Vehicle wheel end braking force determination method, apparatus, device, medium, and product

Abstract

This application relates to the field of vehicle braking technology, and more particularly to a method, device, equipment, medium, and product for determining vehicle wheel-end braking force. The method includes: periodically collecting wheel-end motor parameter data, wherein the wheel-end motor parameter data includes at least the motor armature current, rotor speed, and output shaft angle; preprocessing the wheel-end motor parameter data to obtain a time-series dataset; extracting features from the time-series dataset to obtain a one-dimensional feature vector; and performing time-series prediction on the one-dimensional feature vector to obtain an estimated wheel-end braking force. This application facilitates reducing the cost of vehicle wheel-end braking force detection and improving the reliability of vehicle wheel-end braking force detection while ensuring the accuracy of vehicle wheel-end braking force detection.

Description

Technical Field

[0001] This application relates to the field of vehicle braking technology, and in particular to a method, device, equipment, medium and product for determining the braking force at the wheel end of a vehicle. Background Technology

[0002] When an electric vehicle brakes, it can control the wheel-end calipers, such as EMB calipers (ElectroMechanical BrakeCaliper), to clamp the brake pads, thereby achieving wheel braking. The braking force generated during braking can provide data reference for real-time vehicle functions such as adjusting motor torque / position, compensating for brake pad wear, and diagnosing vehicle faults. Therefore, it is necessary to determine the braking force generated at the wheel end when the caliper clamps the brake pads.

[0003] Currently, the method for determining the braking force at the wheel end of a vehicle is as follows: when the caliper clamps the brake pads, the braking force at the wheel end of the vehicle is detected by the braking force sensor installed on the caliper.

[0004] However, firstly, the cost of brake force sensors is high, resulting in high costs for vehicle wheel-end brake force detection; secondly, brake force sensors are exposed to the harsh working environment of vehicle wheels for a long time. After long-term use, the detection accuracy of brake force sensors will gradually decrease due to factors such as wear and environmental corrosion, and it is difficult to calibrate in real time, resulting in poor reliability of vehicle wheel-end braking force detection. Summary of the Invention

[0005] To reduce the cost of vehicle wheel-end braking force detection and improve the reliability of vehicle wheel-end braking force detection while ensuring the accuracy of vehicle wheel-end braking force detection, this application provides a method, device, equipment, medium, and product for determining vehicle wheel-end braking force.

[0006] In a first aspect, this application provides a method for determining the wheel-end braking force of a vehicle, including: Periodically collect wheel-end motor parameter data, which includes at least the motor armature current, rotor speed, and output shaft angle; The wheel-end motor parameter data is preprocessed to obtain a time-series dataset; Feature extraction is performed on the time-series dataset to obtain a one-dimensional feature vector; The one-dimensional feature vector is used for time-series prediction to obtain the wheel-end braking force estimate.

[0007] Secondly, this application provides a vehicle wheel-end braking force determination device, comprising: The data acquisition module is used to periodically acquire wheel-end motor parameter data, which includes at least the motor armature current, rotor speed, and output shaft angle. The data preprocessing module is used to preprocess the wheel-end motor parameter data to obtain a time-series dataset; The feature extraction module is used to extract features from the time-series dataset to obtain a one-dimensional feature vector; The data prediction module is used to perform time-series prediction on the one-dimensional feature vector to obtain the wheel-end braking force estimate.

[0008] Thirdly, this application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the method described above.

[0009] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the above-described method.

[0010] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the steps in any of the above method embodiments.

[0011] The aforementioned method, apparatus, equipment, medium, and product for determining vehicle wheel-end braking force involves periodically collecting wheel-end motor parameter data, which includes at least the motor armature current, rotor speed, and output shaft angle; preprocessing the wheel-end motor parameter data to obtain a time-series dataset; extracting features from the time-series dataset to obtain a one-dimensional feature vector; and performing time-series prediction on the one-dimensional feature vector to obtain an estimated wheel-end braking force. Through the above implementation, firstly, wheel-end braking force can still be calculated without using braking force sensors at the wheel ends, thus eliminating the need for braking force sensors and effectively reducing the cost of wheel-end braking force detection; secondly, the process of collecting wheel-end motor parameter data is largely unaffected by the harsh environment at the wheel ends, ensuring the stability of the wheel-end braking force calculation process based on this data and effectively improving the reliability of wheel-end braking force detection; thirdly, the wheel-end motor parameter data is preprocessed before being used to calculate the wheel-end braking force, ensuring the accuracy of the braking force detection. In summary, this application can reduce the cost of wheel-end braking force detection and improve the reliability of wheel-end braking force detection while ensuring the accuracy of the detection.

[0012] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent from the following description. Attached Figure Description

[0013] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 This is a flowchart of a method for determining the wheel-end braking force of a vehicle provided in an embodiment of this application; Figure 2 This is a schematic diagram of the structure of a vehicle wheel-end braking force determination device provided in the embodiments of this application; Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application; Figure 4 This is an internal structural diagram of a computer-readable storage medium provided in an embodiment of this application. Detailed Implementation

[0015] To make the objectives, technical solutions, and advantages of this disclosure clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this disclosure.

[0016] It should be noted that the terms "first," "second," etc., used in the specification, claims, and accompanying drawings herein are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, apparatus, product, or device that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0017] In this article, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, and B alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0018] Example 1 Figure 1 This is a flowchart of a method for determining the wheel-end braking force of a vehicle, provided in Embodiment 1 of this application. (Refer to...) Figure 1The method can be executed by a device that performs the method, which can be implemented in software and / or hardware, and the method includes: S110. Periodically collect wheel-end motor parameter data, which includes at least the motor armature current, rotor speed, and output shaft angle.

[0019] It should be noted that this embodiment aims to reduce the cost of vehicle wheel-end braking force detection and improve the reliability of vehicle wheel-end braking force detection while ensuring the accuracy of vehicle wheel-end braking force detection. To this end, this embodiment no longer sets a braking force sensor at the wheel end and directly detects the wheel-end braking force through the braking force sensor. Instead, it selects to estimate the wheel-end braking force based on some collected wheel-end data.

[0020] To collect the aforementioned wheel-end data, this embodiment incorporates multiple sensors within the wheel-end MCU. These sensors include a current sensor, a speed sensor, and a rotation angle sensor. In other embodiments, the sensors may further include a voltage sensor, etc., without limitation. The current sensor periodically collects the motor armature current, the speed sensor periodically collects the rotor speed, and the rotation angle sensor periodically collects the output shaft angle. All sensors periodically collect corresponding data at the same sampling frequency. In this embodiment, the sampling frequency ranges from 0.5kHz to 2kHz; for example, this embodiment uses a sampling frequency of 1kHz. In other embodiments, the selection range and specific value of the sampling frequency are not limited. The unit for motor armature current is "A", the unit for rotor speed is "rpm", and the unit for output shaft angle is "°". In this embodiment, each sampling cycle may collect one set of motor armature current, rotor speed, and output shaft angle data, and these multiple sets of collected data are recorded as wheel-end motor parameter data.

[0021] After the wheel-end motor parameter data is collected once, the collected wheel-end motor parameter data is further stored in the buffer area of ​​the wheel-end MCU. The buffer area has a relatively large capacity of 1024KB, which can store 1024 sets of data to prevent data overflow.

[0022] S120. Preprocess the wheel-end motor parameter data to obtain a time-series dataset.

[0023] It should be noted that the collected wheel-end motor parameter data will be used to calculate the wheel-end braking force. However, the initially collected wheel-end motor parameter data may contain adverse factors such as outliers, noise, and dimensional effects. If the wheel-end motor parameter data is used directly to calculate the wheel-end braking force, the accuracy of the calculated wheel-end braking force will be poor. Therefore, this embodiment intends to preprocess the wheel-end motor parameter data to improve its data accuracy.

[0024] In this embodiment, the preprocessing includes outlier detection and removal, noise filtering, and dimensional unification. In other embodiments, the specifics are not limited. The new data obtained after preprocessing the wheel-end motor parameter data is denoted as the time-series dataset.

[0025] S130. Perform feature extraction on the time series dataset to obtain a one-dimensional feature vector.

[0026] It should be noted that, in order to further calculate the wheel-end braking force based on the time-series dataset, this embodiment pre-configures a neural network. For example, the neural network is a CNN-LSTM composite neural network architecture. In other embodiments, the specific architecture is not limited. The CNN-LSTM composite neural network architecture is used to process the time-series dataset to calculate the wheel-end braking force.

[0027] The CNN-LSTM composite neural network architecture includes a CNN feature extraction layer, which has a structure of 3 convolutional layers + 2 pooling layers. Its input dimension is [batch_size, time_step, feature_num]. Here, batch_size is set to 32 (to balance computing power and real-time performance), time_step is set to 50 (corresponding to a 50ms sliding window, matching a 1kHz sampling frequency to ensure the integrity of time-series data), and feature_num is set to 3 (corresponding to the three input parameters: motor armature current, rotor speed, and output shaft angle).

[0028] The aforementioned time series dataset is first input into the CNN feature extraction layer for processing. The CNN feature extraction layer is used to extract features from the input time series dataset to obtain deep features of the three input parameters. Each deep feature is then processed into a one-dimensional vector so that the LSTM time series prediction layer in the subsequent CNN-LSTM composite neural network architecture can further process each deep feature. This one-dimensional vector is denoted as a one-dimensional feature vector.

[0029] S140. Perform time-series prediction on the one-dimensional feature vector to obtain the wheel-end braking force estimate.

[0030] It should be noted that the LSTM temporal prediction layer in the CNN-LSTM composite neural network architecture consists of one LSTM layer and one fully connected layer, with an input dimension of [32, 832] (corresponding to the output of the CNN feature extraction layer). The number of hidden layer units is set to 128 (to balance prediction accuracy and computational power; experiments have verified that 128 units can achieve high-precision prediction with a memory footprint of ≤64KB). Other parameter settings for this LSTM temporal prediction layer include: the forget gate, input gate, and output gate all use the Sigmoid activation function (output range 0-1, facilitating information control); cell state updates use the tanh activation function (output range -1-1, mitigating gradient explosion); and the dropout coefficient is set to 0.2 (to prevent overfitting, improve generalization ability, and adapt to braking data from different vehicle models and operating conditions). The recurrent_dropout coefficient is set to 0.1 (to reduce overfitting of time series data); Fully connected layer: input dimension is 128, output dimension is 1 (corresponding to a single braking force estimate), activation function is Linear (linear activation, adapted to the prediction of continuous braking force values, with no output range limitation), weight initialization is Xavier normal initialization, and bias initialization is set to 0.

[0031] The LSTM temporal prediction layer is used to process the one-dimensional feature vector, thereby performing temporal prediction on the one-dimensional feature vector to obtain the estimated wheel-end braking force, and the estimated wheel-end braking force is recorded as the wheel-end braking force estimate.

[0032] It should be noted that in this embodiment, wheel-end motor parameter data is collected periodically. The wheel-end motor parameter data includes at least the motor armature current, rotor speed, and output shaft angle. The wheel-end motor parameter data is preprocessed to obtain a time-series dataset. Features are extracted from the time-series dataset to obtain a one-dimensional feature vector. The one-dimensional feature vector is then used for time-series prediction to obtain the wheel-end braking force estimate. Through the above implementation, firstly, wheel-end braking force can still be calculated without using braking force sensors at the wheel ends, thus eliminating the need for braking force sensors and effectively reducing the cost of wheel-end braking force detection; secondly, the process of collecting wheel-end motor parameter data is largely unaffected by the harsh environment at the wheel ends, ensuring the stability of the wheel-end braking force calculation process based on this data and effectively improving the reliability of wheel-end braking force detection; thirdly, the wheel-end motor parameter data is preprocessed before being used to calculate the wheel-end braking force, ensuring the accuracy of the braking force detection. In summary, this application can reduce the cost of wheel-end braking force detection and improve the reliability of wheel-end braking force detection while ensuring the accuracy of the detection.

[0033] Example 2 This application provides a method for determining vehicle wheel-end braking force in Embodiment 2. This method optimizes the "preprocessing of the wheel-end motor parameter data to obtain a time-series dataset" in Embodiment 1. It should be noted that for parts not described in detail in this embodiment, please refer to the descriptions in other embodiments. The method includes: S210. Periodically collect wheel-end motor parameter data, which includes at least the motor armature current, rotor speed, and output shaft angle.

[0034] S221. Perform outlier processing on the wheel-end motor parameter data to obtain de-outlier data.

[0035] It should be noted that there may be a certain number of outliers in the wheel-end motor parameter data. The presence of outliers will affect the accuracy of subsequent wheel-end braking force calculation. Therefore, this embodiment aims to remove outliers in the wheel-end motor parameter data, that is, to implement outlier processing for the wheel-end motor parameter data.

[0036] Specifically, the wheel-end motor parameter data contains three sets of input parameters. In this embodiment, the 3σ criterion is used to remove outliers. The mean μ and standard deviation σ of each set of input parameters are calculated. When the value of a certain set of data exceeds the range of [μ-3σ, μ+3σ], it is determined to be an outlier. The mean of the adjacent 5 sets of data is used to replace it (to avoid outliers affecting model training and prediction). After the outlier processing of the wheel-end motor parameter data, new data is obtained, which is recorded as the outlier-free data.

[0037] S222. The abnormal data is filtered to remove noise, resulting in denoised data.

[0038] It should be noted that the abnormal data may contain noise caused by factors such as electromagnetic interference and mechanical vibration. The presence of this noise will also affect the accuracy of the subsequent wheel-end braking force calculation. Therefore, this embodiment aims to remove the noise from the abnormal data.

[0039] Specifically, this embodiment uses the moving average filtering method, with the sliding window size set to 5ms (corresponding to 5 groups of data). The calculation formula is: x_filtered(i) = (x(i-2) + x(i-1) + x(i) + x(i+1) + x(i+2)) / 5, where x(i) is the i-th group of original data, and x_filtered(i) is the i-th group of filtered data, used to filter high-frequency noise (such as current fluctuations caused by electromagnetic interference). Each data in the anomaly data is processed in the above manner; and the new data obtained after the anomaly data has been filtered to remove noise is recorded as the denoised data.

[0040] S223. Normalize the denoised data to obtain normalized data.

[0041] It should be noted that the three parameters of motor armature current, electric rotor speed, and output shaft angle in the denoised data have different dimensions. To avoid any one parameter having too large a weight in the model prediction, it is necessary to eliminate the influence of dimensions. Therefore, this embodiment intends to normalize all data in the denoised data.

[0042] Specifically, taking one data point from one of the parameters as an example, this embodiment uses the Min-Max normalization method to map the data to the [0,1] interval to eliminate the influence of dimensions. The normalization formula is: x_norm=(x-x_min) / (x_max-x_min), where x is one data point from the above-mentioned parameter, x_min is the minimum value of the parameter (current: 0A, speed: 0rpm, rotation angle: 0°), x_max is the maximum value of the parameter (current: 50A, speed: 3000rpm, rotation angle: 90°), and x_norm is the normalized data. The new data obtained after normalizing all data points in the denoised data is denoted as the normalized data.

[0043] S224. Perform time-series concatenation on the normalized data to obtain a time-series dataset.

[0044] It should be noted that, in order to facilitate further feature extraction by the CNN feature extraction layer, the normalized data also needs to be spliced ​​in time sequence.

[0045] Specifically, the three parameter data in the normalized data are concatenated in chronological order to form a time-series dataset. A 50ms sliding window (corresponding to 50 sets of data) is used to truncate the data, resulting in input data with dimensions [batch_size, time_step, feature_num], which is used to input the CNN feature extraction layer. The new data obtained by concatenating the normalized data in time sequence is denoted as the time-series dataset.

[0046] It should be noted that during the preprocessing of the wheel-end motor parameter data, by sequentially performing outlier processing, noise filtering, and normalization on the wheel-end motor parameter data, outliers and noise in the wheel-end motor parameter data can be effectively removed, and the dimensional influence caused by different parameters can be eliminated. This facilitates the improvement of the accuracy of the preprocessed wheel-end motor parameter data, and in turn, facilitates the improvement of the accuracy of the subsequently calculated wheel-end braking force.

[0047] S230. Perform feature extraction on the time series dataset to obtain a one-dimensional feature vector.

[0048] S240. Perform time-series prediction on the one-dimensional feature vector to obtain the wheel-end braking force estimate.

[0049] Example 3 This application provides a method for determining vehicle wheel-end braking force in Embodiment 3. This method optimizes the step of "extracting features from the time-series dataset to obtain a one-dimensional feature vector" in Embodiment 1. It should be noted that for parts not detailed in this embodiment, please refer to the descriptions in other embodiments. The method includes: S310. Periodically collect wheel-end motor parameter data, which includes at least the motor armature current, rotor speed, and output shaft angle.

[0050] S320. Preprocess the wheel-end motor parameter data to obtain a time-series dataset.

[0051] S331. Convolve the time series dataset to obtain local features of the data.

[0052] It should be noted that the CNN feature extraction layer is used to extract features from time-series datasets; in this embodiment, the CNN feature extraction layer includes three sub-layers: a convolutional layer, a pooling layer, and a feature flattening layer; in other embodiments, the specific implementation is not limited.

[0053] The convolutional layer is used to perform cross-correlation operations on the input time series dataset using the built-in convolutional kernel, thereby obtaining the local features of the time series dataset, and these local features are denoted as data local features.

[0054] In this embodiment, the local features of the data include: current fluctuation features, speed change slope, and angular displacement difference. In other embodiments, the specific features are not limited. The calculation formula for the cross-correlation operation is: y_conv=σ(W_conv×x_in + b_conv), where W_conv is the convolution kernel weight, x_in is the data in the time series dataset input to the convolutional layer, b_conv is the bias, and σ is the ReLU activation function.

[0055] S332. Pool the local features of the data to obtain the key features of the data.

[0056] The pooling layer in the CNN feature extraction layer is used to pool the local features of the data output by the convolutional layer. By pooling the local features of the data, the key features in the local features of the data can be retained and redundant information can be filtered out. The key features retained after the local features of the data are pooled are called the key features of the data.

[0057] It should be noted that by pooling the local features of the data, key features can be retained and redundant features can be filtered out. This results in a significant reduction in the amount of feature data and an increase in the proportion of key features compared to local features. The key features are then used to calculate the wheel-end braking force, which effectively improves the calculation efficiency and accuracy of the wheel-end braking force.

[0058] S333. Perform feature flattening on the key features of the data to obtain a one-dimensional feature vector.

[0059] It should be noted that the key features of the data need to be input into the LSTM time series prediction layer for further processing. To do this, the key features of the data need to be converted from three-dimensional feature data to one-dimensional feature data, that is, feature flattening of the key features of the data.

[0060] Among them, the feature flattening layer in the CNN feature extraction layer can be used to flatten the key features of the input data, thereby converting the key features of the data from three-dimensional feature data to one-dimensional feature data, and the converted data is also recorded as a one-dimensional feature vector.

[0061] S340. Perform time-series prediction on the one-dimensional feature vector to obtain the wheel-end braking force estimate.

[0062] Example 4 This application provides a method for determining the wheel-end braking force of a vehicle in Embodiment 4. This method optimizes the "time-series prediction of the one-dimensional feature vector to obtain the wheel-end braking force estimate" in Embodiment 1. It should be noted that for parts not described in detail in this embodiment, please refer to the descriptions in other embodiments. The method includes: S410. Periodically collect wheel-end motor parameter data, which includes at least the motor armature current, rotor speed, and output shaft angle.

[0063] S420. Preprocess the wheel-end motor parameter data to obtain a time-series dataset.

[0064] S430. Perform feature extraction on the time series dataset to obtain a one-dimensional feature vector.

[0065] S441. Attention weighting is applied to the one-dimensional feature vector to obtain a weighted feature vector.

[0066] It should be noted that a one-dimensional feature vector corresponds to a set of three sets of parameter data that are time-series concatenated. In this embodiment, multiple sets of three sets of parameter data that are time-series concatenated can be obtained as time progresses, which means that multiple one-dimensional feature vectors can be obtained sequentially. Each one-dimensional feature vector corresponds to a braking condition, which is subsequently used to calculate the wheel-end braking force. Different braking conditions have different importance. For example, braking conditions such as sudden current changes and sudden speed drops during emergency braking are more important. In order to make the LSTM time-series prediction layer pay more attention to the more important braking conditions (one-dimensional feature vectors) in the subsequent calculation of wheel-end braking force, this embodiment also adds an attention mechanism module between the CNN feature extraction layer and the LSTM time-series prediction layer. This attention mechanism module is used to weight the one-dimensional feature vector according to its importance, and the weighted one-dimensional feature vector is a new vector, which is denoted as the weighted feature vector.

[0067] Specifically, taking one of the one-dimensional feature vectors as an example, the one-dimensional feature vector is mapped using three preset sets of independent learnable parameters to obtain the query vector q, key vector k, and value vector v corresponding to the one-dimensional feature vector. The weight matrix of the query vector q in relation to the independent learnable parameters is the query vector weight matrix W_q, the weight matrix of the key vector k in relation to the independent learnable parameters is the key vector weight matrix W_k, and the weight matrix of the value vector v in relation to the independent learnable parameters is the value vector weight matrix W_v. The formula for calculating the attention weight used in the weighted calculation of this one-dimensional feature vector is: α = softmax((W_q×q + b_q)×(W_k ×k+b_k)^T / √d_k), where b_q is the bias of the query vector q in relation to the independent learnable parameters, b_k is the bias of the key vector k in relation to the independent learnable parameters, d_k is the dimension of the key vector k, and α is the attention weight.

[0068] The weighting formula for weighting the one-dimensional feature vector using attention weight α is: v_att = α × (W_v × v + b_v), where W_v is the weight matrix corresponding to the value vector v in the independent learnable parameters, b_v is the bias, v is the value vector, i.e., the one-dimensional feature vector, and v_att is the weighted feature vector.

[0069] S442. Perform time-series prediction on the weighted feature vector to obtain the wheel-end braking force estimate.

[0070] It should be noted that the LSTM temporal prediction layer is used to perform temporal prediction on the weighted feature vector to obtain the estimated wheel braking force, that is, the wheel-end braking force estimate.

[0071] It should be noted that the self-attention module in step S441 can assign greater attention weights to important one-dimensional feature vectors, thereby enabling the corresponding weighted feature vectors to receive more attention when processed by the LSTM temporal prediction layer, which facilitates the improvement of the calculation accuracy of wheel-end braking force estimation.

[0072] Example 5 This application provides a method for determining vehicle wheel-end braking force in Embodiment 5. This method optimizes the "time-series prediction of the weighted feature vector to obtain an estimated wheel-end braking force" in Embodiment 4. It should be noted that for parts not described in detail in this embodiment, please refer to the descriptions in other embodiments. The method includes: S410. Periodically collect wheel-end motor parameter data, which includes at least the motor armature current, rotor speed, and output shaft angle.

[0073] S420. Preprocess the wheel-end motor parameter data to obtain a time-series dataset.

[0074] S430. Perform feature extraction on the time series dataset to obtain a one-dimensional feature vector.

[0075] S441. Attention weighting is applied to the one-dimensional feature vector to obtain a weighted feature vector.

[0076] S542A: Based on the forget gate output and input gate output corresponding to the weighted feature vector, and the cell state at the previous time step, update the cell state to obtain the current cell state.

[0077] It should be noted that in the LSTM temporal prediction layer, the output of the forget gate, the output of the input gate, and the cell state at the previous time step can be used to update the cell state, thereby obtaining the updated cell state.

[0078] The forget gate output is calculated as follows: f_t = σ(W_f × [h_{t-1}, v_att] + b_f), where h_{t-1} is the hidden layer output of the previous time step, v_att is the attention-weighted feature vector of the current time step, W_f is the forget gate weight matrix, b_f is the forget gate bias, and f_t is the forget gate output of the current time step. The input gate output is calculated as follows: i_t = σ(W_i × [h_{t-1}, v_att] + b_i), c_t' = tanh(W_c × [h_{t-1}, v_att] + b_c), where i_t is the input gate output of the current time step, c_t' is the candidate cell state, W_i and W_c are the weight matrices of the input gate and candidate cell state, respectively, and b_i and b_c are the corresponding biases. The current cell state is calculated as follows: c_t = f_t×c_{t-1}+i_t×c_t', where c_{t-1} is the cell state at the previous time step, and c_t is the current cell state.

[0079] S542B: Determine the hidden layer output based on the output gate output corresponding to the current cell state and the weighted feature vector.

[0080] The formula for calculating the output of the weighted feature vector v_att is: o_t = σ(W_o×[h_{t-1},v_att]+b_o), where W_o is the output gate weight matrix, b_o is the output gate bias, and o_t is the output gate output; the formula for calculating the hidden layer output is: h_t = o_t×tanh(c_t).

[0081] S542C processes the hidden layer output based on a preset activation function to obtain the wheel-end braking force estimate.

[0082] It should be noted that the fully connected layer in the LSTM timing prediction layer is used to process the hidden layer output h_t, thereby obtaining the estimated braking force at the outgoing wheel end.

[0083] The activation function in the fully connected layer can output h_t to the hidden layer, thereby obtaining the wheel-end braking force estimate F_pred. The calculation formula of F_pred is: F_pred = W_fc×h_t + b_fc, where W_fc is the weight matrix of the fully connected layer and b_fc is the bias of the fully connected layer.

[0084] It should be noted that by using the LSTM time-series prediction layer to calculate the weighted feature vector, the wheel-end braking force estimate can be determined. In the process of determining the wheel-end braking force estimate, there is no need to use a braking force sensor, which can effectively reduce the cost of determining the wheel-end braking force.

[0085] Example 6 This application provides a method for determining the braking force at the wheel end of a vehicle, which supplements the methods shown in any of the embodiments one to five. It should be noted that for parts not detailed in this embodiment, please refer to the descriptions in other embodiments. Taking the supplementation of the embodiments as an example, the method includes: S610. Periodically collect wheel-end motor parameter data, wherein the wheel-end motor parameter data includes at least the motor armature current, rotor speed, and output shaft angle.

[0086] S620. Preprocess the wheel-end motor parameter data to obtain a time-series dataset.

[0087] S630. Perform feature extraction on the time series dataset to obtain a one-dimensional feature vector.

[0088] S640. Perform time-series prediction on the one-dimensional feature vector to obtain the wheel-end braking force estimate.

[0089] S650. Determine the total loss coefficient based on friction loss and gap loss.

[0090] It should be noted that the ball screw involved in the wheel end will cause corresponding efficiency loss during the transmission process. For example, the efficiency loss includes friction loss and clearance loss. The efficiency loss will cause the wheel end braking force estimated in the above steps to be too large. Therefore, the wheel end braking force estimate needs to be corrected.

[0091] Among them, friction loss μ1 is the efficiency loss caused by friction between the balls and the screw, and between the balls and the nut during the ball screw drive process. The formula for calculating friction loss μ1 is: μ1 = a1×T 2 + b1×T + c1×ln(t+1) + d1, where a1, b1, c1, and d1 are fitting coefficients, T is the real-time working temperature of the wheel end, and t is the working time of the ball screw. For example, determined through experimental fitting: a1=-2.5×10^-6, b1=1.2×10^-4, c1=8×10^-3, d1=0.012 (goodness of fit R0). 2 ≥0.98, fitting accuracy meets requirements). Clearance loss μ2 is the small gap between the ball screw and the motor output shaft, and between the ball screw and the piston, resulting in efficiency loss during transmission. The formula for calculating clearance loss μ2 is: μ2 = a2×t^(0.3) + b2×T + c2, where a2, b2, and c2 are fitting coefficients, determined experimentally: a2=3×10^-5, b2=-5×10^-5, c2=0.008 (goodness of fit R0). 2 (≥0.97, fitting accuracy meets requirements).

[0092] The total loss coefficient μ is the sum of friction loss μ1 and gap loss μ2, i.e., μ = μ1 + μ2.

[0093] S660. Based on the total loss coefficient and the preset ideal transmission coefficient, determine the actual transmission coefficient.

[0094] It should be noted that a more accurate wheel-end braking force should be the product of the estimated wheel-end braking force and the transmission coefficient. In the case of no efficiency loss (ideal situation), the transmission coefficient is the ideal transmission coefficient. For example, the ideal transmission coefficient η0 is set to 0.99 based on historical experience. However, in reality, ball screws have corresponding efficiency losses. Therefore, the actual transmission coefficient η should be the difference between the ideal transmission coefficient η0 and the total loss coefficient μ, that is, η = η0 - μ.

[0095] S670. Based on the estimated wheel-end braking force and the actual transmission coefficient, determine the target wheel-end braking force.

[0096] Wherein, the target braking force at the wheel end F_final is the product of the estimated braking force at the wheel end F_pred and the actual transmission coefficient η, that is, F_final = F_pred × F_pred.

[0097] It should be noted that the actual transmission coefficient η eliminates the adverse effects of efficiency loss on the accuracy of the target braking force at the wheel end. By calculating the target braking force at the wheel end using the actual transmission coefficient η and the estimated braking force at the wheel end, the accuracy of the target braking force at the wheel end can be effectively improved.

[0098] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0099] The beneficial effects of the above embodiments include: 1. Overcoming the high cost of existing technologies and achieving significant economic benefits: Existing technologies rely on external force sensors (costing 400 yuan per unit), and the separate structure requires additional components, resulting in high system costs. This invention replaces external force sensors with a "CNN-LSTM composite algorithm + compensation model," eliminating sensor procurement costs. Simultaneously, it adopts a three-in-one integrated structure, reducing the number of components and assembly processes. Combined with lightweight algorithms adapted to existing MCUs, it eliminates the need for additional hardware computing power, ultimately reducing the cost of a single wheel-end actuator and the overall EMB system cost, completely breaking through the cost bottleneck of EMB mass production.

[0100] 2. Overcoming the shortcomings of low reliability and frequent failures in existing technologies, achieving the beneficial effect of high reliability: In existing technologies, external force sensors are exposed to wheel-end salt spray, mud, and extreme temperature changes for a long time, making them susceptible to electromagnetic interference and corrosion, leading to signal drift and failure. Moreover, the separate structure has many failure points, with a failure rate of about 5‰. This invention eliminates the external force sensor, reducing the number of failure points. Combined with an adaptive error calibration mechanism, the model parameters can be dynamically corrected according to temperature and working time, avoiding accuracy drift, reducing the failure rate to below 0.5‰, and extending service life to match the service life requirements of vehicle components, thereby reducing after-sales maintenance costs.

[0101] 3. This invention addresses the shortcomings of existing technologies, such as unstable braking force control accuracy and poor adaptability to various operating conditions, achieving high precision and wide adaptability: Existing technologies use external sensors for direct detection, which are susceptible to wear and interference, leading to decreased accuracy. Furthermore, they lack a dedicated compensation mechanism, making them unsuitable for complex operating conditions such as extreme temperature changes and continuous braking, resulting in significant accuracy fluctuations. This invention utilizes a CNN-LSTM composite neural network (embedded with an attention mechanism) to accurately capture data features of key braking conditions. Combined with a transmission efficiency compensation model that correlates temperature, operating time, and braking force, it achieves braking force control accuracy within ±3%, with long-term stable accuracy. Simultaneously, the optimized ball screw transmission mechanism (transmission clearance ≤0.01mm) and wide temperature range design (-40℃-120℃) adaptable to various regional operating conditions, with a braking response time ≤100ms, improving driving safety.

[0102] Example 7 Based on the same inventive concept, this embodiment also provides a vehicle wheel-end braking force determining device for implementing the above-described vehicle wheel-end braking force determining method. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more vehicle wheel-end braking force determining device embodiments provided below can be found in the limitations of the vehicle wheel-end braking force determining method described above, and will not be repeated here.

[0103] In this embodiment, as Figure 2As shown, a vehicle wheel-end braking force determination device is provided, comprising: The data acquisition module is used to periodically acquire wheel-end motor parameter data, which includes at least the motor armature current, rotor speed, and output shaft angle. The data preprocessing module is used to preprocess the wheel-end motor parameter data to obtain a time-series dataset; The feature extraction module is used to extract features from the time-series dataset to obtain a one-dimensional feature vector; The data prediction module is used to perform time-series prediction on the one-dimensional feature vector to obtain the wheel-end braking force estimate.

[0104] Each module in the aforementioned vehicle wheel-end braking force determination device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in the electronic device, or stored in the memory of the electronic device as software, so that the processor can call and execute the corresponding operations of each module.

[0105] It should be noted that in this embodiment, wheel-end motor parameter data is collected periodically. The wheel-end motor parameter data includes at least the motor armature current, rotor speed, and output shaft angle. The wheel-end motor parameter data is preprocessed to obtain a time-series dataset. Features are extracted from the time-series dataset to obtain a one-dimensional feature vector. The one-dimensional feature vector is then used for time-series prediction to obtain the wheel-end braking force estimate. Through the above implementation, firstly, wheel-end braking force can still be calculated without using braking force sensors at the wheel ends, thus eliminating the need for braking force sensors and effectively reducing the cost of wheel-end braking force detection; secondly, the process of collecting wheel-end motor parameter data is largely unaffected by the harsh environment at the wheel ends, ensuring the stability of the wheel-end braking force calculation process based on this data and effectively improving the reliability of wheel-end braking force detection; thirdly, the wheel-end motor parameter data is preprocessed before being used to calculate the wheel-end braking force, ensuring the accuracy of the braking force detection. In summary, this application can reduce the cost of wheel-end braking force detection and improve the reliability of wheel-end braking force detection while ensuring the accuracy of the detection.

[0106] In an optional embodiment, the preprocessing of the wheel-end motor parameter data to obtain a time-series dataset includes: The outlier data of the wheel-end motor parameters is processed to obtain de-outlier data; The abnormal data is subjected to noise filtering to obtain denoised data; The denoised data is normalized to obtain normalized data; The normalized data is spliced ​​over time to obtain a time-series dataset.

[0107] In an optional embodiment, the step of extracting features from the time-series dataset to obtain a one-dimensional feature vector includes: Convolution is performed on the time-series dataset to obtain local features of the data; Pooling is performed on the local features of the data to obtain the key features of the data; Feature flattening is performed on the key features of the data to obtain a one-dimensional feature vector.

[0108] In an optional embodiment, the step of performing time-series prediction on the one-dimensional feature vector to obtain the wheel-end braking force estimate includes: Attention weighting is applied to the one-dimensional feature vector to obtain a weighted feature vector; The weighted feature vector is used for time-series prediction to obtain the wheel-end braking force estimate.

[0109] In an optional embodiment, the step of performing time-series prediction on the weighted feature vector to obtain the wheel-end braking force estimate includes: Based on the forget gate output, input gate output, and cell state of the previous time step corresponding to the weighted feature vector, the cell state is updated to obtain the current cell state. The hidden layer output is determined based on the output gate output corresponding to the current cell state and the weighted feature vector. The hidden layer output is processed based on a preset activation function to obtain the wheel-end braking force estimate.

[0110] In an optional embodiment, the vehicle wheel-end braking force determining device further includes: The loss coefficient calculation module is used to determine the total loss coefficient based on friction loss and gap loss; The traditional coefficient calculation module is used to determine the actual transmission coefficient based on the total loss coefficient and the preset ideal transmission coefficient; The target braking force determination module is used to determine the target braking force at the wheel end based on the estimated wheel-end braking force and the actual transmission coefficient.

[0111] Example 8 Figure 3A schematic diagram of an electronic device 10, which can be used to implement embodiments of the present invention, is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0112] like Figure 3 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0113] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0114] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as the method for determining the microcirculation resistance index.

[0115] In some embodiments, the method for determining the microcirculation resistance index may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the method for determining the microcirculation resistance index described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the method for determining the microcirculation resistance index by any other suitable means (e.g., by means of firmware).

[0116] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0117] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0118] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0119] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0120] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0121] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0122] Example 9 In this embodiment, a computer-readable storage medium is provided, such as... Figure 4 As shown, a computer program is stored thereon, and when the computer program is executed by the processor, it implements the steps in the above-described method embodiments.

[0123] Example 10 In this embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0124] It should be noted that the information collected is information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data all comply with the relevant laws, regulations and standards of the relevant countries and regions, necessary confidentiality measures have been taken, and it does not violate public order and good morals. Corresponding operation portals are provided for users to choose to authorize or refuse.

[0125] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this disclosure can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this disclosure may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this disclosure may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0126] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0127] The embodiments described above are merely illustrative of several implementations of this disclosure, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent disclosure. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this disclosure, and these all fall within the protection scope of this disclosure. Therefore, the protection scope of this disclosure should be determined by the appended claims.

Claims

1. A vehicle wheel end braking force determination method characterized by comprising: include: Periodically collect wheel-end motor parameter data, which includes at least the motor armature current, rotor speed, and output shaft angle; The wheel-end motor parameter data is preprocessed to obtain a time-series dataset; Feature extraction is performed on the time-series dataset to obtain a one-dimensional feature vector; The one-dimensional feature vector is used for time-series prediction to obtain the wheel-end braking force estimate.

2. The method of claim 1, wherein, The preprocessing of the wheel-end motor parameter data to obtain a time-series dataset includes: The outlier data of the wheel-end motor parameters is processed to obtain de-outlier data; The abnormal data is subjected to noise filtering to obtain denoised data; The denoised data is normalized to obtain normalized data; The normalized data is spliced ​​over time to obtain a time-series dataset.

3. The method of claim 1, wherein, The step of extracting features from the time-series dataset to obtain a one-dimensional feature vector includes: Convolution is performed on the time-series dataset to obtain local features of the data; Pooling is performed on the local features of the data to obtain the key features of the data; Feature flattening is performed on the key features of the data to obtain a one-dimensional feature vector.

4. The method according to claim 1, characterized in that, The step of performing time-series prediction on the one-dimensional feature vector to obtain the wheel-end braking force estimate includes: Attention weighting is applied to the one-dimensional feature vector to obtain a weighted feature vector; The weighted feature vector is used for time-series prediction to obtain the wheel-end braking force estimate.

5. The method according to claim 4, characterized in that, The step of performing time-series prediction on the weighted feature vector to obtain the wheel-end braking force estimate includes: Based on the forget gate output, input gate output, and cell state of the previous time step corresponding to the weighted feature vector, the cell state is updated to obtain the current cell state. The hidden layer output is determined based on the output gate output corresponding to the current cell state and the weighted feature vector. The hidden layer output is processed based on a preset activation function to obtain the wheel-end braking force estimate.

6. The method according to any one of claims 1-5, characterized in that, Also includes: The total loss coefficient is determined based on friction loss and gap loss; The actual transmission coefficient is determined based on the total loss coefficient and the preset ideal transmission coefficient. The target braking force at the wheel end is determined based on the estimated wheel-end braking force and the actual transmission coefficient.

7. A device for determining the braking force at the wheel end of a vehicle, characterized in that, The device includes: The data acquisition module is used to periodically acquire wheel-end motor parameter data, which includes at least the motor armature current, rotor speed, and output shaft angle. The data preprocessing module is used to preprocess the wheel-end motor parameter data to obtain a time-series dataset; The feature extraction module is used to extract features from the time-series dataset to obtain a one-dimensional feature vector; The data prediction module is used to perform time-series prediction on the one-dimensional feature vector to obtain the wheel-end braking force estimate.

8. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the target box grasping method according to any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the target box grasping method according to any one of claims 1-6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images