OTA upgrade task dynamic configuration control method and system based on machine learning

Abstract

The invention discloses an OTA upgrade task dynamic configuration control method and system based on machine learning, and the method comprises the steps: collecting vehicle upgrade data, and carrying out the preprocessing of the vehicle upgrade data; designing a prediction model, carrying out training and effect evaluation on the model, and outputting the upgrade probability of each vehicle in a future time period and a recommended upgrade time window according to the preprocessed vehicle upgrade data; and carrying out OTA upgrade task allocation according to the output of the prediction model to complete vehicle upgrade. According to the method, the vehicle upgrading data are input into the prediction model for prediction, the predicted upgrading time window is obtained, OTA upgrading task allocation is performed, the vehicle upgrading demand can be predicted, the upgrading time window can be recommended, overlarge system load and current limitation caused by centralized task issuing are avoided, and the reliability, stability and success rate of system upgrading are improved; and a greedy algorithm is added in OTA upgrade task allocation, so that the allocation scheduling of the upgrade tasks is optimized, and the network and system loads are balanced to the maximum extent.

Description

Technical Field

[0001] The present invention relates to the field of automotive electronics technology, and particularly to a method and system for dynamically configuring and controlling OTA upgrade tasks based on machine learning. Background Art

[0002] OTA refers to an automotive remote upgrade technology that remotely manages firmware, data, and applications on automotive component terminals through a mobile communication network. With the advancement of the "new four modernizations" of automobiles, namely "electrification, networking, intelligence, and sharing", "software-defined vehicles" has become the development trend of the automotive industry. Under the new situation, higher requirements are put forward for the repair of software faults and the update of personalized customization needs. The vehicle OTA technology can quickly deliver software to users, reduce the upgrade costs of automotive manufacturers and users, and improve the user experience.

[0003] The vehicle OTA architecture includes three parts: the OTA cloud, the OTA engine, and the OTA terminal. Among them, the OTA cloud is mainly responsible for vehicle information management, upgrade package transfer and management, upgrade strategy configuration, upgrade task management, and upgrade data statistics; the OTA engine is mainly responsible for overall management of upgrade tasks, coordination of upgrade tasks, etc., and the OTA terminal mainly refers to the upgrade target controller; the OTA engine is the bridge connecting the OTA cloud and the OTA terminal. And the vehicle OTA success rate is an important indicator to measure the stability and robustness of the OTA system.

[0004] A method and system for dynamically configuring OTA upgrade tasks based on machine learning proposed in this patent, through training on vehicle information, network status, and historical upgrade data, further predicts vehicle upgrade requirements and recommends an upgrade time window to avoid centralized issuance of upgrade tasks, solves the problem of current throttling caused by excessive load on the OTA cloud system due to centralized issuance of upgrade tasks, which affects the upgrade success rate, and improves the reliability and success rate of system upgrades. Summary of the Invention

[0005] To solve the problem in the prior art that the centralized issuance of OTA upgrade tasks leads to excessive system load, thereby affecting stability and success rate, the present invention provides a method and system for dynamically configuring and controlling OTA upgrade tasks based on machine learning.

[0006] For this reason, the technical solution adopted by the present invention is as follows:

[0007] The present invention provides a method for dynamically configuring and controlling OTA upgrade tasks based on machine learning, and the method includes:

[0008] Collect vehicle upgrade data and preprocess the vehicle upgrade data;

[0009] Design a prediction model, train and evaluate the effectiveness of the prediction model, and output the upgrade probability and recommended upgrade time window of each vehicle predicted by the prediction model within a specific future time period based on the preprocessed vehicle upgrade data;

[0010] Perform OTA upgrade task allocation according to the output of the prediction model to complete vehicle upgrades.

[0011] According to the above solution, the vehicle upgrade data includes the number of target upgrade vehicles, the time range during which the OTA task is valid, the number of vehicles compatible with system upgrades, vehicle information, network status information, historical upgrade data, and user behavior patterns.

[0012] According to the above solution, the preprocessing of vehicle upgrade data includes:

[0013] Clean the vehicle upgrade data, dealing with missing values, outliers, and duplicate records;

[0014] Extract or construct features helpful for the prediction task from the cleaned vehicle upgrade data;

[0015] Convert the cleaned vehicle upgrade data into a time series format and normalize the features.

[0016] According to the above solution, designing a prediction model includes: selecting LSTM or GRU as the basic model of the prediction model, designing the number of network layers, the number of units in each layer, and the activation function of the prediction model, defining the loss function of the prediction model, and selecting the optimizer of the prediction model.

[0017] According to the above solution, the loss function includes the mean squared error function or the root mean squared error function.

[0018] According to the above solution, OTA upgrade tasks are constrained by constraint conditions, and the constraint conditions include:

[0019] Each vehicle can only be upgraded once;

[0020] The number of vehicles upgraded simultaneously cannot exceed the upper limit compatible with the system;

[0021] The vehicle upgrade time is within the recommended upgrade time window.

[0022] According to the above solution, OTA upgrade task allocation is carried out by constructing a greedy algorithm, including: initializing the upgrade queue, sorting in descending order according to the upgrade probability of each vehicle within a specific future time period to form a vehicle sequence;

[0023] Traverse the vehicle sequence, obtain the upgrade task queue according to the upgrade time window of each vehicle, and complete the task allocation.

[0024] According to the above solution, the upgrade task queue is obtained based on the upgrade time window of each vehicle as follows: for each vehicle, if the number of vehicles in the current upgrade queue plus this vehicle does not exceed the upper limit compatible with the system, and the total upgrade time of the current upgrade queue plus the vehicle upgrade time of this vehicle does not exceed the valid time range of the OTA task, then this vehicle is added to the upgrade task queue.

[0025] The present invention also provides a dynamic configuration control system for OTA upgrade tasks based on machine learning, and the system includes:

[0026] A data processing module, configured to collect vehicle upgrade data and preprocess the vehicle upgrade data;

[0027] A prediction module, configured to design a prediction model, train and evaluate the effect of the prediction model, and output the upgrade probability and recommended upgrade time window of each vehicle predicted by the prediction model within a specific future time period according to the preprocessed vehicle upgrade data;

[0028] An upgrade module, configured to perform OTA upgrade task allocation according to the output of the prediction model to complete vehicle upgrade.

[0029] The present invention also provides a computer storage medium, which stores a computer program executable by a processor, and the computer program executes a dynamic configuration control method for OTA upgrade tasks based on machine learning as described above in the claims.

[0030] The beneficial effects produced by the present invention are:

[0031] By inputting vehicle upgrade data into the prediction model for prediction, obtaining the predicted upgrade time window, and further performing OTA upgrade task allocation, the present invention can predict vehicle upgrade requirements and recommend upgrade time windows to avoid excessive system load and current limiting caused by centralized issuance of upgrade tasks, effectively improving the reliability, stability, and success rate of system upgrades.

[0032] Furthermore, by adding a greedy algorithm to the OTA upgrade task allocation, the present invention further optimizes the allocation and scheduling of upgrade tasks, achieving maximized balance of network and system load. Description of the Drawings

[0033] Figure 1 is a schematic structural diagram of a dynamic configuration control method for OTA upgrade tasks based on machine learning according to an embodiment of the present invention;

[0034] Figure 2 is a schematic system module diagram of a dynamic configuration control system for OTA upgrade tasks based on machine learning according to an embodiment of the present invention. Detailed Embodiments

[0035] In order to make the objectives, technical solutions and advantages of the present invention more clear and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0036] Embodiment 1

[0037] In the prior art, the centralized distribution of OTA upgrade tasks can lead to an excessive system load, and the server is prone to traffic limiting, resulting in the OTA gateway being melted down or the microservices being restarted. To improve the user experience and enhance the stability and success rate of OTA upgrades, an embodiment of the present invention provides a dynamic configuration control method for OTA upgrade tasks based on machine learning, as Figure 1 shown, the control method includes:

[0038] S1. Collect vehicle upgrade data and preprocess the vehicle upgrade data.

[0039] Specifically, the collected vehicle upgrade data includes the number of target upgrade vehicles N, the effective time range T of the OTA task, the number of vehicles M that the system can compatibly upgrade, vehicle information, network status information, historical upgrade data, and user behavior patterns.

[0040] In a preferred embodiment, the vehicle information includes the vehicle model, service life, mileage, software and hardware configuration; the network status information includes the vehicle's network connection quality and data transmission rate; the historical upgrade data includes the success rate, required time, failure reason, and upgrade type of past upgrades; the user behavior patterns include the vehicle's usage time, usage frequency, and driving route.

[0041] Specifically, preprocessing the vehicle upgrade data includes: performing data cleaning on the vehicle upgrade data to handle missing values, outliers, and duplicate records; extracting or constructing features from the vehicle upgrade data that are helpful for the prediction task; converting the vehicle upgrade data into a time series format to prepare for inputting into the prediction model, and performing normalization processing on the features to accelerate the running speed of the prediction model and improve the model performance.

[0042] S2. Design a prediction model, train and evaluate the prediction model, and output the upgrade probability and recommended upgrade time window predicted by the prediction model for each vehicle within a specific future time period according to the preprocessed vehicle upgrade data.

[0043] Among them, designing the prediction model specifically includes: selecting LSTM or GRU as the basic model of the prediction model; determining the network structure, designing the number of network layers, the number of units in each layer, and the activation function of the prediction model; defining the loss function of the prediction model; selecting the optimizer of the prediction model.

[0044] In a preferred embodiment of the present invention, the mean squared error (MSE) or root mean squared error (RMSE) is used as the loss function for the regression task; Adam or SGD is used as the optimizer for the prediction model to update the network weights.

[0045] Specifically, during the training of the prediction model, the dataset used for training is divided into a training set, a validation set, and a test set; the model is trained using the training set data, and at the same time, the model performance is monitored on the validation set to avoid overfitting; and hyperparameters such as the learning rate, batch size, and number of training epochs are adjusted according to the performance on the validation set; after training, the prediction accuracy of the model is evaluated using evaluation metrics such as MAE or RMSE in the test set; the errors in the model predictions are analyzed to identify possible improvement directions.

[0046] The preprocessed vehicle upgrade data is input into the trained prediction model, and the probability or label P indicating whether each vehicle needs to be upgraded within a specific future time period is output i , and according to the prediction results and possible business logics, considering cost - effectiveness and vehicle performance, the optimal upgrade time window t is recommended i .

[0047] S3. Perform OTA upgrade task allocation according to the output of the prediction model to complete vehicle upgrades.

[0048] According to the upgrade probability P of each vehicle within a specific future time period output by the prediction model i and the recommended upgrade time window t i allocate OTA upgrade tasks. Among them, the goal of the OTA upgrade task is to maximize the number of upgraded vehicles while ensuring that all upgrades are completed within the valid time range T of the OTA task; and the OTA upgrade task is constrained by constraint conditions. Among them, the constraint conditions include: each vehicle can only be upgraded once; the number of vehicles upgraded simultaneously cannot exceed the upper limit compatible with the system; the vehicle upgrade time for each vehicle is within the recommended upgrade time window.

[0049] Specifically, a binary variable U is defined i , if the i - th vehicle is upgraded within the valid time range T of the OTA task, then U i =1, otherwise U i =0. Then the goal of the OTA upgrade task can be expressed as:

[0050]

[0051] Then the constraint conditions can be expressed as: U i ∈(0, 1);

[0052] After the upgrade task allocation is completed, execute the upgrade task and monitor the upgrade process in real time. At the same time, collect the execution results and monitoring data to facilitate providing historical data and optimization goals for the next upgrade task.

[0053] In addition, the embodiment of the present invention also provides a dynamic configuration control system for OTA upgrade tasks based on machine learning, which is used for the dynamic configuration control method of OTA upgrade tasks based on machine learning described above in this embodiment, as Figure 2 shown. The system mainly includes:

[0054] A data processing module, which is used to collect vehicle upgrade data and preprocess the vehicle upgrade data;

[0055] A prediction module, which is used to design a prediction model, train and evaluate the effect of the prediction model, and output the upgrade probability and recommended upgrade time window of each vehicle predicted by the prediction model within a specific future time period according to the preprocessed vehicle upgrade data;

[0056] An upgrade module, which is used to allocate OTA upgrade tasks according to the output of the prediction model to complete vehicle upgrades.

[0057] The present invention also provides a computer storage medium, which stores a computer program executable by a processor, and this computer program executes a dynamic configuration control method of OTA upgrade tasks based on machine learning described above in this embodiment.

[0058] In the embodiment of the present invention, by inputting vehicle upgrade data into the prediction model for prediction, obtaining the predicted upgrade time window, and further allocating OTA upgrade tasks, it is possible to predict vehicle upgrade requirements and recommend upgrade time windows to avoid excessive system load and current limiting caused by centralized issuance of upgrade tasks, effectively improving the reliability, stability, and success rate of system upgrades.

[0059] Embodiment 2

[0060] Based on Embodiment 1, the embodiment of the present invention provides a dynamic configuration control method of OTA upgrade tasks based on machine learning. The difference is that: based on the constraint conditions in step S3 of Embodiment 1, in the embodiment of the present invention, a simple greedy algorithm is constructed to allocate upgrade tasks. This algorithm is a heuristic solution that attempts to maximize the number of upgraded vehicles on the premise of meeting time constraints and system compatibility limitations, as specifically shown below:

[0061] Initialize the upgrade queue, sort it in descending order according to the upgrade probability of each vehicle within a specific future time period to form a vehicle sequence; traverse the vehicle sequence, obtain the upgrade task queue according to the upgrade time window of each vehicle, and complete the upgrade task allocation.

[0062] Specifically, for each vehicle, if the number of vehicles in the current upgrade queue plus this vehicle does not exceed the upper limit M compatible with the system, and the total upgrade time of the current upgrade queue plus the vehicle upgrade time of this vehicle does not exceed the valid time range T of the OTA task, then this vehicle is added to the upgrade task queue.

[0063] The specific greedy selection process is as follows:

[0064] Set the greedy function G, which mainly accepts the current vehicle status array U, the vehicle upgrade probability array P, the currently considered recommended upgrade time window t i and the maximum number of upgradeable vehicles M supported by the system, the number of vehicles to be upgraded N:

[0065]

[0066] Among them, U ′ is the updated status array, which is generated by the following steps:

[0067]

[0068] Among them, Top M (P) represents the index set of the top M vehicles selected from the probability array P from high to low.

[0069] All vehicles added to the upgrade task queue are the vehicles that need to complete the upgrade within the valid time range T of the OTA task.

[0070] Furthermore, the embodiment of the present invention further optimizes the allocation and scheduling of the upgrade task by adding a greedy algorithm to the OTA upgrade task allocation, and realizes the maximization of balancing the network and system load.

[0071] It should be noted that according to the needs of implementation, each step / component described in this application can be split into more steps / components, or two or more steps / components or partial operations of steps / components can be combined into new steps / components to achieve the purpose of the present invention.

[0072] The sequence numbers of the above steps in the above embodiments do not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of this application.

[0073] It should be understood that those of ordinary skill in the art can make improvements or transformations according to the above description, and all such improvements and transformations should fall within the protection scope of the appended claims of the present invention.

Claims

1. A method for dynamic configuration control of OTA upgrade tasks based on machine learning, characterized in that: The method comprises: Collect vehicle upgrade data and pre-process the vehicle upgrade data; Design a prediction model, train and evaluate the prediction model, and output the upgrade probability and recommended upgrade time window of each vehicle in a specific future time period predicted by the prediction model based on the preprocessed vehicle upgrade data; OTA upgrade tasks are assigned based on the output of the prediction model to complete the vehicle upgrade.

2. The method for dynamic configuration control of OTA upgrade tasks based on machine learning according to claim 1, characterized in that: The vehicle upgrade data includes the number of target upgrade vehicles, the effective time range of the OTA task, the number of vehicles compatible with the system upgrade, vehicle information, network status information, historical upgrade data and user behavior patterns.

3. The method for dynamic configuration control of OTA upgrade tasks based on machine learning according to claim 1, characterized in that: Preprocessing of vehicle upgrade data includes: Perform data cleaning on vehicle upgrade data to handle missing values, outliers and duplicate records; Extract or construct features that are helpful for prediction tasks from the cleaned vehicle upgrade data; Convert the cleaned vehicle upgrade data into time series format and normalize the features.

4. The method for dynamic configuration control of OTA upgrade tasks based on machine learning according to claim 1, characterized in that: Designing a prediction model includes: selecting LSTM or GRU as the basic model of the prediction model, designing the number of network layers, the number of units in each layer and the activation function of the prediction model, defining the loss function of the prediction model, and selecting the optimizer of the prediction model.

5. The method for dynamic configuration control of OTA upgrade tasks based on machine learning according to claim 4 is characterized in that: The loss function includes a mean square error function or a root mean square error function.

6. The method for dynamic configuration control of OTA upgrade tasks based on machine learning according to claim 1, characterized in that: The OTA upgrade task is constrained by constraints, which include: Each vehicle can only be upgraded once; The number of vehicles upgraded at the same time cannot exceed the system compatible limit; The vehicle upgrade time is within the recommended upgrade time window.

7. The method for dynamic configuration control of OTA upgrade tasks based on machine learning according to claim 1, characterized in that: The OTA upgrade task allocation is performed by constructing a greedy algorithm, including: initializing the upgrade queue, sorting each vehicle from high to low according to the upgrade probability in a specific time period in the future, and forming a vehicle sequence; Traverse the vehicle sequence, obtain the upgrade task queue according to the upgrade time window of each vehicle, and complete the task allocation.

8. The method for dynamic configuration control of OTA upgrade tasks based on machine learning according to claim 2 or 7, characterized in that: The upgrade task queue is obtained according to the upgrade time window of each vehicle: for each vehicle, if the number of vehicles in the current upgrade queue plus the vehicle does not exceed the upper limit of system compatibility, and the total upgrade time of the current upgrade queue plus the vehicle upgrade time of the vehicle does not exceed the effective time range of the OTA task, then the vehicle will be added to the upgrade task queue.

9. A machine learning-based OTA upgrade task dynamic configuration control system, characterized in that: The system comprises: A data processing module, used for collecting vehicle upgrade data and pre-processing the vehicle upgrade data; The prediction module is used to design the prediction model, train the prediction model and evaluate its effect, and output the upgrade probability and recommended upgrade time window of each vehicle in a specific future time period predicted by the prediction model based on the preprocessed vehicle upgrade data; The upgrade module is used to allocate OTA upgrade tasks according to the output of the prediction model and complete the vehicle upgrade.

10. A computer storage medium, characterized in that: A computer program executable by a processor is stored therein, and the computer program executes a machine learning-based dynamic configuration control method for OTA upgrade tasks as described in any one of claims 1-8.

Images