Steering wheel zero deviation reminder method, device, electronic device and storage medium

Abstract

The present application relates to a steering wheel zero bias reminder method, device, electronic device and storage medium, the method comprising: obtaining real-time driving data of a target vehicle; processing the real-time driving data through a pre-trained random forest model to obtain a steering wheel zero bias value in the target vehicle; and sending a prompt message based on the steering wheel zero bias value, wherein the prompt message is used to prompt the driver to adjust the direction of the steering wheel. The present application improves the safety and stability of vehicle driving.

Description

Technical Field

[0001] This application relates to the technical field of vehicles, and in particular, to a method, device, electronic device, and storage medium for reminding of zero deviation of a steering wheel. Background Art

[0002] Zero deviation of the steering wheel means that when the vehicle is driving straight, the steering wheel is not completely in the central position, but there is a certain deviation angle. This situation is usually the result of the combined action of multiple factors. On the one hand, due to insufficient installation accuracy of components such as the steering wheel, steering tie rod, and steering gear, the steering wheel may not be in the central position when stationary, that is, there is zero deviation. On the other hand, the components of the steering system gradually wear with the use of the vehicle, such as wear of the steering tie rod ball joint and steering knuckle. These wears will cause an increase in the clearance of the steering system, which in turn affects the zero position accuracy of the steering wheel. When the steering wheel signal shows zero, the vehicle may slightly turn to one side and have zero deviation.

[0003] Zero deviation of the steering wheel will destroy the preset conversion relationship between the steering wheel angle and the front wheel angle of the vehicle, resulting in the inability to accurately control the vehicle steering through the steering wheel, and affecting the safety and stability of vehicle driving. Summary of the Invention

[0004] This application provides a method, device, electronic device, and storage medium for reminding of zero deviation of a steering wheel to solve the problem of low safety and stability of vehicle driving.

[0005] In a first aspect, this application provides a method for reminding of zero deviation of a steering wheel, the method including: obtaining real-time driving data of a target vehicle; processing the real-time driving data through a pre-trained random forest model to obtain a zero deviation value of the steering wheel in the target vehicle; and sending a prompt message based on the zero deviation value of the steering wheel, where the prompt message is used to prompt a driver to adjust the direction of the steering wheel.

[0006] Optionally, the pre-training process of the random forest model includes: randomly and with replacement extracting training samples to construct a feature set for each decision tree, where each training sample includes sample driving data of a vehicle and a corresponding steering wheel deviation value; when splitting each node of each decision tree, randomly selecting m types of sample driving data for the node, where m < M, and M is the number of categories of sample driving data in the training samples; traversing the m types of sample driving data in the feature set, finding the minimum splitting item by trying multiple splitting items, and splitting the node based on the minimum splitting item, where the splitting item is used to split the data set in the node into two subsets, and the data set is the feature set or the subset after experiencing the previous split; recursively splitting each node of the decision tree until a set stop condition is met, completing the construction of the decision tree and the generation of the random forest model.

[0007] Optionally, traversing the m types of sample driving data in the feature set and finding the minimum splitting item by trying multiple splitting items includes: traversing the n training samples in the feature set and the m types of sample driving data in the n training samples to obtain n*m splitting items; performing the following operations for each splitting item: using the splitting item to split the data set of the node into two subsets, and calculating the mean square error of the node based on the split subsets, wherein the mean square error refers to the mean square error of the steering wheel deviation value in the data set of the node; selecting the splitting item corresponding to the minimum mean square error from all splitting items as the minimum splitting item.

[0008] Optionally, calculating the mean square error of the node based on the split subsets includes: calculating the mean square error of the steering wheel deviation value in each subset; determining the sample ratio of the subset based on the sample size in the subset and the sample size in the node; taking the sample ratio as a weight, performing weighted summation of the mean square errors of the two subsets to obtain the mean square error of the node.

[0009] Optionally, recursively splitting each node of the decision tree until a set stopping condition is met includes: randomly selecting m types of sample driving data for the root node and finding a minimum splitting term, using the minimum splitting term to split the data set in the root node into two subsets; creating a new child node for each of the split subsets; and executing the steps of selecting m types of sample driving data and splitting the data set based on the minimum splitting term for each of the child nodes until the set stopping condition is met.

[0010] Optionally, the real-time driving data of the target vehicle is processed by the random forest model to obtain the steering wheel zero bias value in the target vehicle, including: using each decision tree in the random forest model to process the real-time driving data of the target vehicle to obtain the steering wheel zero bias value output by each decision tree; taking the average of the steering wheel bias values ​​of each decision tree to obtain the steering wheel zero bias value in the target vehicle.

[0011] Optionally, obtaining sample driving data of the vehicle includes: using a steering wheel angle sensor deployed on the vehicle to collect a steering wheel angle, and using a vehicle speed sensor deployed on the vehicle to collect an actual driving speed of the vehicle; calculating a steering wheel speed based on the steering wheel angle and a time interval of steering wheel rotation; and calculating a steering wheel torque based on the steering wheel angle, a preset tire friction coefficient, a preset tire radius, and a preset steering knuckle arm length.

[0012] In the second aspect, the present application provides a steering wheel zero bias reminder device, which includes: an acquisition module for acquiring real-time driving data of a target vehicle; a processing module for processing the real-time driving data through a pre-trained random forest model to obtain a steering wheel zero bias value in the target vehicle; and a prompt module for issuing a prompt message based on the steering wheel zero bias value, wherein the prompt message is used to prompt the driver to adjust the direction of the steering wheel.

[0013] In a third aspect, the present application provides an electronic device comprising: at least one communication interface; at least one bus connected to the at least one communication interface; at least one processor connected to the at least one bus; and at least one memory connected to the at least one bus.

[0014] In a fourth aspect, the present application further provides a computer storage medium storing computer executable instructions, wherein the computer executable instructions are used to execute the steering wheel zero deviation reminder method described in any one of the above items of the present application.

[0015] The above-mentioned technical solution provided by the embodiment of the present application has the following advantages compared with the prior art: the random forest model can timely monitor the zero offset of the steering wheel according to the real-time driving data of the vehicle, and promptly remind the driver to adjust the steering wheel direction when the steering wheel zero offset is found, thereby avoiding safety hazards caused by the steering wheel zero offset and improving the safety and stability of vehicle driving. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0018] One or more embodiments are exemplarily described by pictures in the corresponding drawings, and these exemplified descriptions do not constitute limitations on the embodiments. Elements with the same reference numerals in the drawings represent similar elements, and unless otherwise stated, the figures in the drawings do not constitute proportional limitations.

[0019] Figure 1 A flow chart of a method for steering wheel zero deviation reminder provided in an embodiment of the present application;

[0020] Figure 2 A flow chart of a method for generating a random forest model provided in an embodiment of the present application;

[0021] Figure 3 A schematic diagram of the structure of a steering wheel zero deviation reminder device provided in an embodiment of the present application;

[0022] Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0023] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0024] The disclosure below provides many different embodiments or examples to implement different structures of the present invention. In order to simplify the disclosure of the present invention, the parts and settings of specific examples are described below. Of course, they are only examples, and the purpose is not to limit the present invention. In addition, the present invention can repeat reference numbers and / or letters in different examples. This repetition is for the purpose of simplification and clarity, and does not itself indicate the relationship between the various embodiments and / or settings discussed.

[0025] The present application provides a steering wheel zero deviation reminder system, including a sensor, a processor and a playback device. The processor constructs a decision tree based on a training sample set and generates a random forest model. The sensor collects current real-time driving data and transmits it to the processor. The processor uses the random forest model to process the data to obtain a steering wheel zero deviation value, and then generates a corresponding prompt message, and uses a playback device to play the prompt message, thereby reminding the driver to correct the direction of the steering wheel. The steering wheel zero deviation reminder system monitors and reminds the driver of the zero deviation state of the steering wheel in real time, thereby ensuring the stability and safety of the driver's vehicle driving and reducing the risk of traffic accidents caused by steering wheel zero deviation.

[0026] Based on the same technical concept, the present application provides a steering wheel zero deviation reminder method, which can be applied to a server or a processor to prompt the driver to adjust the steering wheel direction in time after detecting that the vehicle's steering wheel has zero deviation.

[0027] The present application embodiment is explained by taking application to a server as an example. Figure 1 As shown, the method comprises the following steps:

[0028] Step 101: Acquire real-time driving data of a target vehicle.

[0029] Step 102: Process the real-time driving data through a pre-trained random forest model to obtain the zero-offset value of the steering wheel in the target vehicle.

[0030] Step 103: Send a prompt message based on the zero-offset value of the steering wheel, where the prompt message is used to prompt the driver to adjust the direction of the steering wheel.

[0031] In Steps 101 to 103, the random forest model can timely monitor the zero-offset situation of the steering wheel according to the real-time driving data of the vehicle, and when it detects that the steering wheel has a zero offset, it can timely remind the driver to adjust the direction of the steering wheel, avoiding potential safety hazards caused by the zero offset of the steering wheel, and improving the safety and stability of vehicle driving.

[0032] This application is applicable to any driving scenario. Especially in complex driving scenarios such as winding roads and reverse parking, the accurate position of the steering wheel is crucial for the driving direction. The zero-offset reminder system of the steering wheel can reduce the risk of traffic accidents caused by the driver forgetting the position offset of the steering wheel.

[0033] Before Step 101, the server needs to construct decision trees and generate a random forest model. The steps for generating the random forest model are as Figure 2 shown and include the following content.

[0034] Step 201: Randomly draw training samples with replacement to construct the feature set of each decision tree. Each training sample includes the sample driving data of the vehicle and the corresponding steering wheel deviation value.

[0035] Step 202: When splitting each node of each decision tree, randomly select m types of sample driving data for the node, where m < M, and M is the number of categories of sample driving data in the training samples.

[0036] Step 203: Traverse the m types of sample driving data in the feature set, find the minimum splitting item by trying multiple splitting items, and split the node based on the minimum splitting item. The splitting item is used to split the data set in the node into two subsets, and the data set is the feature set or the subset after the previous split.

[0037] Step 204: Recursively split each node of the decision tree until the set stop condition is met, completing the construction of the decision tree and the generation of the random forest model.

[0038] In Step 201, the server first needs to establish a training sample set. The training sample set includes multiple training samples. Each training sample includes multiple sample driving data of the vehicle and the steering wheel deviation value corresponding to the sample driving data. Each training sample has the same category and number of sample driving data.

[0039] For example, each training sample is X=[θ(t), w(t), T(t), v(t), y], where θ(t) is the steering wheel angle at time t, w(t) is the steering wheel speed at time t, T(t) is the steering wheel torque at time t, v(t) is the vehicle speed at time t, and y is the steering wheel deviation value at time t.

[0040] The process of constructing a training sample set includes the following steps:

[0041] Step 1: Get the data.

[0042] When the vehicle is started and in driving state, the sensor will start to collect data in real time. The steering wheel zero deviation reminder system is equipped with a steering wheel angle sensor and a vehicle speed sensor on the vehicle. The steering wheel angle sensor generates Hall pulses. The steering wheel angle sensor calculates the steering wheel rotation angle travel according to the proportional relationship between the number of pulses and the travel, and converts it into an electrical signal output. After the output signal is processed by the ECU (Electronic Control Unit), the actual steering wheel angle θ(t) can be obtained, so as to accurately capture the steering wheel rotation situation; the vehicle speed sensor detects the speed of the transmission and then generates an electrical signal to represent the speed. After being processed by the ECU, the actual driving speed v(t) of the vehicle is obtained. The sensor sends the actual steering angle and actual driving speed to the server. In addition, the technician connects the OBD-II (On-Board Diagnostics II) diagnostic tool to the diagnostic interface of the car, and then uses the diagnostic tool to find the setting items related to the steering wheel zero deviation. In the setting mode, the steering wheel zero deviation value y at time t can be read. Among them, OBD-II is an automotive electronic diagnostic system, which is mainly used to detect and diagnose vehicle faults. It can read various data of the vehicle by connecting to the vehicle's computer system.

[0043] Step 2: Data cleaning.

[0044] Data cleaning includes but is not limited to processing missing items and outliers. For the input actual turning angle and actual driving speed, the missing values ​​can be filled by interpolation: , where Interpolate represents interpolation, xi represents the i-th data point, xi+1 represents the i+1-th data point, and xi-1 represents the i-1-th data point.

[0045] Step 3: Feature extraction.

[0046] Features that can reflect the instantaneous dynamic changes of the vehicle are extracted from the original data. The purpose of feature extraction is to eliminate unimportant or redundant features from the original feature set, thereby simplifying the model, improving training efficiency and output accuracy.

[0047] Calculate the rate of change of the steering wheel angle: .

[0048] Calculate the steering wheel torque T(t)=μ*r*L*θ(t), where T(t) is the steering wheel torque at time t, μ is the friction coefficient between the tire and the ground, r is the tire radius, L is the length of the steering knuckle arm, and θ(t) is the steering wheel angle at time t.

[0049] Step 4: Data normalization.

[0050] Normalize the data: ,in, is the normalized data, xi is the original data, is the mean of feature x, is the standard deviation of x.

[0051] Step 5: Generate a training sample set.

[0052] The training sample set includes multiple training samples, for example, the following four training samples.

[0053] X1=[θ(t)1,w(t)1,T(t)1,v(t)1,y1];

[0054] X2=[θ(t)2,w(t)2,T(t)2,v(t)2,y2];

[0055] X3=[θ(t)3,w(t)3,T(t)3,v(t)3,y3];

[0056] X4=[θ(t)4, w(t)4, T(t)4, v(t)4, y4].

[0057] The random forest consists of multiple decision trees. The server randomly extracts training samples with replacement to construct a feature set for each decision tree, which includes at least some of the training samples.

[0058] For example, the feature set of decision tree 1 is:

[0059] X1=[θ(t)1,w(t)1,T(t)1,v(t)1,y1];

[0060] X2=[θ(t)2,w(t)2,T(t)2,v(t)2,y2];

[0061] X3=[θ(t)3, w(t)3, T(t)3, v(t)3, y3].

[0062] The feature set of decision tree 2 is:

[0063] X1=[θ(t)1,w(t)1,T(t)1,v(t)1,y1];

[0064] X2=[θ(t)2,w(t)2,T(t)2,v(t)2,y2];

[0065] X4=[θ(t)4, w(t)4, T(t)4, v(t)4, y4].

[0066] The feature set of decision tree 3 is:

[0067] X2=[θ(t)2,w(t)2,T(t)2,v(t)2,y2];

[0068] X3=[θ(t)3,w(t)3,T(t)3,v(t)3,y3];

[0069] X4=[θ(t)4, w(t)4, T(t)4, v(t)4, y4].

[0070] In step 202 to step 204, the decision tree is composed of multiple CART (Classification and Regression Tree) nodes, and the server recursively splits in the order from the root node to the leaf node to generate the decision tree.

[0071] The process of node splitting is as follows: when each node of the decision tree is split, the server randomly selects m types of sample driving data from all categories of sample driving data M. For example, the sample driving data categories include θ(t), w(t), T(t), and v(t), then M is 4. If m is 2, you can choose θ(t), w(t), T(t), v(t), or w(t), T(t). The server traverses the m types of sample driving data in the feature set to obtain multiple split items, and uses each split item to split the node to obtain the mean square error (MSE) corresponding to each split item. The server selects the minimum split item from the multiple split items based on minimizing the mean square error, and then uses the minimum split item to divide the node data set into two subsets. In this way, each subset corresponds to a new child node, and the node that has just been split is used as the parent node of the child node. The child node also performs the process of selecting m types of sample driving data and using the minimum split item to split the node. This continues until the set stop condition is met, thereby completing the construction of the decision tree and the generation of the random forest model.

[0072] The random forest model mentioned in this application can effectively reduce the risk of overfitting of a single model and improve the generalization ability and stability of the overall model by constructing multiple decision trees. In the steering wheel zero deviation warning system, this means that it can more accurately identify slight deviations of the steering wheel, reduce false positives and false negatives, and improve the detection accuracy of the system.

[0073] As an optional implementation method, traversing m types of sample driving data in a feature set and finding the minimum splitting item by trying multiple splitting items includes: traversing n training samples in the feature set and m types of sample driving data in the n training samples to obtain n*m splitting items; performing the following operations for each splitting item: using the splitting item to split the node's data set into two subsets, and calculating the mean square error of the node based on the split subsets, wherein the mean square error refers to the mean square error of the steering wheel deviation value in the node's data set; selecting the splitting item corresponding to the minimum mean square error from all splitting items as the minimum splitting item.

[0074] The feature set of the current decision tree includes n training samples, such as the following three training samples.

[0075] X1=[θ(t)1,w(t)1,T(t)1,v(t)1,y1];

[0076] X2=[θ(t)2,w(t)2,T(t)2,v(t)2,y2];

[0077] X3=[θ(t)3, w(t)3, T(t)3, v(t)3, y3].

[0078] The current node selects m types of sample driving data from the feature set. Since each training sample includes the m types of sample driving data, n*m is the number of split items. The split item is a threshold, and the data set in the node is split into two subsets according to the threshold.

[0079] For example, if m is 2, and we choose θ(t), w(t), then the splitting terms are θ(t)1, w(t)1, θ(t)2, w(t)2, θ(t)3, w(t)3.

[0080] If θ(t)1 is used as the threshold, then in the feature set, the θ(t) values ​​greater than θ(t)1 are divided into one subset, and the θ(t) values ​​less than or equal to θ(t)1 are divided into another subset. If the feature set has both θ(t) values ​​greater than θ(t)1 and θ(t) values ​​less than or equal to θ(t)1, then it can be divided into two subsets; if the feature set only has θ(t) values ​​greater than θ(t)1, or only has θ(t) values ​​less than or equal to θ(t)1, then one of the subsets is an empty set.

[0081] The feature set of the decision tree is shown in Table 1.

[0082]

[0083] Table 1

[0084] If the splitting term is θ(t) = 30, then it can be divided into two subsets, one subset with θ(t) values ​​less than or equal to 30 is [30, 52, 3, 70, 2], and the other subset with θ(t) values ​​greater than 30 is [120, 50, 2, 60, -3], [60, 56, 5, 90, 5].

[0085] If the splitting term is θ(t) = 60, then it can be divided into two subsets, the ones with θ(t) values ​​less than or equal to 60 are [30, 52, 3, 70, 2], [60, 56, 5, 90, 5], and the ones with θ(t) values ​​greater than 60 are [120, 50, 2, 60, -3].

[0086] If the splitting term is θ(t) = 120, then it can be divided into a subset. The subsets with θ(t) values ​​less than or equal to 120 are [30, 52, 3, 70, 2], [60, 56, 5, 90, 5], [120, 50, 2, 60, -3], and the subsets with θ(t) values ​​greater than 120 are empty subsets.

[0087] Similarly, if w(t)1 is used as the threshold, then in the feature set, the w(t) values ​​greater than w(t)1 are divided into one subset, and the w(t) values ​​less than or equal to w(t)1 are divided into another subset. If the feature set has both w(t) values ​​greater than w(t)1 and w(t) values ​​less than or equal to w(t)1, then it can be divided into two subsets; if the feature set only has w(t) values ​​greater than w(t)1, or only has w(t) values ​​less than or equal to w(t)1, then one of the subsets is an empty set.

[0088] If the splitting term is w(t)=50, then it can be divided into two subsets, one subset with w(t) values ​​less than or equal to 50 is [120,50,2,60,-3], and the other subset with w(t) values ​​greater than 50 is [30,52,3,70,2], [60,56,5,90,5].

[0089] If the splitting term is w(t)=52, then it can be divided into two subsets, the ones with w(t) values ​​less than or equal to 52 are [120,50,2,60,-3], [30,52,3,70,2], and the ones with w(t) values ​​greater than 52 are [60,56,5,90,5].

[0090] If the splitting term is w(t)=56, then it can be divided into a subset. The subsets with w(t) values ​​less than or equal to 56 are [120,50,2,60,-3], [30,52,3,70,2], [60,56,5,90,5], and the subsets with w(t) values ​​greater than 56 are the empty set.

[0091] According to the above example, 3 training samples and 2 types of sample driving data can generate 6 split items, each split item splits the data set of the node into two subsets. Among them, this data set can be the original data set or the subset left after the last split. If the current node is the root node, then the data set is the feature set of the decision tree. If the current node is a non-root node, then the data set is the subset left after the parent node splits.

[0092] For each split item, the two subsets are split out, and the mean square error corresponding to the split item is calculated, and the split item corresponding to the minimum mean square error is selected from all the split items as the minimum split item.

[0093] The mean square error is an indicator to measure the difference between the predicted value and the actual value. Its calculation formula is:

[0094] ,in, represents the mean square error, N is the number of samples in node t, is the actual value of the sample in node t, is the predicted value (usually the mean) of the sample in node t. After splitting, node t is divided into two child nodes, and their mean square errors are:

[0095] ,in, Represents the left child node The mean square error, is the number of samples in the left child node, is the actual value in the left child node, is the predicted value in the left child (usually the mean).

[0096] ,in, Represents the right child node The mean square error, is the number of samples in the right child node, is the actual value in the right child node, is the predicted value in the right child (usually the mean).

[0097] is the sample ratio of the left child node, The sample ratio of child nodes.

[0098] The mean square error of the node is: .

[0099] For example, if the split term is θ(t)=30, the process of calculating MSE is:

[0100] 1. Divide into two nodes.

[0101] θ(t) values ​​less than or equal to 30: [30,52,3,70,2];

[0102] θ(t) values ​​greater than 30: [120,50,2,60,-3], [60,56,5,90,5].

[0103] 2. Calculate the MSE.

[0104] Left node (θ(t) value is less than or equal to 30): the average steering wheel deviation value is 2, is (2 - 2)^2=0.

[0105] Right node (θ(t) value is greater than 30): the average steering wheel deviation value is -3+5=2, It is {(-3 - 2)^2+(5 - 2)^2} / 2=17.

[0106] MSE= *0+ *17=11.33.

[0107] According to the above method, the MSE of six different split items is calculated respectively, and then the split item with the smallest MSE is selected as the minimum split item. Among all the generated split items, the one with the smallest mean square error is selected as the optimal split item, which can minimize the uncertainty and noise of the data.

[0108] After a node is split into two child nodes by the minimum splitting term, if the set stop condition is not met, the two child nodes need to continue to be split according to the splitting term. The splitting term based on which the child nodes are split can be any splitting term. Preferably, the types of splitting terms used for splitting different nodes are different. For example, the root node 1 is split into two child nodes, namely child node 2 and child node 3, using the minimum splitting term θ(t)1. Child node 2 is split into child node 4 and child node 5 using w(t)1. Child node 3 is split into child node 6 and child node 7 using T(t)2. Child node 4 is then split using v(t)1, and so on, until the set stop condition is met. Using different splitting terms to split multiple nodes in a decision tree can reduce the deviation caused by a single splitting strategy, making the overall output result of the random forest model more reliable.

[0109] In summary, by traversing the n training samples in the feature set and randomly selecting m types of sample data, n*m splitting terms can be obtained. Then, each splitting term is used as a threshold to split the data set in the node into two subsets, and the mean square error is calculated based on each subset to evaluate the splitting effect. Finally, the splitting term with the smallest mean square error is selected as the minimum splitting term. The minimum splitting term represents the most representative feature in the data, rather than relying solely on the features of certain specific samples. Therefore, selecting the minimum splitting term can reduce the model's excessive dependence on training data, thereby reducing the risk of overfitting. In addition, the minimum splitting term can divide the data set more accurately, so the random forest model based on these splitting terms can more accurately determine the steering wheel deviation value, improving the overall performance and accuracy of the model.

[0110] As an optional implementation method, recursively splitting each node of the decision tree until a set stopping condition is met includes: randomly selecting m types of sample driving data for the root node and finding the minimum splitting term, and using the minimum splitting term to split the data set in the root node into two subsets; creating a new child node for each of the split subsets; and executing the steps of selecting m types of sample driving data and splitting the data set based on the minimum splitting term for each child node until the set stopping condition is met.

[0111] The splitting process of the root node: first select m types of sample driving data from the n training samples in the feature set to obtain n*m split items, then split the root node data set into two subsets for each split item, and calculate the MSE of the data set by combining the two subsets, and use the MSE as the MSE corresponding to the split item, where the root node data set is the feature set of the decision tree. Then select the split item corresponding to the minimum MSE as the minimum split item, and finally use the minimum split item to split the root node data set into two subsets, and establish the child nodes corresponding to each subset.

[0112] The splitting process of the child node is similar to that of the root node: first, m types of sample driving data are selected from the n training samples in the feature set to obtain n*m split items, and then the data set of the child node is split into two subsets for each split item, and the MSE of the data set is calculated by combining the two subsets, and the MSE is used as the MSE corresponding to the split item, where the data set of the child node is the subset after the parent node is split, and then the split item corresponding to the minimum MSE is selected as the minimum split item, and finally the minimum split item is used to split the data set of the child node into two subsets, and the child node corresponding to each subset is established. The difference is that the data set of the child node is different from the data set of the root node. The data set of the root node is the feature set of the decision tree, and the data set of the child node is the subset after the parent node is split.

[0113] Each node continues to recursively split according to the above steps. If the set stop condition is met, the last layer of child nodes is used as leaf nodes. For example, if the maximum depth is 3 layers, the process stops after the first layer of child nodes is split, achieving: root node-child node-leaf node.

[0114] Among them, if any of the following three situations is met, it can be considered that the set stop condition is met, and the three situations are: the node depth reaches the maximum depth; the number of samples of the node is less than the preset minimum number of samples, and the purity of the node has reached the preset value.

[0115] As an optional implementation method, the real-time driving data of the target vehicle is processed by a random forest model to obtain the steering wheel zero bias value in the target vehicle, including: using each decision tree in the random forest model to process the real-time driving data of the target vehicle to obtain the steering wheel zero bias value output by each decision tree; taking the average of the steering wheel bias values ​​of each decision tree to obtain the steering wheel zero bias value in the target vehicle.

[0116] In the actual processing of the random forest model, the server extracts the real-time driving data at the current moment and inputs it into the random forest model. Each decision tree obtains a steering wheel zero bias value. The random forest model takes the average of the steering wheel zero bias values ​​output by each decision tree to obtain a comprehensive steering wheel zero bias value. This steering wheel zero bias value can reflect the comprehensive judgment of the entire random forest model on the driving status of the target vehicle.

[0117] The formula for calculating the random forest model is:

[0118] in, is the online output result of the random forest model, Q is the number of decision trees, i is the i-th decision tree, is the online output of the i-th decision tree for the real-time driving data X.

[0119] By constructing multiple decision trees and integrating their output results, the present application can effectively reduce the risk of overfitting of a single model, improve the generalization ability and stability of the overall model, and thus improve the accuracy of the steering wheel zero deviation warning system. In addition, the feature set of each decision tree is random, and the split items of the nodes in each decision tree are also random, so that the features used in the construction of different decision trees are random, which further improves the generalization ability of the overall model.

[0120] In step 103, the content of the prompt message sent by the server based on the steering wheel zero bias value includes: the server determines the correction amount of the steering wheel according to the final steering wheel zero bias value. Generally speaking, the correction amount is the negative number of the steering wheel zero bias value. Assuming that the steering return angle is 0 degrees, the left side is set to a negative value and the right side is set to a positive value.

[0121] If the steering wheel zero offset value is positive, the correction amount is a negative value. The driver is prompted by voice that "the steering wheel is detected to be offset to the right, please check and correct" to make the steering wheel move closer to the desired position. The warning message displayed on the instrument panel reminds the driver. Specifically, the driver can be prompted with the angle of the steering wheel offset to the right, so that the driver knows the specific angle of controlling the steering wheel.

[0122] If the steering wheel zero offset value is negative, the correction amount is a positive value. The driver will be prompted through a voice message that "the steering wheel is detected to be offset to the left, please check and correct" to make the steering wheel move closer to the desired position. The warning message displayed on the instrument panel reminds the driver. Specifically, the driver can be prompted with the angle of the steering wheel offset to the left, so that the driver knows the specific angle of controlling the steering wheel.

[0123] In this application, the working process of the steering wheel zero deviation reminder system is a highly integrated and automated process. It mainly realizes real-time monitoring and reminder of steering wheel zero deviation through five main links: sensor monitoring, data preprocessing, model training, online output and correction feedback. The following is a detailed explanation of the working process of the system:

[0124] 1. Data acquisition.

[0125] The steering wheel zero deviation reminder system is equipped with a steering wheel angle sensor and a vehicle speed sensor on the vehicle. The steering wheel angle sensor collects the actual steering wheel angle θ(t), and the vehicle speed sensor collects the actual vehicle speed v(t). The OBD-II diagnostic tool is used to read the steering wheel zero deviation value y at time t.

[0126] 2. Data preprocessing.

[0127] After data cleaning, the steering wheel speed is calculated based on the actual steering angle θ(t) and steering wheel torque , and then normalize the data.

[0128] Calculate the steering wheel speed: .

[0129] Calculate the steering wheel torque T(t)=μ*r*L*θ(t), where T(t) is the steering wheel torque at time t, μ is the friction coefficient between the tire and the ground, r is the tire radius, L is the length of the steering knuckle arm, and θ(t) is the steering wheel angle at time t.

[0130] 3. Model training.

[0131] The preprocessed data will be sent to the system's data processing module for further analysis. In the steering wheel zero deviation reminder system, random forest, as a powerful integrated learning algorithm, can effectively process complex nonlinear relationships and data. By constructing multiple decision trees and using preset algorithms and models to process the data, feature information related to steering wheel zero deviation can be extracted. The specific implementation process is as follows.

[0132] (1) Construct training samples for model training: X = [θ(t), w(t), T(t), v(t), y].

[0133] (2) By randomly selecting training samples with replacement, construct the feature set of each decision tree. For example, the feature set of decision tree 1 includes three training samples:

[0134] X1=[θ(t)1,w(t)1,T(t)1,v(t)1,y1];

[0135] X2=[θ(t)2,w(t)2,T(t)2,v(t)2,y2];

[0136] X3=[θ(t)3, w(t)3, T(t)3, v(t)3, y3].

[0137] (3) When each node of the decision tree is split, m sample driving data are selected to obtain n*m split items, where n is the number of training samples in the feature set. For example, if n is 3 and m is 2 (θ(t), w(t)), 6 split items are obtained, namely: θ(t)1, w(t)1, θ(t)2, w(t)2, θ(t)3, w(t)3. The mean square error corresponding to each split item is calculated, and the minimum split item is selected through the minimum mean square error. The minimum split item is used to split the node data set into two subsets, and a corresponding child node is established for each subset.

[0138] (4) Each node is recursively split according to step (3) until the set stopping condition is met, a decision tree is constructed and a random forest model is generated.

[0139] 4. Online output.

[0140] Real-time driving data is input into the random forest model, which integrates the results of each decision tree to output the steering wheel zero bias value.

[0141] 5. Correction feedback.

[0142] The driver is prompted by voice to steer the wheel closer to the desired position, and a warning message is displayed on the instrument panel to remind the driver.

[0143] In this application, in the online identification of the zero bias of the steering wheel, the random forest algorithm realizes the accurate, stable, and real-time identification of the zero bias of the steering wheel through technical points such as the advantages of ensemble learning, random sampling of samples and features, construction and fusion of decision trees, feature importance evaluation, real-time performance, and adaptability.

[0144] Based on the same technical concept, this application provides a device for reminding of the zero bias of the steering wheel, as Figure 3 shown. The device includes:

[0145] An acquisition module 301, configured to acquire real-time driving data of a target vehicle;

[0146] A processing module 302, configured to process the real-time driving data through a pre-trained random forest model to obtain the zero bias value of the steering wheel in the target vehicle;

[0147] A prompting module 303, configured to send a prompting message based on the zero bias value of the steering wheel, where the prompting message is used to prompt the driver to adjust the direction of the steering wheel.

[0148] Optionally, the device further includes:

[0149] A construction module, configured to construct the feature set of each decision tree by randomly sampling training samples with replacement, where each training sample includes the sample driving data of the vehicle and the corresponding steering wheel deviation value;

[0150] A selection module, configured to randomly select m types of sample driving data for a node when splitting each node of each decision tree, where m < M, and M is the number of categories of sample driving data in the training samples;

[0151] A traversal module, configured to traverse the m types of sample driving data in the feature set, find the minimum splitting item by trying multiple splitting items, and split the node based on the minimum splitting item, where the splitting item is used to split the data set in the node into two subsets, and the data set is the feature set or the subset after the previous split;

[0152] A recursion module, configured to recursively split each node of the decision tree until a set stop condition is met, and complete the construction of the decision tree and the generation of the random forest model.

[0153] Optionally, the traversal module is configured to:

[0154] Traverse n training samples in the feature set and m types of sample driving data in the n training samples to obtain n * m splitting items;

[0155] The following operations are performed for each splitting item: the node data set is split into two subsets using the splitting item, and the mean square error of the node is calculated based on the split subsets, where the mean square error refers to the mean square error of the steering wheel deviation value in the node data set;

[0156] The split item corresponding to the minimum mean square error is selected from all split items as the minimum split item.

[0157] Optionally, the traversal module is used to:

[0158] Calculate the mean square error of the steering wheel deviation values ​​in each subset;

[0159] Determine the sample proportion of the subset according to the sample size in the subset and the sample size in the node;

[0160] Taking the sample proportion as the weight, the mean square error of the two subsets is weighted and summed to obtain the mean square error of the node.

[0161] Optionally, the recursive module is used to:

[0162] Randomly select m types of sample driving data for the root node and find the minimum split term, and use the minimum split term to split the data set in the root node into two subsets;

[0163] Create new child nodes for each subset into which it is split;

[0164] The steps of selecting m types of sample driving data and splitting the data set based on the minimum splitting term are executed for each child node until the set stopping condition is met.

[0165] Optionally, the processing module 302 is used to:

[0166] Each decision tree in the random forest model is used to process the real-time driving data of the target vehicle to obtain the steering wheel zero bias value output by each decision tree;

[0167] The steering wheel deviation value of each decision tree is averaged to obtain the steering wheel zero deviation value in the target vehicle.

[0168] Optionally, the acquisition module 301 is used to:

[0169] The steering wheel angle is collected by using a steering angle sensor deployed on the vehicle, and the actual driving speed of the vehicle is collected by using a speed sensor deployed on the vehicle;

[0170] The steering wheel speed is calculated according to the steering wheel angle and the time interval of the steering wheel rotation;

[0171] The steering wheel torque is calculated according to the steering wheel angle, a preset tire friction coefficient, a preset tire radius, and a preset steering knuckle arm length.

[0172] As Figure 4 shown, an embodiment of the present application provides an electronic device, including a processor 401, a communication interface 402, a memory 403, and a communication bus 404. Among them, the processor 401, the communication interface 402, and the memory 403 complete mutual communication through the communication bus 404.

[0173] The memory 403 is used to store computer programs.

[0174] In an embodiment of the present application, when the processor 401 is used to execute the program stored on the memory 403, the steering wheel zero-offset reminder method provided by any one of the foregoing method embodiments is implemented as follows.

[0175] Obtain the real-time driving data of the target vehicle;

[0176] Process the real-time driving data through a pre-trained random forest model to obtain the steering wheel zero-offset value in the target vehicle;

[0177] Send a reminder message based on the steering wheel zero-offset value, where the reminder message is used to prompt the driver to adjust the direction of the steering wheel.

[0178] Optionally, before processing the real-time driving data through a pre-trained random forest model, the method further includes:

[0179] Construct the feature set of each decision tree by randomly sampling training samples with replacement, where each training sample includes the sample driving data of the vehicle and the corresponding steering wheel deviation value;

[0180] When each node of each decision tree is split, randomly select m types of sample driving data for the node, where m < M, and M is the number of categories of sample driving data in the training samples;

[0181] Traverse the m types of sample driving data in the feature set, find the minimum splitting item by trying multiple splitting items, and split the node based on the minimum splitting item, where the splitting item is used to split the data set in the node into two subsets, and the data set is the feature set or the subset after the previous split;

[0182] Recursively split each node of the decision tree until the set stop condition is met, and complete the construction of the decision tree and the generation of the random forest model.

[0183] Optionally, traversing the m types of sample driving data in the feature set and finding the minimum splitting item by trying multiple splitting items includes:

[0184] Traverse the n training samples in the feature set and the m types of sample driving data in the n training samples to obtain n * m splitting items;

[0185] The following operations are performed for each splitting item: the node data set is split into two subsets using the splitting item, and the mean square error of the node is calculated based on the split subsets, where the mean square error refers to the mean square error of the steering wheel deviation value in the node data set;

[0186] The split item corresponding to the minimum mean square error is selected from all split items as the minimum split item.

[0187] Optionally, calculating the mean square error of the node according to the split subsets includes:

[0188] Calculate the mean square error of the steering wheel deviation values ​​in each subset;

[0189] Determine the sample proportion of the subset according to the sample size in the subset and the sample size in the node;

[0190] Taking the sample proportion as the weight, the mean square error of the two subsets is weighted and summed to obtain the mean square error of the node.

[0191] Optionally, recursively split each node of the decision tree until a set stop condition is met, including:

[0192] Randomly select m types of sample driving data for the root node and find the minimum split term, and use the minimum split term to split the data set in the root node into two subsets;

[0193] Create new child nodes for each subset into which it is split;

[0194] The steps of selecting m types of sample driving data and splitting the data set based on the minimum splitting term are executed for each child node until the set stopping condition is met.

[0195] Optionally, the real-time driving data of the target vehicle is processed by a random forest model to obtain a steering wheel zero bias value in the target vehicle including:

[0196] Each decision tree in the random forest model is used to process the real-time driving data of the target vehicle to obtain the steering wheel zero bias value output by each decision tree;

[0197] The steering wheel deviation value of each decision tree is averaged to obtain the steering wheel zero deviation value in the target vehicle.

[0198] Optionally, obtaining sample driving data of the vehicle includes:

[0199] The steering wheel angle is collected by using a steering angle sensor deployed on the vehicle, and the actual driving speed of the vehicle is collected by using a speed sensor deployed on the vehicle;

[0200] The steering wheel speed is calculated according to the steering wheel angle and the time interval of the steering wheel rotation;

[0201] The steering wheel torque is calculated according to the steering wheel angle, a preset tire friction coefficient, a preset tire radius, and a preset steering knuckle arm length.

[0202] An embodiment of the present application also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the steering wheel zero deviation reminder method provided in any of the aforementioned method embodiments are implemented.

[0203] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0204] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a general hardware platform, and of course, by hardware. Based on this understanding, the above technical solution, in essence, or the part that contributes to the relevant technology, can be embodied in the form of a software product, which can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiment.

[0205] It should be understood that the terms used herein are only for the purpose of describing specific example embodiments and are not intended to be limiting. Unless the context clearly indicates otherwise, the singular forms "one", "an" and "said" as used herein may also be meant to include plural forms. The terms "include", "comprise", "contain", and "have" are inclusive, and therefore specify the existence of stated features, steps, operations, elements and / or parts, but do not exclude the existence or addition of one or more other features, steps, operations, elements, parts, and / or combinations thereof. The method steps, processes, and operations described herein are not interpreted as necessarily requiring them to be performed in the specific order described or illustrated, unless the execution order is clearly indicated. It should also be understood that additional or alternative steps may be used.

[0206] The foregoing is merely a specific embodiment of the present invention, which enables those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A steering wheel zero deviation reminder method, characterized in that: The method includes: Obtaining real-time driving data of the target vehicle; Processing the real-time driving data through a pre-trained random forest model to obtain the zero-offset value of the steering wheel in the target vehicle; Sending a prompt message based on the zero-offset value of the steering wheel, where the prompt message is used to prompt the driver to adjust the direction of the steering wheel; Wherein, processing the real-time driving data of the target vehicle through the random forest model to obtain the zero-offset value of the steering wheel in the target vehicle includes: Using each decision tree in the random forest model to process the real-time driving data of the target vehicle to obtain the zero-offset value of the steering wheel output by each decision tree; Taking the mean value of the steering wheel deviation values of each decision tree to obtain the zero-offset value of the steering wheel in the target vehicle.

2. The method according to claim 1, characterized in that The pre-training process of the random forest model includes: Constructing the feature set of each decision tree by randomly sampling training samples with replacement, where each training sample includes the sample driving data of the vehicle and the corresponding steering wheel deviation value; When splitting each node of each decision tree, randomly selecting m types of sample driving data for the node, where m < M, and M is the number of categories of sample driving data in the training samples; Traversing the m types of sample driving data in the feature set, finding the minimum splitting item by trying multiple splitting items, and splitting the node based on the minimum splitting item, where the splitting item is used to split the data set in the node into two subsets, and the data set is the feature set or the subset after the previous split; Recursively splitting each node of the decision tree until a set stop condition is met, completing the construction of the decision tree and the generation of the random forest model.

3. The method according to claim 2, characterized in that Traversing the m types of sample driving data in the feature set and finding the minimum splitting item by trying multiple splitting items includes: Traversing the n training samples in the feature set and the m types of sample driving data in the n training samples to obtain n * m splitting items; Performing the following operations for each splitting item: Using the splitting item to split the data set of the node into two subsets, and calculating the mean square error of the node according to the split subsets, where the mean square error refers to the mean square error of the steering wheel deviation values in the data set of the node; Selecting the splitting item corresponding to the minimum mean square error from all splitting items as the minimum splitting item.

4. The method according to claim 3, characterized in that Calculating the mean square error of the node according to the split subsets includes: Calculating the mean square error of the steering wheel deviation values in each subset; Determining the sample proportion of the subset according to the sample size in the subset and the sample size in the node; Taking the sample proportion as the weight, performing a weighted sum of the mean square errors of the two subsets to obtain the mean square error of the node.

5. The method according to claim 2, characterized in that: Recursively splitting each node of the decision tree until a set stop condition is met includes: Randomly selecting m types of sample driving data for the root node and finding the minimum splitting item, and using the minimum splitting item to split the data set in the root node into two subsets; Creating new child nodes for each subset after splitting; The steps of selecting m types of sample driving data and splitting the data set based on the minimum splitting term are executed for each of the subnodes until a set stop condition is met.

6. The method according to claim 2, characterized in that Acquiring real-time driving data or sample driving data of a vehicle includes: A steering wheel angle sensor is used to collect the steering wheel angle of the vehicle, and a vehicle speed sensor is used to collect the actual driving speed of the vehicle; Calculate the steering wheel speed according to the steering wheel angle and the time interval of the steering wheel rotation; The steering wheel torque is calculated according to the steering wheel angle, a preset tire friction coefficient, a preset tire radius and a preset steering knuckle arm length.

7. A steering wheel zero deviation reminder device, characterized in that: The device comprises: An acquisition module is used to acquire real-time driving data of a target vehicle; A processing module, used for processing the real-time driving data through a pre-trained random forest model to obtain a steering wheel zero bias value in the target vehicle; a prompt module, configured to issue a prompt message based on the steering wheel zero offset value, wherein the prompt message is used to prompt the driver to adjust the direction of the steering wheel; Wherein, the processing module is used for: Using each decision tree in the random forest model to process the real-time driving data of the target vehicle to obtain a steering wheel zero bias value output by each decision tree; The steering wheel deviation values ​​of each decision tree are averaged to obtain the steering wheel zero deviation value in the target vehicle.

8. An electronic device, characterized in that: It includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor, for implementing any of the methods described in claims 1-6 when executing a program stored in a memory.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images