishes between DIY items (such as air filter replacement) and items requiring professional operation (such as engine repair), and provides corresponding cost estimates and operation instructions. For each maintenance item, the system clearly specifies the execution standards (such as engine oil type), required materials, estimated costs, recommended maintenance institutions (such as the nearest partner 4S store), and suggested execution time windows (such as within one month or when the vehicle reaches 70,000 kilometers). Finally, a complete optimal maintenance plan is output, which is stored in a structured data format and can be directly used for subsequent maintenance reminders and execution monitoring.

[0134] In one embodiment of the invention, vehicle 001 is a fuel-powered logistics vehicle, used for urban and intercity logistics (high-frequency use), driven aggressively, and primarily used on urban roads. User preference prioritizes cost, but some DIY projects are acceptable. Optimal maintenance schedule: routine maintenance every 10,000 km, brake system maintenance every 5,000 km, tire inspection every 10,000 km. The optimal maintenance schedule is dynamically adjusted. Priority sorting: Brake system maintenance is set to high priority (safety components), tire inspection is set to medium priority, routine maintenance (oil change, etc.) is set to medium priority, and chassis inspection is set to low priority (not included in this list, but may be dynamically added due to subsequent environmental changes).

[0135] Dynamic Periodic Adjustment: Due to the aggressive driving style of vehicle 001 (frequent hard braking), the system shortened the brake system maintenance interval from the standard 10,000 km to 5,000 km, consistent with the optimal maintenance list. In the following 10 days, vehicle 001 frequently drove on rough suburban roads, resulting in increased vibration of chassis components. The system detected the changes in environmental road conditions in real time and automatically adjusted the maintenance plan: advancing the original maintenance time by 5 days and adding "chassis lubrication" to the maintenance items.

[0136] Personalized adaptation: Based on vehicle type (gasoline vehicle), battery-related items were not added. Based on user preferences (cost priority), the system marks the air filter as a DIY item for car owners in the solution, prompting users to replace it themselves to save labor costs.

[0137] The final optimal maintenance plan is as follows: Maintenance Item 1: Brake System Maintenance. The standard is to replace the brake pads (wear-resistant type). The estimated cost is 400 yuan. The recommended maintenance institution is the nearest cooperating 4S store. The maintenance cycle is within 65,000 kilometers or 1 month. The priority is high.

[0138] Maintenance Item 2: Tire Inspection. The standard for this is to check the tread depth and tire pressure. The estimated cost is 50 yuan. The performance cycle is the same as regular maintenance, and the priority is medium.

[0139] Maintenance Item 3: Routine maintenance, the standard is to change the engine oil SN 5W-30, oil filter, and air filter (DIY is possible), the estimated cost is 350 yuan, the recommended maintenance institution is the cooperating 4S store, the performance cycle is within 70,000 kilometers or 1 month, the priority is medium.

[0140] Maintenance item 4 (dynamic addition): Chassis lubrication, the standard for execution is lubrication of key chassis hinge points, the estimated cost is 100 yuan, the execution cycle is immediate (5 days in advance), and the priority is high.

[0141] In another embodiment of the present invention, for the new energy logistics vehicle (No. 501), the system formulates the optimal maintenance plan according to the personalized maintenance standard: battery testing and equalization maintenance are performed every 15,000 kilometers or 6 months, and optimized charging strategy suggestions are output (avoiding overcharging and deep discharge) to extend battery life.

[0142] This invention utilizes a rule engine, a full lifecycle cost analysis model, and a layered processing of personalized standards to generate an optimal maintenance plan that balances safety, economy, and user preferences. It also supports dynamic adjustments based on real-time operating conditions, ensuring that the maintenance plan always matches the actual needs of the vehicle and avoiding over-maintenance or under-maintenance.

[0143] Step 204: Monitor vehicle maintenance behavior data.

[0144] In this embodiment of the invention, maintenance behavior data refers to the actual data recorded during the execution of the maintenance plan, including but not limited to: maintenance reminder sending records (time, channel, recipient), maintenance start time, maintenance completion time, actual maintenance items performed, replaced parts (brand, model, quantity), actual costs incurred, information on the organization and personnel performing the maintenance, and process evidence such as photos, videos, and test reports uploaded during the maintenance process.

[0145] In this embodiment of the invention, after generating the optimal maintenance plan, the process enters the execution supervision phase. The execution supervision phase includes: maintenance reminders, execution tracking, process monitoring, and data synchronization.

[0146] Specifically, based on the maintenance time or mileage milestones specified in the optimal maintenance plan, a maintenance reminder message is automatically generated before the scheduled execution time (e.g., 3 days in advance). This message includes at least the following fields: vehicle identification, recommended maintenance time or mileage, maintenance item list, priority order, and recommended maintenance service provider. The reminder message is sent to preset user roles (such as drivers and fleet managers) through at least one communication channel (including in-vehicle terminal display, driver's mobile APP, SMS gateway, and fleet management platform).

[0147] Establish data interfaces with the systems of partner maintenance organizations (such as 4S dealerships and third-party repair shops). When a user schedules maintenance and begins the process, the maintenance organization's system or the user triggers a "Start Maintenance" event on their terminal, recording the maintenance start time (accurate to the minute). During the maintenance process, the maintenance organization's system reports in real time the completed maintenance items, ongoing items, and detailed information on replaced parts (including part name, brand, model, and quantity). After maintenance is completed, the system records the maintenance completion time and summarizes the actual execution content, the actual list of replaced parts, and the actual cost, forming a complete execution tracking record.

[0148] Maintenance centers must upload supporting documentation at key operational points, including but not limited to: before-and-after photos, videos of critical operations, and test reports (such as brake fluid water content tests and tire tread depth measurements). Uploaded files will undergo format verification and timestamp verification. If any non-standard procedures are found (such as failure to upload necessary evidence or replacement of non-designated brand parts), an automatic warning will be issued, requiring the maintenance center to rectify the issues and resubmit.

[0149] After maintenance is completed, all the aforementioned execution tracking records and process monitoring documents are integrated to generate structured maintenance behavior data. This maintenance behavior data is written to the vehicle's full lifecycle archive database in real time, updating the vehicle's maintenance record list to cover the most recent maintenance time, maintenance items, replaced parts, and costs, providing a data foundation for subsequent effectiveness verification and iterative optimization.

[0150] In one embodiment of the present invention, taking the fuel logistics vehicle numbered 001 as an example, the optimal maintenance plan stipulates that: brake system maintenance (high priority) shall be performed within 65,000 kilometers or 1 month, and routine maintenance and tire inspection shall be performed within 70,000 kilometers or 1 month.

[0151] Three days before the maintenance is due (i.e., when the mileage reaches 64,500 kilometers or the due date is approaching), a reminder message will be sent simultaneously through the driver's mobile APP and the fleet management terminal. The message will read: "Vehicle 001, brake system maintenance is due. High priority. It is recommended to complete the maintenance before September 5, 2024. Recommended institution: Partner 4S store A".

[0152] The driver scheduled maintenance at partner 4S store A. At 9:00 AM on September 5, 2024, the maintenance system recorded the start time of the maintenance; the system tracked the progress in real time and recorded the completion time of the maintenance at 10:30 AM on the same day; the executed content obtained simultaneously was: changing engine oil, oil filter, brake pads, and chassis lubrication; the replaced parts included: engine oil SN 5W-30 (4 liters), oil filter (original), and brake pads (wear-resistant type); the actual cost was 820 yuan.

[0153] The maintenance organization uploaded photos of the oil change process, before-and-after photos of brake pad replacement, and a chassis lubrication inspection report; the system of this invention remotely verified that all documents were clearly identifiable, the timestamps met the requirements, no abnormal maintenance behavior was found (such as not tightening bolts to the standard torque, or using inferior parts), and no warning was issued.

[0154] After maintenance is completed, the above records (start time, completion time, execution content, details of replaced parts, cost, and links to supporting documents) will be synchronized in real time to the full life cycle file of vehicle 001, updating the corresponding maintenance data and fault data.

[0155] This invention enables the system to achieve full traceability of the maintenance plan from reminder to completion by real-time monitoring and recording of maintenance behavior data, ensuring that maintenance items are implemented according to standards and providing a real and complete data foundation for subsequent effect verification and model iteration.

[0156] Furthermore, maintenance effectiveness is quantitatively verified based on maintenance behavior data. Maintenance effectiveness verification results refer to a data set that quantitatively evaluates the effectiveness of the maintenance plan by comparing component status, failure rate, cost-effectiveness, and user feedback before and after maintenance. This includes at least changes in component status (e.g., percentage improvement in brake sensitivity), changes in failure rate (comparison of the number of failures within a certain period before and after maintenance), cost-effectiveness indicators (cost savings per maintenance, total lifecycle cost reduction rate), and user satisfaction scores.

[0157] Specifically, the system extracts the operating parameters of core components collected by onboard sensors within a period prior to maintenance (e.g., one month before maintenance) and a period after maintenance (e.g., seven days after maintenance). For the braking system, the comparison parameters include brake response time and brake pad wear rate; for the engine, the comparison parameters include speed fluctuation variance and fuel consumption stability; for new energy vehicle batteries, the comparison parameters include SOC degradation rate. The system calculates the difference or percentage change of each parameter before and after maintenance and generates a component status verification report. If the change exceeds a preset pass threshold (e.g., brake sensitivity improvement of no less than 20%), the maintenance effect of the component is deemed qualified.

[0158] From the fault records in the entire lifecycle file, count the number of faults occurring within a fixed time window (e.g., 1 month) before maintenance, and the number of faults occurring within the same length time window after maintenance. Calculate the fault incidence rate change rate using the formula: (Number of faults before maintenance) Number of failures after maintenance / Number of failures before maintenance × 100%. This indicator is used to evaluate the effectiveness of the maintenance plan in preventing failures.

[0159] Obtain the actual cost of this maintenance, as well as the changes in repair costs and vehicle residual value (based on the residual value assessment model) caused by the maintenance over a period of time after the maintenance. Simultaneously, retrieve the estimated cost under traditional fixed-cycle maintenance from the TCO analysis model as a benchmark; calculate the cost savings rate per maintenance session: (benchmark maintenance cost) (Actual maintenance cost) / Baseline maintenance cost × 100%; Calculate the total lifecycle cost reduction rate (requires accumulating multiple maintenance and repair data); Output a cost-benefit assessment report.

[0160] Questionnaires or rating forms are pushed to drivers, fleet managers, and maintenance organizations through user interfaces (mobile apps, fleet management terminals). The collected feedback includes: the rationality of the maintenance plan (whether it is on-demand and timely), satisfaction with the maintenance results (degree of vehicle performance improvement), and ease of operation (reminder channels, appointment process). The system converts qualitative feedback into quantitative scores (e.g., 1-5 points) and stores them.

[0161] In one embodiment of the present invention, after vehicle 001 completes maintenance, an effect verification is performed: Operating parameters were collected before maintenance (August 2024) and after maintenance (September 5th to September 12th, 2024). Comparison results: Brake sensitivity improved by 30% from the pre-maintenance baseline, engine speed stability improved by 20%, tire pressure returned to normal, and chassis vibration amplitude decreased. All parameter changes exceeded the preset acceptable thresholds, indicating that the maintenance effect met expectations.

[0162] The number of malfunctions for vehicle 001 in the month prior to maintenance (August 5th to September 4th, 2024) was 1 (engine abnormal noise). The number of malfunctions in the month following maintenance (September 5th to October 4th, 2024) was 0. The malfunction rate decreased by 100%. For the entire fleet (700 vehicles): the total number of malfunctions in the month prior to maintenance was 80, and the total number of malfunctions in the month following maintenance was 15, a decrease of 81.25%.

[0163] The actual cost of a single maintenance service for vehicle 001 is 820 yuan, while the benchmark cost for traditional fixed-cycle maintenance is 1,000 yuan, resulting in an 18% reduction in cost per service. The overall fleet maintenance cost has decreased from 150,000 yuan per month to 105,000 yuan, a 30% reduction. The total lifecycle cost (based on model calculations) has decreased by 26%. Assessment conclusion: Cost-effectiveness optimization has been achieved.

[0164] Questionnaires were sent to drivers, fleet managers, and maintenance organizations. Feedback results showed that the maintenance plan met actual needs (average rating 4.8 / 5), maintenance reminders were timely (4.9 / 5), maintenance results were good (4.7 / 5), and operation was convenient (4.6 / 5). Drivers specifically reported a significant improvement in driving safety after brake system maintenance.

[0165] The above component status verification results, failure rate changes, cost-effectiveness indicators, and user feedback scores are integrated into structured maintenance effectiveness verification results for subsequent iterative optimization.

[0166] This invention, through quantitative verification of maintenance effects, enables the system to objectively evaluate the rationality and effectiveness of maintenance plans, identify deficiencies in maintenance schemes (such as the condition of components not meeting expectations after a certain maintenance), provide quantitative basis for subsequent optimization of maintenance strategy models and rule engines, and ensure continuous improvement in maintenance quality.

[0167] Step 205: Update the vehicle's multimodal profile data based on maintenance behavior data.

[0168] Specifically, a new record is added to the maintenance data, with fields including but not limited to: maintenance time, maintenance items, replaced parts (brand, model, replacement cycle), maintenance cost, maintenance agency, and maintenance personnel. If new faults or anomalies are discovered during the maintenance process (e.g., records before fault codes were cleared), the fault data is updated.

[0169] In this embodiment of the invention, the update operation employs a transaction mechanism to ensure data consistency. After the update is completed, the vehicle's multimodal profile data reflects the latest maintenance history and component status, providing an up-to-date data foundation for subsequent rule engine optimization and knowledge report generation.

[0170] In one embodiment of the present invention, vehicle 001 completed maintenance on September 5, 2024. The maintenance data includes: maintenance start time at 9:00 AM on September 5, 2024, completion time at 10:30 AM, the performed items being oil change, oil filter replacement, brake pad replacement, and chassis lubrication replacement, the replaced parts being SN 5W-30 (4 liters) engine oil, original oil filter, and wear-resistant brake pads, the actual cost being 820 yuan, the maintenance institution being the cooperating 4S store A, and the maintenance personnel being A.

[0171] Perform the following update operations: A new record is inserted into the maintenance data of vehicle 001: Maintenance ID=001-20240905, Maintenance Time=2024-09-05, Maintenance Mileage=65000 km, Maintenance Items="Change engine oil, oil filter, brake pads, chassis lubrication", Maintenance Cost=820 yuan, Maintenance Institution="Cooperating 4S Store A", Maintenance Personnel="Jia", Replaced Parts="Engine Oil (SN 5W-30, replacement mileage 65000 km), Oil Filter (Original, 65000 km), Brake Pads (Wear-resistant, 65000 km)". The updated multimodal profile data is saved to the database for subsequent steps.

[0172] This step maintains the timeliness and completeness of vehicle lifecycle information by updating multimodal profile data in real time, providing the latest training data for the continuous optimization of the rule engine and avoiding erroneous decisions based on outdated data.

[0173] Step 206: Based on the updated multimodal archive data, optimize the rule engine to obtain the updated rule engine.

[0174] In this embodiment of the invention, the optimization process of the rule engine is triggered periodically (e.g., every 24 hours, or whenever a certain number of new maintenance records are accumulated). Specifically, the maintenance records, fault records, and effect verification results newly added or updated in the most recent optimization cycle (e.g., the past 24 hours) are read; historical input data (lifespan information, operating status, driving behavior, environmental data) are re-input into the current rule engine to obtain a predicted list of maintenance needs; this predicted list is compared with the actual maintenance behavior data, the accuracy and recall rate of each rule are calculated, and rules that frequently generate false alarms (predicting unnecessary maintenance) or false negatives (not predicting but actually needing maintenance) are identified.

[0175] For rules with discrepancies, the system automatically adjusts their condition thresholds. For example, if a rule "generate brake maintenance if the remaining life of the brake system is ≤15,000 km" frequently results in missed reports (when maintenance is actually needed at 16,000 km remaining life), the threshold will be increased from 15,000 km to 18,000 km. The adjustment is based on the statistical distribution of historical errors (e.g., taking the mean error). New rules can be generated for newly added fault modes or maintenance requirement modes. For example, if association rule mining reveals a high correlation between "ambient humidity >80% and mileage >20,000 km" and "chassis corrosion," a new chassis rust prevention rule will be added.

[0176] Furthermore, the adjusted rule engine will be validated in historical data backtesting. If the accuracy and recall are not lower than the original version, it will be deployed as the updated rule engine in the production environment.

[0177] In one embodiment of the invention, the rule engine is optimized every 24 hours. Based on historical data, maintenance effectiveness data, and fault data of 700 vehicles in the fleet, the current rule engine sets the maintenance cycle for vehicles in the accelerated wear-out period to 5000 km / 3 months. By analyzing the maintenance effectiveness data, it is identified that the actual failure rate of vehicles in this stage begins to rise after 4000 km, and the condition of components improves significantly after maintenance. The maintenance cycle threshold for the accelerated wear-out period is lowered from 5000 km to 4000 km, while chassis and engine maintenance items are added as default rules.

[0178] For vehicles with aggressive driving styles (≥5 times / hour of hard braking), the original rule only required the brake system maintenance interval to be shortened to 5,000 kilometers. However, actual data shows that tire wear is also significantly accelerated under aggressive driving styles. The system has added a new rule: "If the number of hard braking times is ≥5 times / hour and the remaining tire life is ≤20,000 kilometers, the tire inspection priority is raised to high priority."

[0179] The original rule set the remaining lifespan threshold for the braking system at 15,000 kilometers. However, by comparing actual replacement records, it was found that when the predicted remaining lifespan was 15,000 kilometers, the brake pads had already worn to their limit after only 12,000 kilometers of actual driving. The system has since adjusted the threshold to 13,000 kilometers.

[0180] After making the above rule adjustments, backtesting was conducted on historical datasets: the accuracy of maintenance demand prediction improved from 96% to 97%. The updated rule engine was then deployed and went live.

[0181] Through continuous optimization of the rule engine, this invention generates a more accurate maintenance requirement list, which can adapt to individual vehicle differences and dynamic changes, reduce false alarms and missed alarms, and further improve the rationality of maintenance plans.

[0182] Step 207: Perform statistical analysis on the multimodal archive data, optimal maintenance plan and maintenance behavior data of vehicles in the deterioration and scrapping period throughout their entire life cycle, and generate a knowledge report.

[0183] In this embodiment of the invention, the knowledge report includes selection recommendations based on vehicle scenarios and user preferences. Specifically, the knowledge report is generated through multi-dimensional statistical analysis of the full lifecycle multimodal archive data of vehicles in their decline and scrapping period, the optimal maintenance plans throughout the entire lifecycle, and actual maintenance behavior data. The knowledge report includes, but is not limited to: component failure patterns under different vehicle scenarios and user preferences (such as the probability of early failure of specific components due to certain driving habits), maintenance cost-benefit analysis (the impact of different maintenance strategies on the total lifecycle cost and residual value), and quantitative recommendations for future vehicle selection (such as recommending the purchase of wear-resistant and easy-to-maintain vehicle configurations).

[0184] Specifically, all vehicles currently in the "decline and scrapping period" of their life cycle are identified periodically (e.g., monthly or quarterly), along with their multimodal profile data, optimal maintenance plans, and maintenance behavior data. Statistical analysis is then performed on this data regarding component wear and failure patterns, and the cost-effectiveness of maintenance strategies, to generate selection recommendations.

[0185] The statistical analysis of component wear and failure patterns includes: statistically analyzing the average service life, main failure modes, and early warning signs of each key component (engine, transmission, braking system, tires, battery, etc.) according to vehicle type (fuel / new energy), usage scenario (urban / intercity / rough roads), driving style (aggressive / stable).

[0186] Statistical analysis of the cost-benefit of maintenance strategies specifically includes comparing the total life-cycle cost and residual value changes under different maintenance strategies (such as different maintenance cycles and different component replacement standards) and calculating the benefit ratio of the optimal strategy.

[0187] Based on the above analysis, quantitative recommendations are provided. For example: "For high-frequency logistics scenarios in cities, we recommend purchasing model X, whose braking system life is 20% longer than that of model Y." "For aggressive driving styles, we recommend standard equipment with wear-resistant brake pads, which can reduce the total life cycle cost by 15%."

[0188] As an optional approach, the statistical analysis results can be organized into tables, charts, and text descriptions to generate a knowledge report in PDF or HTML format, which can then be stored in a knowledge base for vehicle selection decision-makers to access and use.

[0189] In one embodiment of the present invention, 20 vehicles in the fleet are entering their deterioration and scrapping period. The full life cycle data of these vehicles are extracted and statistically analyzed. The 20 vehicles were evaluated, and it was found that 15 of them had severe wear and tear on core components (engine and transmission), and the repair cost exceeded the residual value of the vehicles. They were determined to have reached the reasonable service life and were recommended to be scrapped and dismantled. The remaining 5 vehicles had good component condition and were recommended to be transferred to the used car market.

[0190] Based on complete maintenance records and good component condition, a residual value assessment was conducted on the five vehicles to be transferred. The assessment result was 15% higher than that of traditional methods (for example, the traditional assessment residual value was 10,000 yuan, while the system assessment was 11,500 yuan), helping the fleet to maximize asset value.

[0191] By summarizing the engine wear patterns of all scrapped vehicles, it was found that a certain engine model had a high rate of abnormal camshaft wear after 180,000 kilometers. Meanwhile, maintenance data showed that changing the engine oil every 5,000 kilometers significantly reduced the incidence of this failure. Based on this, the system generated a selection recommendation in a knowledge report: "When purchasing fuel-powered logistics vehicles in the future, it is recommended to prioritize engine model Y (with a 20% lower wear rate), or to mandate a 5,000-kilometer oil change strategy for existing models, which is expected to reduce the total life-cycle maintenance cost of each vehicle by 3,000 yuan." Furthermore, for new energy vehicles, the analysis showed that vehicles with a battery SOC degradation exceeding 20% ​​experienced a 35% decrease in residual value; it is recommended to prioritize models with a battery cycle life of ≥3,000 cycles when selecting a vehicle.

[0192] The above analysis results are used to generate a knowledge report, which is provided to fleet managers for procurement decisions on the next batch of vehicles. Practical application statistics show that this invention has achieved significant operational results after being implemented in the logistics fleet. The average monthly vehicle failure rate decreased from 80 to 15, a reduction of 81.25%, effectively solving the problem of frequent failures under the traditional maintenance model; monthly maintenance costs decreased from 150,000 yuan to 105,000 yuan, a reduction of 30%, and the total life-cycle cost decreased by 26%; the average vehicle lifespan increased by 22%, reducing vehicle replacement frequency and lowering purchase costs. The accuracy rate of core component life prediction reached 97.8%, enabling early identification of potential problems and truly achieving "maintenance instead of repair," avoiding losses caused by sudden failures. Closed-loop management throughout the entire process ensures the traceability and evaluability of maintenance results, improving fleet operation and maintenance efficiency by 45% and reducing the workload of management personnel by 60%. Furthermore, data feedback optimizes vehicle selection, further reducing subsequent operation and maintenance costs. This system is fully adapted to the actual needs of high-frequency use and multi-scenario operation of logistics fleets, and can be easily promoted to other logistics fleets and public transportation vehicles and other vehicle operation scenarios, with extremely high industrial applicability and promotion value.

[0193] This invention utilizes comprehensive data mining of vehicles in their decline and scrapping stages. The system transforms the aging patterns of components, maintenance effectiveness, and cost data from practice into quantifiable selection knowledge, which is then fed back into the vehicle planning and selection stage. This forms a closed-loop system covering the entire lifecycle of "selection-operation-maintenance-disposal-selection," helping users reduce the total lifecycle cost of their fleet and improve asset operation efficiency.

[0194] This invention, centered on full lifecycle management, achieves precise, personalized, and closed-loop preventative maintenance, comprehensively meeting the maintenance needs of vehicles at each stage. Specifically, it covers the entire vehicle lifecycle, from planning and selection, procurement and access, operation and use, maintenance and repair to disposal and scrapping. It develops targeted maintenance strategies based on the characteristics of each stage, achieving "prevention before the event, control during the event, and optimization after the event," effectively reducing vehicle failure rates by over 80%. Through deep integration of multi-source data, it breaks down data silos and utilizes intelligent algorithms to mine data correlations, ensuring a core component lifespan prediction accuracy of no less than 97% and a maintenance demand prediction accuracy of no less than 96%, identifying aging trends and potential hazards in advance, truly achieving "maintenance instead of repair." Simultaneously, it develops and dynamically adjusts differentiated maintenance plans based on vehicle type, usage scenario, driving style, and usage environment, avoiding over-maintenance or under-maintenance, reducing maintenance costs by over 30%, extending vehicle lifespan by over 20%, and clearly distinguishing between DIY and professional maintenance items to further optimize costs. It also constructs a "data collection..." analyze plan implement verify optimization This invention employs a closed-loop management system for the entire "disposal" process, ensuring that maintenance results are traceable and assessable, and that maintenance plans continuously adapt to the actual needs of vehicles. Through full lifecycle cost analysis, it optimizes maintenance strategies to balance effectiveness and cost, achieving the lowest total cost of ownership (TCO) and increasing vehicle residual value. Furthermore, this invention is compatible with various vehicle types, including fuel-powered vehicles and new energy vehicles, and can be applied to various scenarios such as personal vehicles, ride-hailing vehicles, logistics fleets, and public transportation groups. It requires no large-scale modification of existing architecture, resulting in low deployment costs and convenient operation. Moreover, through iterative optimization modules, it continuously improves algorithm accuracy and the rationality of maintenance strategies, adapting to new vehicle models, components, and industry technology iterations, and feeds disposal data back into the selection process, forming a virtuous cycle throughout the entire lifecycle.

[0195] It is worth noting that the acquisition, storage, use, and processing of data in the technical solution of this application all comply with relevant laws and regulations. The user information in the embodiments of this application was obtained through legal and compliant means, and the acquisition, storage, use, and processing of user information have been authorized and agreed upon by the client.

[0196] It is worth noting that the information collected in this application is information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data all comply with the relevant laws, regulations and standards of the relevant countries and regions, necessary confidentiality measures have been taken, and they do not violate public order and good morals. Corresponding operation portals are provided for users to choose to authorize or refuse.

[0197] It is worth noting that the technical solution provided in this application provides users with a corresponding operation entry point, allowing users to choose to agree to or reject the automated decision-making result; if the user chooses to reject, the process will proceed to the expert decision-making process.

[0198] The technical solution of the vehicle preventive maintenance control method based on the whole life cycle provided in this invention embodiment identifies the current life cycle of the vehicle based on the collected multimodal archive data of the vehicle; through a pre-built maintenance strategy model, the optimal maintenance plan is generated according to the current life cycle and multimodal archive data, and the vehicle's maintenance behavior data is monitored; the maintenance strategy model is iteratively optimized based on the maintenance behavior data, and a knowledge report for guiding vehicle selection is generated; through multimodal fusion of LSTM-Transformer architecture and attention mechanism, the prediction accuracy and feature representation capability of the remaining life of key components are significantly improved; the maintenance behavior data driven by closed-loop monitoring drives model iteration, realizing the adaptive and rapid convergence of the maintenance strategy model; at the same time, combined with whole life cycle cost analysis and rule engine, the computational complexity of maintenance decision-making under complex working conditions is reduced, and a high-dimensional knowledge report that can guide vehicle selection is generated.

[0199] Figure 4 This is a schematic diagram of a vehicle preventive maintenance control device based on the entire life cycle, provided in an embodiment of the present invention. This device is used to execute the aforementioned vehicle preventive maintenance control method based on the entire life cycle, such as... Figure 4 As shown, the device includes: a life cycle identification unit 11, a maintenance plan generation unit 12, and a maintenance strategy optimization unit 13.

[0200] The lifecycle identification unit 11 is used to identify the current lifecycle of a vehicle based on the collected multimodal vehicle profile data.

[0201] The maintenance plan generation unit 12 is used to generate the optimal maintenance plan based on the current life cycle and multimodal archive data through a pre-built maintenance strategy model, and to monitor the vehicle's maintenance behavior data.

[0202] The maintenance strategy optimization unit 13 is used to iteratively optimize the maintenance strategy model based on maintenance behavior data and generate a knowledge report to guide vehicle selection.

[0203] In this embodiment of the invention, the multimodal archive data includes mileage and service life; the life cycle identification unit 11 is specifically used to determine the current life cycle of the vehicle as the deterioration and scrapping period if the mileage is greater than or equal to a preset first mileage threshold, or if the service life is greater than or equal to a preset first year threshold; if the mileage is greater than or equal to a preset second mileage threshold, or if the service life is greater than or equal to a preset second year threshold, the current life cycle of the vehicle is determined as the accelerated wear period; if the mileage is greater than or equal to a preset third mileage threshold, or if the service life is greater than or equal to a preset third year threshold, the current life cycle of the vehicle is determined as the stable operation period; if the mileage is less than the third mileage threshold and the service life is less than the third year threshold, the current life cycle of the vehicle is determined as the break-in period.

[0204] In this embodiment of the invention, the maintenance plan generation unit 12 is specifically used to predict the lifespan information of key components based on the current lifespan and multimodal archive data using a pre-built lifespan prediction model. The lifespan prediction model is trained based on a long short-term memory network and a Transformer architecture. Based on the maintenance strategy model, the optimal maintenance plan is generated based on the lifespan information and multimodal archive data.

[0205] In this embodiment of the invention, the maintenance plan generation unit 12 is specifically used to perform data fusion on multimodal archive data through a multimodal fusion algorithm based on an attention mechanism to obtain a fused feature vector; and to perform life prediction on key components based on the fused feature vector and the current life cycle through a life prediction model to generate life information of key components.

[0206] In this embodiment of the invention, the multimodal archive data includes operational status data, driving behavior data, and environmental data; the maintenance plan generation unit 12 is specifically used to generate a maintenance requirement list based on a rule engine, according to lifespan information, operational status data, driving behavior data, and environmental data; through a pre-built full lifecycle cost analysis model, to optimize costs based on the maintenance requirement list and generate an optimal maintenance list; and based on personalized maintenance standards and basic maintenance standards, to dynamically adjust according to the optimal maintenance list and generate an optimal maintenance plan, whereby personalized maintenance standards include vehicle scenarios and user preferences.

[0207] In this embodiment of the invention, the maintenance strategy optimization unit 13 is specifically used to update the multimodal profile data of the vehicle based on the maintenance behavior data; optimize the rule engine based on the updated multimodal profile data to obtain the updated rule engine; and perform statistical analysis on the multimodal profile data, optimal maintenance plan and maintenance behavior data of the entire life cycle of the vehicle in the deterioration and scrapping period to generate a knowledge report, which includes selection suggestions based on vehicle scenarios and user preferences.

[0208] In the solution of this invention embodiment, the current life cycle of the vehicle is identified based on the collected multimodal vehicle profile data; an optimal maintenance plan is generated based on the current life cycle and multimodal profile data through a pre-built maintenance strategy model, and the vehicle's maintenance behavior data is monitored; the maintenance strategy model is iteratively optimized based on the maintenance behavior data, and a knowledge report for guiding vehicle selection is generated; through multimodal fusion of LSTM-Transformer architecture and attention mechanism, the prediction accuracy and feature representation capability of the remaining life of key components are significantly improved; the model iteration driven by maintenance behavior data based on closed-loop monitoring achieves adaptive and rapid convergence of the maintenance strategy model; at the same time, combined with full life cycle cost analysis and rule engine, the computational complexity of maintenance decisions under complex working conditions is reduced, and a high-dimensional knowledge report that can guide vehicle selection is generated.

[0209] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer device, specifically, a computer device can be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or any combination of these devices.

[0210] This invention provides a computer device including a memory and a processor. The memory stores information including program instructions, and the processor controls the execution of the program instructions. When the program instructions are loaded and executed by the processor, they implement the steps of the above-described embodiment of the vehicle preventive maintenance control method based on the entire life cycle. For a detailed description, please refer to the above-described embodiment of the vehicle preventive maintenance control method based on the entire life cycle.

[0211] The following is for reference. Figure 5 It shows a schematic diagram of the structure of a computer device 600 suitable for implementing the embodiments of this application.

[0212] like Figure 5 As shown, the computer device 600 includes a central processing unit (CPU) 601, which can perform various appropriate tasks and processes based on programs stored in read-only memory (ROM) 602 or programs loaded from storage section 608 into random access memory (RAM) 603. The RAM 603 also stores various programs and data required for the operation of the computer device 600. The CPU 601, ROM 602, and RAM 603 are interconnected via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0213] The following components are connected to I / O interface 605: an input section 606 including a keyboard, mouse, etc.; an output section 607 including a cathode ray tube (CRT), liquid crystal feedback (LCD), etc., and speakers, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card such as a LAN card, modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to I / O interface 605 as needed. A removable medium 611, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on drive 610 as needed so that computer programs read from it can be installed in storage section 608 as needed.

[0214] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program tangibly embodied on a machine-readable medium, the computer program including program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 609, and / or installed from removable medium 611.

[0215] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0216] For ease of description, the above devices are described separately by function as various units. Of course, in implementing this application, the functions of each unit can be implemented in one or more software and / or hardware.

[0217] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0218] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0219] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0220] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0221] The acquisition, storage, use, and processing of data in this application all comply with the relevant provisions of national laws and regulations.

[0222] It should be noted that in the embodiments of this application, certain software, components, models and other existing solutions in the industry may be mentioned. These should be regarded as exemplary and are only intended to illustrate the feasibility of implementing the technical solution of this application. However, it does not mean that the applicant has used or necessarily used the solution.

[0223] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0224] This application can be described in the general context of computer-executable instructions, such as program modules, that are executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0225] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0226] The above description is merely an embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of this application should be included within the scope of the claims of this application.

Claims

1. A vehicle preventive maintenance control method based on the entire life cycle, characterized in that, The method includes: Based on the collected multimodal vehicle profile data, the current lifecycle of the vehicle is identified; By using a pre-built maintenance strategy model, an optimal maintenance plan is generated based on the current life cycle and multimodal profile data, and the vehicle's maintenance behavior data is monitored. The maintenance strategy model is iteratively optimized based on the maintenance behavior data, and a knowledge report is generated to guide vehicle selection.

2. The vehicle preventive maintenance control method based on the entire life cycle as described in claim 1, characterized in that, The multimodal archive data includes mileage and years of use; The identification of the vehicle's current lifecycle based on the collected multimodal vehicle profile data includes: If the mileage is greater than or equal to a preset first mileage threshold, or if the service life is greater than or equal to a preset first year threshold, the current life cycle of the vehicle is determined as the deterioration and scrapping period. If the mileage is greater than or equal to a preset second mileage threshold, or if the service life is greater than or equal to a preset second service life threshold, the current life cycle of the vehicle is determined as the period of accelerated wear and tear. If the mileage is greater than or equal to a preset third mileage threshold, or if the service life is greater than or equal to a preset third year threshold, the current life cycle of the vehicle is determined as a stable operating period. If the mileage is less than the third mileage threshold and the service life is less than the third service life threshold, the current life cycle of the vehicle is determined as the break-in period.

3. The vehicle preventive maintenance control method based on the entire life cycle as described in claim 1, characterized in that, The process of generating an optimal maintenance plan based on the current lifecycle and multimodal profile data using a pre-built maintenance strategy model includes: Using a pre-built lifespan prediction model, the lifespan information of key components is predicted based on the current lifespan and multimodal archive data. The lifespan prediction model is trained based on a long short-term memory network and a Transformer architecture. Based on the maintenance strategy model, an optimal maintenance plan is generated according to the lifespan information and multimodal archive data.

4. The vehicle preventive maintenance control method based on the entire life cycle as described in claim 3, characterized in that, The method of predicting the lifespan information of key components using a pre-built lifespan prediction model, based on the current lifespan and multimodal profile data, includes: The multimodal archive data is fused using an attention-based multimodal fusion algorithm to obtain a fused feature vector; The lifetime prediction model is used to predict the lifetime of key components based on the fused feature vector and the current life cycle, thereby generating lifetime information for the key components.

5. The vehicle preventive maintenance control method based on the entire life cycle according to claim 3, characterized in that, The multimodal archive data includes operational status data, driving behavior data, and environmental data; The optimal maintenance plan is generated based on the maintenance strategy model, according to the lifespan information and multimodal profile data, including: Based on the rule engine, a maintenance requirement list is generated according to the lifespan information, operating status data, driving behavior data, and environmental data. By using a pre-built full life cycle cost analysis model, cost optimization is performed based on the maintenance requirement list to generate an optimal maintenance list; Based on personalized maintenance standards and basic maintenance standards, the optimal maintenance plan is generated by dynamically adjusting according to the optimal maintenance list. The personalized maintenance standards include vehicle scenarios and user preferences.

6. The vehicle preventive maintenance control method based on the entire life cycle according to claim 5, characterized in that, The step of iteratively optimizing the maintenance strategy model based on the maintenance behavior data and generating a knowledge report to guide vehicle selection includes: Update the vehicle's multimodal profile data based on the maintenance behavior data; Based on the updated multimodal archive data, the rule engine is optimized to obtain an updated rule engine; Statistical analysis is performed on the multimodal archive data, optimal maintenance plan, and maintenance behavior data of vehicles in the deterioration and scrapping period throughout their entire life cycle to generate the knowledge report, which includes selection recommendations based on vehicle scenarios and user preferences.

7. A vehicle preventive maintenance control device based on the entire life cycle, characterized in that, The device includes: The lifecycle identification unit is used to identify the current lifecycle of the vehicle based on the collected multimodal vehicle profile data. The maintenance plan generation unit is used to generate the optimal maintenance plan based on the current life cycle and multimodal archive data through a pre-built maintenance strategy model, and to monitor the vehicle's maintenance behavior data. The maintenance strategy optimization unit is used to iteratively optimize the maintenance strategy model based on the maintenance behavior data and generate a knowledge report to guide vehicle selection.

8. A computer-readable medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the vehicle preventive maintenance control method based on the entire life cycle as described in any one of claims 1 to 6.

9. A computer device comprising a memory and a processor, the memory for storing information including program instructions, and the processor for controlling the execution of the program instructions, characterized in that, When the program instructions are loaded and executed by the processor, they implement the vehicle preventive maintenance control method based on the entire life cycle as described in any one of claims 1 to 6.

10. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the vehicle preventive maintenance control method based on the entire life cycle as described in any one of claims 1 to 6.

Images