A vehicle-mounted operating system running scene real-time identification method and system

Abstract

The application discloses a kind of vehicle-mounted operating system running scene real-time identification method and system, the method includes: real-time acquisition system log data, and the system log data is preprocessed, and scene parameter is obtained;According to the algorithm and the scene parameter of preestablished, the actual scene recognition value is calculated;Through the relationship model of pre-trained scene correlation value and scene recognition value and actual scene recognition value, actual scene correlation value is obtained, and the actual scene correlation value is the scene of system identification.This embodiment of the application can dynamically identify different running scenes, thereby providing intelligent basis for subsequent automatic adjustment of resource allocation, thereby significantly improving the response speed of system, resource utilization efficiency and user experience, meanwhile, the technical scheme also has good expansibility, can adapt to new application scenarios and functional requirements, ensure that system always maintains optimal performance in complex and changeable environment.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of intelligent vehicle technology, specifically, a vehicle operating system running scene real-time identification method and system. BACKGROUND

[0002] With the rapid development of automotive electronics technology, intelligent vehicle operating systems have become an important part of modern vehicles. Intelligent vehicle operating systems provide users with navigation, entertainment, safety alerts, vehicle monitoring and other functions. However, existing vehicle operating systems are usually optimized for a single or specific scenario, and existing vehicle operating systems usually use fixed resource allocation and performance optimization strategies, lacking the ability to identify and adapt to complex and diverse scenarios in real time. This results in vehicle operating systems that may be slow to react and have low resource utilization in certain scenarios, thus failing to provide the best user experience.

[0003] In existing solutions, most vehicle operating systems pre-allocate fixed system resources such as CPU, memory and storage space for various applications and services when designed. The advantage of this method is that it is simple to implement and can be optimized for specific applications at the design stage.

[0004] In existing technology, some vehicle operating systems also use event-driven resource scheduling. When a specific event occurs (such as navigation starting, phone access), the system will start a predefined resource allocation scheme to ensure the priority and response speed of critical tasks.

[0005] At the same time, some more advanced vehicle operating systems may also have several common scenarios pre-set, such as navigation scenarios, entertainment scenarios, safety alert scenarios, etc., and performance optimization is carried out for these scenarios. For example, in the navigation scenario, the system will prioritize resource requirements for navigation applications.

[0006] Although existing in-vehicle operating systems have been able to ensure the normal operation of functions to some extent, with the diversification of car functions and the improvement of user needs, the existing technology has obvious shortcomings in the following aspects: (1) Lack of real-time scene recognition capability: The existing fixed resource allocation strategy is generally based on pre-set scenarios and cannot identify real-time complex and variable scenarios. In actual driving process, the in-vehicle operating system may frequently switch between different scenarios, such as mixed use of navigation and entertainment, sudden safety events, etc. If this scenario switching cannot be identified in time and resource optimization is not performed, it may cause the system response to slow down or appear to be stuck, affecting user experience. (2) Low system resource utilization: Since the resource allocation scheme is usually predefined, the diversity and dynamic changes of actual scenarios are not fully considered, resulting in low utilization of system resources. For example, in some cases, the system may allocate too many resources to unimportant background applications, while critical tasks cannot be supported with sufficient resources. (3) Poor scene scalability: Most existing technologies are based on predefined scenarios and fixed optimization strategies, which are difficult to extend or adapt to new application scenarios and functional requirements. With the continuous increase of in-vehicle operating system functions, this limitation will further manifest, restricting the improvement of system performance. (4) Poor user experience: Since the system cannot dynamically optimize according to the actual scenario, there may be some user experience problems, such as navigation delay, music playback stuttering, voice control sluggishness, etc. These problems not only affect user satisfaction, but also may cause safety hazards in critical moments.

[0007] Therefore, there is an urgent need for a technical solution that can identify different scenarios of an in-vehicle operating system in real time, provide a basis for subsequent automatic adjustment of operating system running parameters, and improve the overall performance and user experience of the in-vehicle system. SUMMARY

[0008] One of the purposes of the embodiments of the present application is to provide an in-vehicle operating system running scenario real-time identification method and system, which can identify different scenarios of an in-vehicle operating system in real time and provide a basis for subsequent automatic adjustment of operating system running parameters, thereby improving the overall performance and user experience of the in-vehicle system.

[0009] To solve the above technical problems, in a first aspect, the embodiments of the present application provide an in-vehicle operating system running scenario real-time identification method, which comprises:

[0010] Real-time collection of system log data and preprocessing of the system log data to obtain scenario parameters;

[0011] According to a pre-set algorithm and the scenario parameters, an actual scenario recognition value is calculated;

[0012] An actual scene correlation value is obtained through a pre-trained relationship model between a scene correlation value and a scene recognition value and the actual scene recognition value, wherein the actual scene correlation value is a scene recognized by the system.

[0013] Preferably, the system log data is represented as D={d1, d2, …, d N} and each data record d i corresponds to the scene parameter.

[0014] Further, the method further comprises:

[0015] The running scenes of the vehicle-mounted operating system are pre-classified, including but not limited to driving scene classification, entertainment scene classification, navigation scene and vehicle-mounted phone scene, and the running state under each type of scene is divided.

[0016] Preferably, the training of the pre-trained relationship model between the scene correlation value and the scene recognition value specifically comprises:

[0017] A scene correlation value is obtained in advance to obtain sample data of the scene correlation value;

[0018] A scene recognition value is calculated in advance to obtain sample data of the scene recognition value;

[0019] A relationship model between the scene correlation value and the scene recognition value is established, and the sample data of the scene correlation value and the sample data of the scene recognition value are input into the relationship model for training to obtain a trained relationship model.

[0020] Preferably, the obtaining of the scene correlation value specifically comprises:

[0021] Vehicle-mounted state data and system log data are collected, and the system log data is pre-processed;

[0022] Feature extraction is performed on the pre-processed data to obtain different types of feature vectors, wherein the features include time features, text features and numerical features;

[0023] The extracted different types of feature vectors are spliced to obtain a log vector, and a formula for obtaining the scene correlation value according to the log vector is as follows:

[0024] Scene correlation value SA=driving scene weight Wd×driving scene score Sd+entertainment scene weight We×entertainment scene score Se+navigation scene weight Wn×navigation scene score Sn+vehicle-mounted phone scene weight Wp×vehicle-mounted phone scene score Sp.

[0025] Preferably, the calculation of the actual scene recognition value according to the preset algorithm and the scene parameter specifically comprises:

[0026] According to the scene parameter, a characteristic function of the scene is obtained;

[0027] The calculation formula of the preset algorithm is: SR = ∑ [ω i × F i ], wherein SR represents an actual scene recognition value, F i represents a characteristic function of different scenes, the characteristic function represents different system running scenes, ω i is a weight coefficient corresponding to the characteristic function, the weight coefficient is adjusted according to actual conditions, and:

[0028] ∑ω i = 1.

[0029] Preferably, the relationship model between the scene correlation value and the scene recognition value is,

[0030] SA = θ0+ θ1SR+ θ2x1+ θ3x2+ … + θ m+1 x m + ∈;

[0031] wherein the scene recognition value SR and the characteristic vector X = (x1, x2, … x m ) are inputs, θ0, θ1, …, θ m+1 are parameters of the model, ∈ is an error term, and SA is the scene correlation value;

[0032] In the process of training the relationship, a least square method is used to minimize the error between the predicted scene correlation value and the actual scene correlation value, so as to obtain the trained relationship model between the scene correlation value and the scene recognition value:

[0033]

[0034] wherein N is the number of training samples, SA i and SR i are the actual scene correlation value and the scene recognition value of the i th training sample, and x ij is the i th characteristic value of the i th training sample.

[0035] In a second aspect, the embodiment of the present application further provides a vehicle-mounted operating system running scene real-time identification system, and the system comprises:

[0036] A data acquisition module is configured to acquire system log data and pre-process the system log data;

[0037] A calculation module is configured to calculate an actual scene recognition value according to a preset algorithm and the scene parameter;

[0038] A scene recognition module is configured to obtain an actual scene correlation value by using a pre-trained relationship model between a scene correlation value and a scene recognition value and the actual scene recognition value, wherein the actual scene correlation value is a scene recognized by the system.

[0039] In a third aspect, the embodiments further provide a vehicle-mounted computing device, which comprises a processor or a computer, and a memory configured to store a computer program, wherein the computer program comprises program instructions, and the processor or the computer is configured to invoke the program instructions to execute the method as described above.

[0040] In a fourth aspect, the embodiments further provide a computer-readable storage medium, which stores a computer program, wherein the computer program comprises program instructions, and the program instructions, when executed by a processor or a computer, cause the processor or the computer to execute the method as described above.

[0041] Compared with the prior art, the vehicle-mounted operating system running scene real-time recognition method and system provided by the embodiments of the present application at least have the following beneficial effects:

[0042] The vehicle-mounted operating system of the embodiments of the present application can dynamically recognize different running scenes, thereby providing an intelligent basis for subsequent automatic adjustment of resource configuration, thereby significantly improving the response speed, resource utilization efficiency and user experience of the system. Meanwhile, the technical solution has good expansibility, can adapt to new application scenarios and functional requirements, and ensures that the system always maintains the best performance in a complex and changeable environment. BRIEF DESCRIPTION OF DRAWINGS

[0043] The above-mentioned features, technical characteristics, advantages and implementation manners of the present application will be further described in the following preferred embodiments in a clear and understandable manner in combination with the accompanying drawings.

[0044] Figure 1 A vehicle-mounted operating system running scene real-time recognition method flowchart of the embodiments of the present application;

[0045] Figure 2 A vehicle-mounted operating system running scene real-time recognition system schematic diagram of the embodiments of the present application;

[0046] Figure 3 A vehicle-mounted computing device structure schematic diagram of the embodiments of the present application for realizing vehicle-mounted operating system running scene real-time recognition. DETAILED DESCRIPTION

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, specific implementations of the present application will be described below with reference to the drawings. Obviously, the drawings in the following description only represent some of the embodiments of the present application, and for those skilled in the art, other drawings can be obtained from these drawings without creative effort, and other embodiments can also be obtained.

[0048] For the sake of simplicity of the drawings, only the parts related to the present application are shown in each drawing, which do not represent the actual structure of the product. In addition, in order to make the drawings simple and easy to understand, in some drawings, only one of the components with the same structure or function is shown schematically, or only one of them is marked. In this document, "one" not only means "only one", but also means "more than one" situation.

[0049] The technical solutions of the present application will be described in detail below mainly taking some specific embodiments as examples.

[0050] As shown in the drawings, in order to achieve the purpose of the present application, the present application provides a vehicle-mounted operating system running scene real-time identification method, which comprises: Figure 1

[0051] S1, real-time collection of system log data, and preprocessing of the system log data to obtain scene parameters;

[0052] S2, preset algorithm and the scene parameters, calculation of actual scene recognition value;

[0053] S3, through the pre-trained scene correlation value and the relationship model of scene recognition value, and the actual scene recognition value, the actual scene correlation value is obtained, which is the scene recognized by the system.

[0054] The embodiments of the present application collect vehicle-mounted running state data and system performance data in real time through vehicle-mounted sensors, system logs, user operation records and other means. These collected data include but are not limited to vehicle speed, acceleration, navigation state, CPU usage, memory occupation and other data, and the collected data are preprocessed, including the following:

[0055] For system log data, log data set D = {d1, d2, …, d N}, each log record d i is associated with a specific system running scene, and the system log data reflects the system running state parameters of the system in different application scenes.

[0056] Preferably, the training of the pre-trained scene correlation value and the relationship model of scene recognition value specifically comprises: ​

[0057] obtaining a scene correlation value in advance to obtain sample data of the scene correlation value;

[0058] calculating a scene recognition value in advance to obtain sample data of the scene recognition value;

[0059] establishing a relationship model of the scene correlation value and the scene recognition value, and inputting the sample data of the scene correlation value and the sample data of the scene recognition value into the relationship model for training to obtain a trained relationship model.

[0060] Preferably, the obtaining of the scene correlation value specifically comprises:

[0061] collecting vehicle-mounted state data and system log data, and pre-processing the collected data set;

[0062] extracting features from the pre-processed data to obtain different types of feature vectors, wherein the features include time features, text features and numerical features;

[0063] splicing the extracted different types of feature vectors to obtain a log vector, and obtaining a formula of the scene correlation value according to the log vector as follows:

[0064] Scene correlation value SA = driving scene weight Wd x driving scene score Sd + entertainment scene weight We x entertainment scene score Se + navigation scene weight Wn x navigation scene score Sn + vehicle-mounted phone scene weight Wp x vehicle-mounted phone scene score Sp.

[0065] That is, the collected log data is pre-processed to remove irrelevant information and noise, and to ensure the quality and effectiveness of the data. The pre-processed log data is extracted for features, including time features, text features and numerical features.

[0066] For time features, time information is extracted from log data, and a timestamp can be converted into a specific time format, such as year, month, day, hour, minute and second. The time interval can also be calculated, such as the time difference between two events. These time features can be part of the vector. For example, if the timestamp of the event recorded in the log is 1627225200 (representing July 25, 2021 12:00:00), it can be converted into a time vector such as [2021, 7, 25, 12, 0, 0].

[0067] For text features, the text information in the log data, such as event description, error message, etc., is extracted using natural language processing technology, and the bag-of-words model can be used for feature extraction. For example, when using the bag-of-words model, the text in the log can be segmented, and then the number of occurrences of each word in the log is counted to form a word frequency vector.

[0068] If the event description in the log is "out of memory error", after tokenization, we get ["memory", "not enough", "error"]. Assuming that the different words contained in the entire log set are ["memory", "not enough", "error", "disk", "space"], the word frequency vector of this log can be represented as [1, 1, 1, 0, 0].

[0069] In addition, numerical feature extraction is also performed. Since the log data may contain some numerical information, such as CPU usage, memory occupancy, network traffic, etc., these numerical values are directly taken as part of the vector. For example, if the CPU usage recorded in the log is 75% and the memory occupancy is 60%, these numerical values can be directly taken as two dimensions of the vector, i.e. [75, 60].

[0070] Again, the extracted feature information is normalized to ensure that different features have comparable numerical values, so that the neural network can better learn the feature patterns in different scenarios. The normalization formula is:

[0071]

[0072] where x ij is the original feature value, x′ ij is the normalized feature value, min(x j ) and max(x j ) are the minimum and maximum values of feature j, respectively.

[0073] For example, using Min-Max normalization to normalize CPU usage and memory occupancy to the [0, 1] interval. Assuming that the CPU usage range is [0, 100] and the memory occupancy range is [0, 80], for CPU usage of 75%, the normalized value is 75 / 100 = 0.75; for memory occupancy of 60%, the normalized value is 60 / 80 = 0.75. Then the normalized vector becomes [0.75, 0.75].

[0074] Finally, the vector is spliced to splice the extracted feature vectors of different types to form a comprehensive log vector. The splicing order can be determined according to the importance of the features or specific requirements.

[0075] For example, the time feature vector [2021, 7, 25, 12, 0, 0], the text feature vector [1, 1, 1, 0, 0], and the numerical feature vector [0.75, 0.75] are spliced to obtain a complete log vector [2021, 7, 25, 12, 0, 0, 1, 1, 1, 0, 0, 0.75, 0.75].

[0076] The embodiment of the application classifies the scenarios of the intelligent vehicle operating system in advance, including but not limited to: driving scenario classification, entertainment scenario classification, navigation scenario, and vehicle phone scenario.

[0077] Taking the driving scenario classification as an example, it is further divided into the following categories:

[0078] High-speed driving scenario: the vehicle speed is usually above 80 km / h, the road is relatively flat, and the vehicle runs stably.

[0079] Urban road scenario: the vehicle speed is relatively low, there are many traffic lights and pedestrians, and the road conditions are relatively complex.

[0080] Mountain road scenario: the terrain is large, there are many curves and slopes, and the requirements for vehicle power and control are high.

[0081] Congestion scenario: the vehicle runs slowly, stops and walks, and the traffic flow is large.

[0082] Taking the entertainment scenario classification as an example, it is further divided into the following categories:

[0083] Music playing scenario: the vehicle audio system is playing music, and the user may perform song switching, volume adjustment, etc.

[0084] Video playing scenario: the vehicle display screen is playing a video, which usually occurs when the vehicle is parked or passengers are riding.

[0085] Radio listening scenario: the user is listening to a radio program.

[0086] Taking the navigation scenario classification as an example, it is further divided into the following categories:

[0087] Route planning scenario: the user is using the navigation system to plan a route, inputting destination information, etc.

[0088] Navigation guidance scenario: the navigation system is providing real-time navigation guidance, including voice prompts and map display.

[0089] And the call scenario includes the following categories:

[0090] Call-in scenario: the user is making a vehicle phone call.

[0091] Incoming call waiting scenario: there is an incoming call waiting for the user to answer.

[0092] Call end scenario: just ended a vehicle phone call.

[0093] The embodiment of the application also labels the log data as a category label corresponding to the scenario.

[0094] For example, the scenario categories can be divided according to different running states, business process stages or abnormal situations of the system.

[0095] Further, the embodiment of the present application can also mark the log data as different scenario categories according to actual needs.

[0096] For example, the log can be divided into normal running scenarios, abnormal warning scenarios, serious failure scenarios and the like.

[0097] Each label y i represents a specific scenario, which provides a clear target for subsequent model training.

[0098] Preferably, the scenario marking of the log data specifically includes:

[0099] Information such as time, speed, acceleration, navigation state, entertainment system usage and vehicle-mounted phone state is extracted from the log of the vehicle-mounted system for marking.

[0100] For example, when the speed is above 80 km / h, the navigation system is in an active state, the entertainment system is not used, and the vehicle-mounted phone is in a call state, the scenario is marked as a high-speed driving scenario, a navigation guidance scenario and a call-in-scenario.

[0101] For another example, if there is a call waiting and the speed is low, the scenario is marked as a city road scenario and a call waiting scenario.

[0102] Preferably, data of vehicle-mounted sensors such as acceleration sensors and gyroscopes are also used to mark the driving state of the vehicle, and information such as signal strength of the vehicle-mounted phone is combined.

[0103] For example, when the acceleration sensor detects a large longitudinal acceleration, it may indicate that the vehicle is accelerating or decelerating, and information such as speed can be used to mark different scenarios. If the signal strength of the vehicle-mounted phone is weak at this time, the corresponding scenario can be marked and the phone signal state can be recorded.

[0104] According to the marked scenario categories, a scenario correlation value is used for identification, specifically including:

[0105] Scenario correlation value (S) = driving scenario weight (Wd) x driving scenario score (Sd) + entertainment scenario weight (We) x entertainment scenario score (Se) + navigation scenario weight (Wn) x navigation scenario score (Sn) + vehicle-mounted phone scenario weight (Wp) x vehicle-mounted phone scenario score (Sp)

[0106] Through this formula, a scenario correlation value can be calculated according to the scores and weights of different scenarios, so as to comprehensively judge the scenario state of the current vehicle-mounted system and provide a basis for adaptive adjustment of the system.

[0107] The above formula can be extended according to the definition of the scene category.

[0108] The driving scene score (Sd) is set as:

[0109] High-speed driving scene: if the vehicle speed is greater than or equal to 80 km / h and the road is relatively flat with small acceleration change, 80 points are recorded; if some conditions are met, the score is appropriately reduced, that is, the score can be fine-tuned, and the present embodiment is not limited thereto.

[0110] Urban road scene: the vehicle speed is low, and there are many traffic signals and pedestrians; the score is comprehensively judged according to factors such as congestion degree and signal light frequency, and the range is 40-70 points, which can be fine-tuned, and the present embodiment is not limited thereto.

[0111] Mountain road scene: the score is comprehensively judged according to factors such as terrain undulation, curve and slope conditions, and vehicle control difficulty, and the range is 50-80 points, which can be fine-tuned, and the present embodiment is not limited thereto.

[0112] Congestion scene: the vehicle drives slowly, stops and walks, and the traffic flow is large; the score is judged according to the congestion severity, and the range is 30-60 points, which can be fine-tuned, and the present embodiment is not limited thereto.

[0113] The entertainment scene score (Se) is set as:

[0114] Music playing scene: the score is judged according to factors such as playing time and song switching frequency, and the range is 40-70 points, which can be fine-tuned, and the present embodiment is not limited thereto.

[0115] Video playing scene: the score is judged according to factors such as playing time and video type, and the range is 30-60 points, which can be fine-tuned.

[0116] Radio listening scene: the score is judged according to factors such as listening time and channel switching frequency, and the range is 30-60 points, which can be fine-tuned.

[0117] The navigation scene score (Sn) is set as:

[0118] Route planning scene: the score is judged according to factors such as planning time and destination complexity, and the range is 40-70 points, which can be fine-tuned, and the present embodiment is not limited thereto.

[0119] Navigation guidance scene: the score is judged according to factors such as navigation accuracy and voice prompt frequency, and the range is 50-80 points, which can be fine-tuned, and the present embodiment is not limited thereto.

[0120] The score (Sp) of the car phone scene is set as:

[0121] The in-call scene: judging according to the call duration, signal strength and other factors, the range is 40-70 points, the range can be fine-tuned, and the embodiment of the application is not limited.

[0122] The incoming call waiting scene: judging according to the waiting time, importance judgment and other factors, the range is 30-60 points, the range can be fine-tuned, and the embodiment of the application is not limited.

[0123] The call end scene: judging according to the call duration, whether there is a subsequent operation and other factors, the range is 20-50 points, the range can be fine-tuned, and the embodiment of the application is not limited.

[0124] For the weight setting part, it can be set as:

[0125] The driving scene weight (Wd), the entertainment scene weight (We), the navigation scene weight (Wn) and the car phone scene weight (Wp) can be set according to actual needs and importance.

[0126] For example, if driving safety is the primary consideration, the driving scene weight can be set higher, such as 0.4; the entertainment scene and the car phone scene weight are relatively low, such as 0.2; the navigation scene weight can be set to 0.2.

[0127] The above score and weight parameter settings should not be understood as the only limitation of the technical solution of the application. The above parameter settings can be fine-tuned according to the specific situation after the general technical personnel in the field debug the car operation system in the actual implementation process. The fine-tuned score or weight parameter value should be within the protection scope of the application.

[0128] Preferably, the calculation of the actual scene recognition value according to the preset algorithm and the scene parameter specifically comprises:

[0129] Obtaining a characteristic function of the scene according to the scene parameter;

[0130] The calculation formula of the preset algorithm is: SR = ∑[ω i × F i ], wherein F i represents the characteristic function of different scenes, the characteristic function represents different system running scenes, ω i is a weight coefficient corresponding to the characteristic function, the weight coefficient is adjusted according to the actual situation, and:

[0131] ∑ω i = 1.

[0132] F irepresent different feature functions, and represent various factors related to a specific scenario, for example, in some embodiments:

[0133] f1 (Speed and Acceleration Features) is a speed and acceleration feature function, F1 = V α × A β .

[0134] The current vehicle speed, acceleration rate of change, and other factors can be considered comprehensively. Among them, V is the vehicle speed, A is the acceleration, and a and β are adjustable weight parameters used to adjust the importance of vehicle speed and acceleration in this feature function.

[0135] The speed and acceleration feature function provides an effective feature description for vehicle system scenario recognition by comprehensively considering the two important vehicle dynamic parameters of speed and acceleration, helping the system to more accurately judge the scenario in which the vehicle is located, and thus to realize corresponding optimization and adjustment.

[0136] F2 (Navigation Status Features) is a navigation status feature function, F2 = N γ × C δ × D τ .

[0137] Whether the navigation is turned on, the complexity of the current navigation route, the distance to the destination, and other factors are considered.

[0138] Among them, N represents the navigation on-off state (1 for on and 0 for off), C represents the quantified value of route complexity, D represents the distance, and γ, δ, and τ are weight parameters.

[0139] The complexity of the navigation route (C) is composed of three parts:

[0140] N = T × R × L;

[0141] The number of turns (T) is used to count the number of expected turns in the navigation route. The more turns, the more complex the route usually means. The number of turns can be determined according to the route planning information provided by the navigation software. The number of turns is normalized.

[0142] The road type (R) is used to assign values to different types of roads and then count the number of different types of roads in the route. For example, expressways can be assigned a value of 1, urban trunk roads a value of 2, secondary trunk roads a value of 3, and branch roads a value of 4, etc. The higher the sum of the road type assignment values appearing in the route, the greater the road type diversity, and the more complex the route. If a route contains a section of expressway and two sections of urban secondary trunk road, the numerical representation of road type diversity can be 1+3+3=7.

[0143] The value of the traffic restriction factor (L) is used to comprehensively consider the traffic restriction factors in the route, such as one-way streets, construction areas, traffic congestion areas, and the like.

[0144] The embodiment of the present application can assign a weight to each traffic restriction factor, and then multiply the number of traffic restriction factors appearing in the route by the corresponding weight to obtain the value.

[0145] For example, the weight of a one-way street is 1, the weight of a construction area is 2, and the weight of a traffic congestion area is 3. If a route has one construction area and one traffic congestion area, the value of the traffic restriction factor is 2+3=5.

[0146] F3 (Entertainment Usage Features) is a characteristic function of the entertainment usage state:

[0147]

[0148] It includes music playing state, video playing state, radio listening state, and the like.

[0149] Wherein, M represents the music playing state (1 for playing and 0 for not playing), V represents the video playing state, B represents the radio listening state, ∈, θ and are weight parameters.

[0150] F4 (Phone Call Features) is a characteristic function of the car phone:

[0151] F4=P μ ×S π ;

[0152] It covers the phone call state, signal strength, and the like, wherein P represents the call state (1 for calling and 0 for not calling), S represents the quantized value of signal strength, μ and π are weight parameters.

[0153] F5 (Sensor Data Features) is a characteristic function of sensor data:

[0154] F5=T ρ ×L σ ;

[0155] It can use temperature sensor, light sensor, and the like data, wherein T represents the quantized value of temperature, L represents the quantized value of light intensity, ρ and σ are weight parameters.

[0156] ω i is the weight coefficient corresponding to the characteristic function, which can be adjusted according to the actual situation to reflect the importance of different characteristics in scene recognition, and the sum of the weight coefficients is 1, that is:

[0157] ∑ω i = 1.

[0158] An empirical method can be used to assign an initial weight value to each feature function, or a gradient descent method can be used to gradually adjust the weight coefficients by calculating the gradient of the scene recognition value with respect to each weight coefficient, so that the scene recognition value more accurately reflects the scene label of the sample data.

[0159] By adjusting the weight coefficients of each feature function, a scene recognition model suitable for the vehicle-mounted system can be constructed according to different application requirements and data characteristics. This formula can flexibly adapt to different combinations of scene factors, providing accurate scene recognition results for the vehicle-mounted system to make corresponding intelligent adjustments and optimizations.

[0160] Preferably, the embodiment of the present application trains the relationship model between the scene correlation value and the scene recognition value as:

[0161] SA = θ0+ θ1SR+ θ2x1+ θ3x2+ … + θ m+1 x m + ∈;

[0162] Wherein, the scene recognition value SR and the feature vector X = (x1, x2, … x m ) are inputs, θ0, θ1, …, θ m+1 are model parameters, ∈ is an error term, and SA is a scene correlation value.

[0163] In the process of training the relationship, the least squares method is used to minimize the error between the predicted scene correlation value and the actual scene correlation value, thereby obtaining the trained relationship model between the scene correlation value and the scene recognition value:

[0164]

[0165] Wherein, N is the number of training samples, SA i and SR i are the actual scene correlation value and the scene recognition value of the i-th training sample, and x ij is the i-th feature value of the i-th training sample.

[0166] The trained model is deployed to the actual system for real-time analysis and prediction of newly generated log data, and by quickly and accurately identifying the scene to which the log data belongs, timely decision support is provided for system monitoring, fault diagnosis and performance optimization.

[0167] According to the different scenarios identified by the model, the working frequency of the CPU is adjusted in real time. For example, in navigation mode, the CPU frequency priority of navigation-related processes is increased to ensure that tasks such as map loading and route planning are completed quickly. In entertainment mode, the frequency of navigation processes can be appropriately reduced to allocate more frequency resources to entertainment applications.

[0168] Processes for critical tasks are set to high priority, such as setting the process priority of the vehicle safety monitoring system to the highest in driving mode to ensure timely response. For non-critical tasks such as background updates, set a lower priority.

[0169] Adjust the cache strategy for different scenarios. For example, in navigation mode, cache more map data and route information to reduce loading time. In entertainment mode, cache music and video files to improve playback smoothness.

[0170] Dynamically allocate memory according to scenario requirements. For example, when watching high-definition videos, allocate more memory to the video playback application, and reduce its memory usage in other scenarios.

[0171] Determine the allocation of network bandwidth according to the scenario. In navigation mode, prioritize the bandwidth required for data updates and real-time traffic acquisition of navigation software. In entertainment mode, allocate bandwidth reasonably to music and video streaming services.

[0172] Automatically switch network connection methods according to scenarios. For example, when driving at high speed, prefer to connect to 4G / 5G networks with higher stability; when parked, switch to Wi-Fi networks to save traffic.

[0173] Data preloading: according to scenario prediction of user demand, pre-load the data that may be used into the storage. For example, on the route that the user often drives, pre-load the map data and point of interest information of the region.

[0174] Periodically clean up unnecessary temporary files and cache to release storage resources. In low storage state, prefer to clean up cache data of non-critical applications.

[0175] The vehicle-mounted operating system of the embodiment of the present application can dynamically identify different use scenarios and automatically adjust resource configuration, thereby significantly improving the response speed, resource utilization efficiency and user experience of the system. At the same time, the technical scheme also has good expansibility, can adapt to new application scenarios and functional requirements, and ensures that the system always maintains optimal performance in complex and variable environments.

[0176] As shown in Figure 2 The embodiment of the present application also provides a vehicle-mounted operating system running scenario real-time identification system, which comprises:

[0177] A data collection module is configured to collect system log data and pre-process the system log data.

[0178] A calculation module is configured to calculate an actual scene recognition value according to a preset algorithm and the scene parameter.

[0179] A scene recognition module is configured to obtain an actual scene correlation value through a pre-trained relationship model between a scene correlation value and a scene recognition value and the actual scene recognition value, wherein the actual scene correlation value is a scene recognized by the system.

[0180] In the system embodiment, the implementation manner is the same as that of the method embodiment, which will not be described herein.

[0181] Compared with the prior art, the vehicle-mounted operating system running scene real-time recognition method and system provided by the embodiment has at least the following beneficial effects:

[0182] The vehicle-mounted operating system can dynamically recognize different running scenes, thereby providing intelligent basis for subsequent automatic adjustment of resource configuration, and thereby significantly improving the response speed, resource utilization efficiency and user experience of the system.

[0183] The trained model is deployed to the actual system to perform real-time analysis and prediction on newly generated log data.

[0184] In the third aspect, the embodiment further provides a vehicle-mounted computing device, which comprises a processor or a calculator and a memory.

[0185] In the fourth aspect, the embodiment further provides a computer readable storage medium, which stores a computer program.

[0186] As Figure 3As shown, the vehicle-mounted computing device provided by the embodiments of the present application includes a processor or computer (not shown in the figure) 1001 and a memory 1002, and the processor or computer 1001 and the memory 1002 can be connected to each other through a communication bus 1003. The communication bus 1003 can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus 1003 can be divided into an address bus, a data bus, a control bus, etc. For the convenience of representation, the memory 1002 is used for storing a computer program, the computer program includes program instructions, and the processor 1001 is configured to invoke the program instructions, and the above program includes instructions for executing part or all of the steps in the method contained above.

[0187] The processor 1001 can be a general central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of the programs of the above solutions.

[0188] The memory 1002 can be a read-only memory (ROM) or other type of static storage device that can store static information and instructions, a random access memory (RAM) or other type of dynamic storage device that can store information and instructions, and can also be an Electrically Erasable Programmable Read-Only Memory (EEPROM), a Compact Disc Read-Only Memory (CD-ROM) or other optical disk storage, an optical disk storage (including a compact optical disk, a laser disk, an optical disk, a digital versatile optical disk, a Blu-ray disk, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and capable of being accessed by a computer, but not limited to this. The memory can exist independently and be connected to the processor through a bus. The memory can also be integrated with the processor.

[0189] The electronic device 1000 can also include a communication module 1004 and a display 1005. The communication module 1004 can be in communication connection with the optical tracking device. The communication module 1004 can be a wireless communication module (such as a WiFi module, a Bluetooth module, etc.) or a wired communication module.

[0190] In addition, the electronic device 1000 can further include a communication interface (such as a USB interface, a microphone interface, etc.), an antenna, and the like, general components, which are not described here in detail.

[0191] It should be noted that, for the foregoing method embodiments, in order to simply describe, the foregoing method embodiments are all described as a combination of a series of actions, but those skilled in the art should know that the present application is not limited to the order of the actions described, because according to the present application, certain steps can be performed in other order or at the same time. Secondly, those skilled in the art should know that the embodiments described in the specification all belong to preferred embodiments, and the actions and modules involved are not necessarily necessary for the present application.

[0192] In the foregoing embodiments, the description of each embodiment is focused on, and the part not described in detail in a certain embodiment can be referred to the related description of other embodiments.

[0193] In several embodiments provided by the present application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely schematic. The division of the units is only a logical function division. There can be another division manner in actual implementation. For example, a plurality of units or components can be combined or integrated into another apparatus, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interfaces, apparatuses or units, and can be electrical or other forms.

[0194] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, that is, they can be located in one place, or can be distributed on a plurality of network units. According to actual needs, part or all of the units can be selected to achieve the purpose of the embodiment scheme.

[0195] In addition, the functional units in each embodiment of the application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software program module.

[0196] The integrated unit, if implemented in the form of a software program module and sold or used as an independent product, can be stored in a computer readable memory. Based on such understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or the whole or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a memory and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server or a network device, etc.) to execute all or part of the steps of the method described in the various embodiments of the present application. The aforementioned memory includes: a U disk, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk, and various media that can store program codes.

[0197] A person of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by a program instructing relevant hardware, and the program can be stored in a computer readable memory, which can include a flash disk, a read-only memory, a random access memory, a magnetic disk or an optical disk, etc.

[0198] The embodiments of the present application are described in detail above, and the principles and implementation manners of the present application are described by applying specific examples. The above description of the embodiments is only used to help understand the method of the present application and its core idea; at the same time, for those of ordinary skill in the art, according to the idea of the present application, the specific implementation manner and application range can be changed, and the above description of the embodiments should not be understood as limiting the present application.

[0199] It should be noted that the above embodiments can be freely combined as needed. The above is only a preferred embodiment of the present application, and it should be noted that, for those of ordinary skill in the art, without departing from the principles of the present application, a number of improvements and refinements can be made, and these improvements and refinements should also be considered as the protection scope of the present application.

Claims

1. A vehicle-mounted operating system running scene real-time identification method, characterized in that, The method comprises: real-time acquisition of system log data, and preprocessing of the system log data to obtain a scene parameter; calculation of an actual scene recognition value according to a preset algorithm and the scene parameter; obtaining of an actual scene correlation value through a pre-trained scene correlation value and scene recognition value relationship model and the actual scene recognition value, the actual scene correlation value being a scene recognized by the system, wherein the training of the pre-trained scene correlation value and scene recognition value relationship model specifically comprises: previously obtaining a scene correlation value to obtain sample data of the scene correlation value; previously calculating a scene recognition value to obtain sample data of the scene recognition value; establishing a relationship model of the scene correlation value and the scene recognition value, and inputting the sample data of the scene correlation value and the sample data of the scene recognition value into the relationship model for training to obtain a trained relationship model; wherein the previously obtaining of the scene correlation value specifically comprises: acquisition of system log data, and preprocessing of the system log data; feature extraction from the preprocessed data to obtain different types of feature vectors, the features including time features, text features and numerical features; splicing of the extracted different types of feature vectors to obtain a log vector, and a formula for obtaining the scene correlation value according to the log vector being as follows: scene correlation value SA = driving scene weight Wd × driving scene score Sd + entertainment scene weight We × entertainment scene score Se + navigation scene weight Wn × navigation scene score Sn + vehicle-mounted phone scene weight Wp × vehicle-mounted phone scene score Sp.

2. The method of claim 1, wherein the running scene of the in-vehicle operation system is identified in real time. The system log data is represented as D = {d1, d2,..., dn}, each data record d N corresponds to the scene parameter. i The system log data is represented as D = {d1, d2,..., dn}, each data record d N corresponds to the scene parameter. i The system log data is represented as D = {d1, d2,..., dn 3. The method of claim 2, wherein the running scene of the in-vehicle operation system is identified in real time. The method further comprises: pre-classification of running scenes of the vehicle-mounted operating system, the running scenes including but not limited to driving scene classification, entertainment scene classification, navigation scene and vehicle-mounted phone scene, and division of running states under each type of scene.

4. The method of claim 1, wherein the running scene of the in-vehicle operation system is identified in real time. The calculation of the actual scene recognition value according to the preset algorithm and the scene parameter specifically comprises: obtaining of a feature function of the scene according to the scene parameter; The calculation formula of the preset algorithm is: SR=∑[ω i ×F i ], wherein SR represents an actual scene recognition value, F i represents a feature function of different scenes, the feature function represents different system running scenes, ω i is a weight coefficient corresponding to the feature function, the weight coefficient is adjusted according to actual conditions, and the preset algorithm is: ∑ω i = 1.

5. The vehicle-mounted operating system running scene real-time identification method of claim 1 or 4, characterized in that, the pre-trained scene correlation value and scene recognition value relationship model being: SA = θ0+ θ1SR+ θ2x1+ θ3x2+... + θ m+1 x m + ∈; wherein the scene recognition value SR and the feature vector X = (x1, x2, … x m ) are inputs, θ0, θ1, …, θ m+1 are parameters of the model, ∈ is an error term, and SA is a scene association value; in the process of training the relationship, a least square method is used to minimize the error between the predicted scene correlation value and the actual scene correlation value, thereby obtaining the trained scene correlation value and scene recognition value relationship model: where N is the number of training samples, SA i and SR i are the actual scene relevance value and the scene recognition value of the ith training sample, x ij is the jth feature value of the ith training sample.

6. A vehicle-mounted operating system running scene real-time identification system, characterized in that, The system comprises: a data acquisition module for real-time acquisition of system log data, and preprocessing of the system log data to obtain a scene parameter; a calculation module for calculation of an actual scene recognition value according to a preset algorithm and the scene parameter; a scene recognition module for obtaining of an actual scene correlation value through a pre-trained scene correlation value and scene recognition value relationship model and the actual scene recognition value, the actual scene correlation value being a scene recognized by the system, wherein the training of the pre-trained scene correlation value and scene recognition value relationship model specifically comprises: previously obtaining a scene correlation value to obtain sample data of the scene correlation value; previously calculating a scene recognition value to obtain sample data of the scene recognition value; A relationship model of the scene correlation value and the scene identification value is established, and sample data of the scene correlation value and sample data of the scene identification value are input into the relationship model for training to obtain a trained relationship model; The pre-obtained scene correlation value specifically includes: System log data is collected and pre-processed; Feature extraction is performed on the pre-processed data to obtain different types of feature vectors, and the features include time features, text features, and numerical features; The different types of feature vectors are spliced to obtain a log vector, and a formula for obtaining the scene correlation value according to the log vector is as follows: Scene correlation value SA = driving scene weight Wd × driving scene score Sd + entertainment scene weight We × entertainment scene score Se + navigation scene weight Wn × navigation scene score Sn + vehicle-mounted phone scene weight Wp × vehicle-mounted phone scene score Sp.

7. An in-vehicle device characterized by comprising: The computer readable storage medium stores a computer program, and the computer program includes program instructions. When the program instructions are executed by the processor or the computer, the processor or the computer executes the method according to any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, and the computer program includes program instructions. When the program instructions are executed by the processor or the computer, the processor or the computer executes the method according to any one of claims 1-5.

Images