A whole vehicle dts measurement data prediction method and device

Abstract

The application discloses a whole vehicle DTS measurement data prediction method and device. The method comprises the following steps: obtaining DTS measured data corresponding to the Nth batch of vehicles obtained by measuring the whole vehicle DTS of the Nth batch of vehicles, and adding the DTS measured data corresponding to the Nth batch to a historical DTS measured data set; obtaining DTS measured data corresponding to the N-Tth to N-1th batches from the historical DTS measured data set, and generating a DTS measured data sequence sorted by batches based on the DTS measured data corresponding to the N-Tth to Nth batches; inputting the DTS measured data sequence into a trained sequence model, so that the sequence model predicts DTS prediction data corresponding to the N+1th batch based on the DTS measured data sequence; wherein the sequence model is a machine learning model obtained by supervised training based on the historical DTS measured data set; the N is a positive integer, the T is a preset first threshold value, and the value of the N is greater than the value of the T.

Description

Technical Field

[0001] This specification relates to the field of communication technology, and in particular to a method and apparatus for predicting DTS measurement data of a whole vehicle. Background Technology

[0002] DTS (Dimension Technical Specifications) refers to the design requirements for the gaps and surface differences between various components of the vehicle's interior and exterior. It is an important indicator for measuring the manufacturing quality of the vehicle's interior and exterior, reflecting the overall dimensional quality of the vehicle.

[0003] DTS measurement, specifically, involves setting up a certain number of measuring points for the gaps and planes of various car components after the car design and production are completed, collecting measurement data for each measuring point, and measuring the manufacturing quality of the car's interior and exterior based on whether the measurement data for each measuring point is within the corresponding standard tolerance range.

[0004] With the development of society, competition in the automotive industry is becoming increasingly fierce. The manufacturing quality of the interior and exterior of automobiles is an important part of measuring the overall quality of automobiles. In current technology, historical data obtained by DTS (Device Testing System) measurement of completed automobiles is generally analyzed to identify problems in the current automobile production process and seek solutions to these problems.

[0005] In other words, current technology can only analyze historical data obtained from DTS measurements of already manufactured cars to identify problems and propose corresponding solutions, but it cannot predict the DTS measurement data of cars that have not yet been manufactured. Summary of the Invention

[0006] This application provides a method and apparatus for predicting DTS measurement data of a whole vehicle, the technical solution of which is as follows:

[0007] According to a first aspect of this application, a method for predicting DTS measurement data of a whole vehicle is provided, the method comprising:

[0008] Obtain the DTS measured data corresponding to the Nth batch of vehicles by performing whole-vehicle DTS measurement, and add the DTS measured data corresponding to the Nth batch to the historical DTS measured data set.

[0009] Obtain DTS measured data corresponding to batches NT to N-1 from the historical DTS measured data set, and generate a DTS measured data sequence sorted by batch based on the DTS measured data corresponding to batches NT to N.

[0010] The DTS measured data sequence is input into the trained sequence model so that the sequence model can predict the DTS prediction data corresponding to the (N+1)th batch based on the DTS measured data sequence; wherein, the sequence model is a machine learning model obtained by supervised training based on the historical DTS measured data set; N is a positive integer, T is a preset first threshold, and the value of N is greater than the value of T.

[0011] Optionally, the method further includes:

[0012] Obtain the DTS measured data corresponding to the N+1th batch of vehicles by performing whole-vehicle DTS measurement, and add the DTS measured data corresponding to the N+1th batch to the historical DTS measured data set.

[0013] Based on the DTS prediction data corresponding to the N+1th batch, the DTS measured data corresponding to the N+1th batch, and the average value of the DTS measured data in the updated historical DTS measured data set, the coefficient of determination is calculated, and it is determined whether the coefficient of determination reaches a preset second threshold.

[0014] If the coefficient of determination does not reach the second threshold, the sequence model is retrained in a supervised manner based on the updated historical DTS measured data set.

[0015] Optionally, the method further includes:

[0016] If the coefficient of determination reaches the second threshold, the sequence model is incrementally trained periodically based on the DTS measured data obtained within the preset time period.

[0017] Optionally, the sequence model can be trained in a supervised manner using the following methods:

[0018] Training samples are determined from the historical DTS measured data set; wherein, the training samples include a sequence of historical DTS measured data sorted by batch, generated based on the DTS measured data corresponding to batches MT to M respectively, and a label for the historical DTS measured data sequence; the label is the DTS measured data corresponding to batch M+1; M is a positive integer; the value of M is greater than the value of T;

[0019] The training samples are input into the sequence model to perform supervised training on the sequence model until the coefficient of determination corresponding to the sequence model reaches a preset second threshold.

[0020] Optionally, the DTS measured data includes DTS measured sub-data corresponding to several DTS measurement points respectively;

[0021] Before generating a DTS measured data sequence ordered by batch based on the DTS measured data corresponding to batches NT to N respectively, the method further includes:

[0022] Determine whether the amount of missing data in the DTS measured data corresponding to any target DTS measurement point in the DTS measured data corresponding to batches NT to N reaches a preset third threshold.

[0023] If so, delete the DTS measured sub-data corresponding to the target DTS measurement point from the DTS measured data corresponding to batches NT to N respectively;

[0024] If not, based on the average value of the DTS measured sub-data corresponding to the target DTS measurement point, the DTS measured data corresponding to batches NT to N are supplemented.

[0025] Optionally, the sequence model includes an LSTM model or a CNN-LSTM model.

[0026] Optionally, the method further includes:

[0027] Determine whether all the prediction data included in the DTS prediction data corresponding to the N+1th batch are within the corresponding preset range;

[0028] If any of the predicted data in the DTS prediction data corresponding to the N+1th batch is not within the corresponding preset range, an early warning prompt will be output.

[0029] According to a second aspect of this application, a whole vehicle DTS measurement data prediction device is provided, the device comprising:

[0030] The acquisition unit is used to acquire the DTS measured data corresponding to the Nth batch of vehicles obtained by performing whole-vehicle DTS measurement on the Nth batch of vehicles, and add the DTS measured data corresponding to the Nth batch to the historical DTS measured data set.

[0031] The generation unit is used to obtain DTS measured data corresponding to batches NT to N-1 from the historical DTS measured data set, and generate a DTS measured data sequence sorted by batch based on the DTS measured data corresponding to batches NT to N.

[0032] The prediction unit is used to input the DTS measured data sequence into the trained sequence model, so that the sequence model can predict the DTS prediction data corresponding to the (N+1)th batch based on the DTS measured data sequence; wherein, the sequence model is a machine learning model obtained by supervised training based on the historical DTS measured data set; N is a positive integer, T is a preset first threshold, and the value of N is greater than the value of T.

[0033] According to a third aspect of this application, an electronic device is provided, including a communication interface, a processor, a memory, and a bus, wherein the communication interface, the processor, and the memory are interconnected via the bus;

[0034] The memory stores machine-readable instructions, and the processor executes the method by invoking the machine-readable instructions.

[0035] According to a fourth aspect of this application, a machine-readable storage medium is provided, the machine-readable storage medium storing machine-readable instructions that, when invoked and executed by a processor, implement the method.

[0036] Through the above embodiments, based on the DTS measurement data corresponding to the Nth batch of vehicles obtained from whole-vehicle DTS measurements, and the DTS measurement data corresponding to batches NT to N-1 obtained from the historical DTS measurement data set, a DTS measurement data sequence ordered by batch can be generated. This DTS measurement data sequence is then input into a trained sequence model, enabling the sequence model to predict the DTS prediction data corresponding to the N+1th batch. By using the DTS measurement data corresponding to the Nth batch and the DTS measurement data corresponding to batches NT to N-1 from the historical DTS measurement data set, the DTS prediction data corresponding to the next batch of vehicles that has not yet been completed can be predicted. Furthermore, the production parameters of the factory's vehicle production equipment can be adjusted based on the DTS prediction data, thereby improving the manufacturing quality of the interior and exterior of the next batch of vehicles that has not yet been completed.

[0037] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings.

[0039] Figure 1 This is a flowchart of a whole vehicle DTS measurement data prediction method in one embodiment of this specification;

[0040] Figure 2 This is a schematic diagram of the architecture of a whole vehicle DTS measurement data prediction system in one embodiment of this specification;

[0041] Figure 3 This is a schematic diagram of the electronic device containing the whole vehicle DTS measurement data prediction device in one embodiment of this specification;

[0042] Figure 4 This is a block diagram of a vehicle DTS measurement data prediction device in one embodiment of this specification. Detailed Implementation

[0043] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with one or more embodiments of this specification. Rather, they are merely examples of apparatuses and methods consistent with some aspects of one or more embodiments of this specification as detailed in the appended claims.

[0044] It should be noted that the steps of the corresponding methods are not necessarily performed in the order shown and described in this specification in other embodiments. In some other embodiments, the methods may include more or fewer steps than described in this specification. Furthermore, a single step described in this specification may be broken down into multiple steps in other embodiments; and multiple steps described in this specification may be combined into a single step in other embodiments.

[0045] Current technologies can only analyze historical data obtained from DTS (Device Testing and Measurement) measurements of completed vehicles to identify problems and propose corresponding solutions. However, they cannot prevent out-of-tolerance issues in vehicles currently in production, thus failing to improve the manufacturing quality of the interior and exterior of the current production batch. Therefore, there is an urgent need for a technical solution that can prevent out-of-tolerance issues in vehicles yet to be completed, thereby improving the manufacturing quality of the interior and exterior of vehicles currently in production.

[0046] In view of this, this specification aims to propose a technical solution that can predict the DTS measurement data of a car before it has been manufactured.

[0047] In implementation, the DTS measured data corresponding to the Nth batch of vehicles can be obtained by performing whole-vehicle DTS measurement on the Nth batch of vehicles, and the DTS measured data corresponding to the Nth batch can be added to the historical DTS measured data set.

[0048] Furthermore, DTS measured data corresponding to batches NT to N-1 can be obtained from the historical DTS measured data set, and a DTS measured data sequence sorted by batch can be generated based on the DTS measured data corresponding to batches NT to N.

[0049] Furthermore, the DTS measured data sequence can be input into the trained sequence model, so that the sequence model can predict the DTS prediction data corresponding to the (N+1)th batch based on the DTS measured data sequence; wherein, the sequence model can be a machine learning model obtained by supervised training based on the historical DTS measured data set; N is a positive integer, T is a preset first threshold, and the value of N is greater than the value of T.

[0050] Through the above embodiments, a DTS measurement data sequence corresponding to the Nth batch of vehicles can be generated based on the DTS measurement data obtained from the whole vehicle DTS measurement of the Nth batch of vehicles, and the DTS measurement data corresponding to the Nth to N-1th batches respectively obtained from the historical DTS measurement data set. This sequence is then input into a trained sequence model, enabling the model to predict the DTS prediction data corresponding to the N+1th batch. By using the DTS measurement data corresponding to the Nth batch and the DTS measurement data corresponding to the Nth to N-1th batches respectively from the historical DTS measurement data set, the DTS prediction data corresponding to the next batch of vehicles that has not yet been completed can be predicted. Furthermore, the production parameters of the factory's vehicle production equipment can be adjusted based on the DTS prediction data, thereby improving the manufacturing quality of the interior and exterior of the next batch of vehicles that has not yet been completed.

[0051] Please see Figure 1 , Figure 1 This is a flowchart illustrating an exemplary method for predicting DTS measurement data for a whole vehicle.

[0052] like Figure 1 As shown, the method may include the following steps:

[0053] S102: Obtain the DTS measured data corresponding to the Nth batch of vehicles obtained by performing whole-vehicle DTS measurement, and add the DTS measured data corresponding to the Nth batch to the historical DTS measured data set.

[0054] N is a positive integer, and the DTS measured data refers to the actual measurement data obtained by DTS measurement of the vehicle. The historical DTS measured data set may include the DTS measured data corresponding to the first to N-1 batches of vehicles obtained by whole-vehicle DTS measurement.

[0055] It can obtain the DTS measured data corresponding to the Nth batch of vehicles by performing whole-vehicle DTS measurement, and add the DTS measured data corresponding to the Nth batch to the historical DTS measured data set.

[0056] For example, assuming N is 100, the historical DTS measured data set may include the DTS measured data corresponding to the first to 99th batches of vehicles obtained by performing whole-vehicle DTS measurements. Whole-vehicle DTS measurements may be performed on the 100th batch of vehicles to obtain the measured data corresponding to the 100th batch of vehicles. The DTS measured data corresponding to the 100th batch may also be added to the historical DTS measured data set.

[0057] Regarding the whole-vehicle DTS measurement for the Nth batch of vehicles, it can be performed on one vehicle in the Nth batch, or on ten vehicles in the Nth batch, or on all vehicles in the Nth batch. This manual does not limit this to any particular method.

[0058] S104: Obtain the DTS measured data corresponding to batches NT to N-1 from the historical DTS measured data set, and generate a DTS measured data sequence sorted by batch based on the DTS measured data corresponding to batches NT to N.

[0059] The DTS measured data corresponding to batches NT to N-1 can be obtained from the historical DTS measured data set, and a DTS measured data sequence sorted by batch can be generated based on the DTS measured data corresponding to batches NT to N.

[0060] For example, assuming N is 100 and T is 10, DTS measured data corresponding to batches 90 to 99 can be obtained from the historical DTS measured data set, and a DTS measured data sequence sorted by batch can be generated based on the DTS measured data corresponding to batches 90 to 100.

[0061] In one embodiment, the DTS measured data may include DTS measured sub-data corresponding to several DTS measurement points. Before generating a DTS measured data sequence ordered by batch based on the DTS measured data corresponding to batches NT to N, it may be determined whether the amount of missing data in the DTS measured sub-data corresponding to any target DTS measurement point in the DTS measured data corresponding to batches NT to N reaches a preset third threshold. If the amount of missing data in the DTS measured sub-data corresponding to any target DTS measurement point in the DTS measured data corresponding to batches NT to N reaches the preset third threshold, the DTS measured sub-data corresponding to the target DTS measurement point can be deleted from the DTS measured data corresponding to batches NT to N. If the amount of missing data in the DTS measured sub-data corresponding to any target DTS measurement point in the DTS measured data corresponding to batches NT to N does not reach the preset third threshold, the DTS measured data corresponding to batches NT to N can be supplemented based on the average value of the DTS measured sub-data corresponding to the target DTS measurement point.

[0062] It should be noted that the DTS measured sub-data corresponding to any target DTS measurement point can be the DTS measured sub-data corresponding to any DTS measurement point included in the DTS measured data corresponding to batches NT to N respectively, and does not specifically refer to the DTS measured sub-data corresponding to a certain DTS measurement point.

[0063] For example, assuming N is 100, T is 10, and the preset third threshold is 40% of the data volume of the DTS measured sub-data corresponding to the DTS measurement point containing complete data, then before generating the DTS measured data sequence sorted by batch based on the DTS measured data corresponding to batches 90 to 100, it can be determined whether the data missing amount of the DTS measured sub-data corresponding to any target DTS measurement point in the DTS measured data corresponding to batches 90 to 100 reaches 40% of the data volume of the DTS measured sub-data corresponding to the DTS measurement point containing complete data. If the data missing amount of the DTS measured sub-data corresponding to any target DTS measurement point reaches 40% of the data volume of the DTS measured sub-data corresponding to the DTS measurement point containing complete data, then... If the data volume of the DTS measured sub-data corresponding to the DTS measurement point containing complete data is 40%, then the DTS measured sub-data corresponding to the target DTS measurement point can be deleted from the DTS measured data corresponding to batches 90 to 100 respectively. If the data missing amount of the DTS measured sub-data corresponding to any target DTS measurement point does not reach 40% of the data volume of the DTS measured sub-data corresponding to the DTS measurement point containing complete data, then the average value of the DTS measured sub-data corresponding to the target DTS measurement point can be calculated, and the DTS measured sub-data corresponding to the target DTS measurement point contained in the DTS measured data corresponding to batches 90 to 100 can be supplemented based on the average value.

[0064] In one embodiment shown, the DTS measured data corresponding to batches NT to N can be normalized, and a DTS measured data sequence ordered by batch can be generated based on the normalized DTS measured data.

[0065] For example, assuming N is 100 and T is 10, the DTS measured data corresponding to batches 90 to 100 can be normalized, and a DTS measured data sequence ordered by batch can be generated based on the normalized DTS measured data.

[0066] Normalization refers to processing data to limit it to a certain range. For example, data can be limited to the range [0,1].

[0067] Regarding the specific normalization method, it can be max-min standardization, but this specification does not limit it.

[0068] For example, the max-min standardization method can be used, and the specific formula can be:

[0069]

[0070] Wherein, x' can be the normalized data of the i-th measured data at any measuring point, x can be the i-th measured data at any measuring point, max(x) can be the maximum measured data contained in the measured data at any measuring point, and min(x) can be the minimum measured data contained in the measured data at any measuring point.

[0071] S106: Input the DTS measured data sequence into the trained sequence model, so that the sequence model predicts the DTS prediction data corresponding to the N+1th batch based on the DTS measured data sequence; wherein, the sequence model is a machine learning model obtained by supervised training based on the historical DTS measured data set; N is a positive integer, T is a preset first threshold, and the value of N is greater than the value of T.

[0072] The sequence model can be a machine learning model obtained by supervised training based on the historical DTS measured data set, and the DTS prediction data can be the predicted data corresponding to the vehicles in the (N+1)th batch, where N is a positive integer, T is a preset first threshold, and the value of N is greater than the value of T.

[0073] The actual DTS data sequence can be input into the trained sequence model so that the sequence model can predict the DTS prediction data corresponding to the N+1th batch based on the actual DTS data sequence.

[0074] For example, assuming N is 100 and T is preset to 10, the DTS measured data sequence can be input into the already trained sequence model so that the sequence model can predict the DTS prediction data corresponding to the 101st batch based on the DTS measured data sequence.

[0075] The sequence model can be a machine learning model obtained by supervised training based on the historical DTS measured data set. Supervised training, also known as supervised learning, refers to the machine learning task of inferring functions from a labeled training dataset. The training data consists of a set of training instances, each of which includes an input object and a corresponding output value.

[0076] For example, supervised training can be performed by using the first T batches of DTS measured data contained in the historical DTS measured data as input objects and obtaining the DTS measured data corresponding to the (T+1)th batch as the corresponding output value.

[0077] In one embodiment, the supervised training based on the historical DTS measured data set can specifically involve determining training samples from the historical DTS measured data set. The training samples can include a sequence of historical DTS measured data, sorted by batch, generated based on the DTS measured data corresponding to batches MT to M, and labels affixed to the historical DTS measured data sequence. The labels can be DTS measured data corresponding to batch M+1. M is a positive integer, and the value of M is greater than the value of T. The training samples can be input into the sequence model to perform supervised training on the sequence model until the coefficient of determination corresponding to the sequence model reaches a preset second threshold.

[0078] For example, assuming T is 3, the historical DTS measured data includes DTS measured data corresponding to batches 1 to 8, and M can be 4, then the training sample can include a sequence of historical DTS measured data generated according to batches based on the DTS measured data corresponding to batches 1 to 4, and a label for the historical DTS measured data sequence, where the label can be the DTS measured data corresponding to batch 5; if M can be 5, then the training sample can include a sequence of historical DTS measured data generated according to batches based on the DTS measured data corresponding to batches 2 to 5, and a label for the historical DTS measured data sequence. The labels can be DTS measured data corresponding to the 6th batch; if M can be 6, then the training samples can include historical DTS measured data sequences sorted by batch, generated based on DTS measured data corresponding to the 3rd to 6th batches respectively, and the labels annotating the historical DTS measured data sequences can be DTS measured data corresponding to the 7th batch; if M can be 7, then the training samples can include historical DTS measured data sequences sorted by batch, generated based on DTS measured data corresponding to the 4th to 7th batches respectively, and the labels annotating the historical DTS measured data sequences can be DTS measured data corresponding to the 8th batch.

[0079] The coefficient of determination, also known as the determination coefficient, is a numerical characteristic used to represent the relationship between a random variable and multiple random variables. It reflects the proportion of the total variation of the dependent variable that can be explained by the independent variable through the regression relationship, and is generally represented by the symbol "R2".

[0080] For example, the preset second threshold may include 0.8, and the training samples may be input into the sequence model to perform supervised training on the sequence model until the coefficient of determination of the sequence model reaches 0.8.

[0081] By setting a coefficient of determination, the accuracy of the sequence model's predictions can be determined, making it easier to select a suitable training method for updating the sequence model based on the accuracy.

[0082] In one embodiment shown, the sequence model may include an LSTM model or a CNN-LSTM model.

[0083] The LSTM (Long Short Term Memory) model is a type of recurrent neural network designed specifically to address the long-term dependency problem inherent in general RNNs (Recurrent Neural Networks).

[0084] CNN (Convolutional Neural Networks) is a type of feedforward neural network that includes convolutional computation and has a deep structure. It is one of the representative algorithms of deep learning. It is constructed by imitating the visual perception mechanism of biological organisms. It can perform supervised and unsupervised learning. The sharing of convolutional kernel parameters in its hidden layers and the sparsity of inter-layer connections enable convolutional neural networks to learn gridded features with a small amount of computation.

[0085] For example, the sequence model may include a CNN-LSTM model, which can extract data features and sequence features of the DTS measured data sequence, calculate weight values ​​corresponding to the DTS measured data sequence based on the data features and sequence features, and predict DTS prediction data corresponding to the N+1th batch based on the weight values.

[0086] The model parameters of the CNN-LSTM model may include the number of convolutional kernels of the CNN, the hidden neurons of the LSTM, the output neurons of the LSTM, and the activation function, but this specification does not limit these parameters.

[0087] The specific values ​​of the model parameters of the CNN-LSTM model may include 32 convolutional kernels for the CNN, 180 hidden neurons for the LSTM, and 256 output neurons for the LSTM. This specification does not limit these values.

[0088] In one embodiment shown, DTS measured data corresponding to the N+1th batch of vehicles obtained by performing whole-vehicle DTS measurements can also be acquired and added to the historical DTS measured data set. Based on the DTS predicted data corresponding to the N+1th batch, the DTS measured data corresponding to the N+1th batch, and the average value of the DTS measured data in the updated historical DTS measured data set, a coefficient of determination is calculated, and it is determined whether the coefficient of determination reaches a preset second threshold. If the coefficient of determination does not reach the second threshold, the sequence model is retrained in a supervised manner based on the updated historical DTS measured data set.

[0089] The step of retraining the sequence model in a supervised manner based on the updated historical DTS measured data set may include: adjusting the model parameters of the sequence model, and retraining the sequence model in a supervised manner based on the model with adjusted parameters and new training samples constructed from the updated historical DTS measured data set.

[0090] For example, assuming N is 100 and the preset second threshold is 0.8, the DTS measured data corresponding to the 101st batch of vehicles obtained by performing whole-vehicle DTS measurements can be acquired and added to the historical DTS measured data set. Based on the average value of the DTS predicted data, the DTS measured data, and the updated historical DTS measured data set, a coefficient of determination can be calculated, and it can be determined whether the coefficient of determination reaches 0.8. If the coefficient of determination does not reach 0.8, the model parameters of the sequence model can be adjusted, and the sequence model can be retrained in supervised manner based on the model with adjusted parameters and new training samples constructed from the updated historical DTS measured data set.

[0091] The specific method for calculating the coefficient of determination may include the following calculation method:

[0092] First, the regression sum of squares (SSR) can be calculated. The SSR represents the error between the average value of the DTS predicted data and the average value of the measured DTS data. It reflects the sum of squared deviations in the correlation between the independent and dependent variables. The specific formula is as follows:

[0093]

[0094] Wherein, the y i It can represent DTS measured data, y2 It can represent the average value of the measured data from DTS. It can represent DTS prediction data.

[0095] Furthermore, the residual sum of squares (SSE) can be calculated. This residual sum of squares represents the error between the DTS predicted data and the DTS measured data, reflecting the degree of model fit. The specific formula can be:

[0096]

[0097] Furthermore, the total sum of squared deviations (SST) can be calculated. The SST represents the error between the average value of the DTS measured data and the actual DTS measured data, reflecting the degree of deviation from the mathematical expectation. The specific formula can be:

[0098] SST = SSR + SSE

[0099] That is:

[0100]

[0101] Furthermore, R2 can be defined as:

[0102]

[0103] That is:

[0104]

[0105] It is important to note that the closer R² is to 1, the better the independent variable explains the dependent variable in the regression analysis. In other words, the closer R² is to 1, the closer the measured DTS data is to the predicted DTS data. When R² = 1, it means that the measured DTS data and the predicted DTS data are completely equal, with no error.

[0106] The specific time for retraining the sequence model in a supervised manner can be set between midnight and 2 a.m. each day; this specification does not impose any restrictions on this.

[0107] By retraining the sequence model in a supervised manner, the model parameters can be adjusted to update the sequence model in a timely manner, making the predictions of the sequence model more accurate.

[0108] In one embodiment shown, if the coefficient of determination reaches the second threshold, the sequence model is incrementally trained periodically based on the DTS measured data obtained within the time period, according to a preset time period.

[0109] Incremental training refers to adding new training samples to a model that has already been trained.

[0110] The preset time period can be from 2:00 AM to 2:00 AM the next day, but this specification does not limit it.

[0111] For example, assuming the preset second threshold is 0.8 and the preset time period is from 2:00 AM to 2:00 AM the next day, if the coefficient of determination is determined to reach 0.8, the sequence model can be incrementally trained periodically from 2:00 AM to 2:00 AM the next day based on the DTS measured data obtained from 2:00 AM to 2:00 AM the next day.

[0112] The specific time for incremental training of the sequence model can be set from midnight to 2 a.m. the next day of the preset time period, and this specification does not limit this.

[0113] By periodically performing incremental training on the sequence model, the sequence model can be updated in a timely manner, making the predictions of the sequence model more accurate while saving the time and energy required for model training.

[0114] In one embodiment shown, it can also be determined whether all the predicted data contained in the DTS predicted data corresponding to the N+1th batch are within the corresponding preset range. If any of the predicted data contained in the DTS predicted data corresponding to the N+1th batch is not within the corresponding preset range, an early warning prompt is output.

[0115] For example, assuming N is 100, the vehicle may include measurement point 1 and measurement point 2. It can be determined whether the prediction data corresponding to measurement point 1 in the DTS prediction data corresponding to the 101st batch is within the preset range corresponding to measurement point 1, and whether the prediction data corresponding to measurement point 2 is within the preset range corresponding to measurement point 2. If the prediction data corresponding to measurement point 1 is not within the preset range corresponding to measurement point 1, or the prediction data corresponding to measurement point 2 is not within the preset range corresponding to measurement point 2, a warning prompt can be output.

[0116] The warning prompt may be a warning prompt for the measurement point where the predicted data is not within the corresponding preset range, and this specification does not limit it.

[0117] By issuing early warnings for measurement points where predicted data exceeds a preset range, production parameters corresponding to the measurement points in the factory can be adjusted based on the early warnings, thereby improving the manufacturing quality of the interior and exterior of the next batch of vehicles.

[0118] Please see Figure 2 , Figure 2This is a schematic diagram of the architecture of a vehicle DTS measurement data prediction system in one embodiment of this specification. Figure 2 As shown, assuming the sequence model is a CNN-LSTM model, whole-vehicle DTS measurement can be performed on the 100th batch of vehicles, and the measured data corresponding to the 100th batch of vehicles can be obtained. The DTS measured data corresponding to the 100th batch can also be added to the historical DTS measured data set, and the DTS measured data corresponding to the 90th to 99th batches can be obtained from the historical DTS measured data set.

[0119] Assuming the preset third threshold is 40% of the data volume of the DTS measured sub-data corresponding to the DTS measurement point containing complete data, it can also be determined whether the data missing amount of the DTS measured sub-data corresponding to any one of the target DTS measurement points in the DTS measured data corresponding to batches 90 to 100 reaches 40% of the data volume of the DTS measured sub-data corresponding to the DTS measurement point containing complete data. If the data missing amount of the DTS measured sub-data corresponding to any one target DTS measurement point reaches 40% of the data volume of the DTS measured sub-data corresponding to the DTS measurement point containing complete data, then... Then, DTS measured sub-data corresponding to the target DTS measurement point can be deleted from the DTS measured data corresponding to batches 90 to 100 respectively; if the amount of missing data in the DTS measured sub-data corresponding to any target DTS measurement point does not reach 40% of the amount of data in the DTS measured sub-data corresponding to the DTS measurement point containing complete data, the average value of the DTS measured sub-data corresponding to the target DTS measurement point can be calculated, and the DTS measured sub-data corresponding to the target DTS measurement point contained in the DTS measured data corresponding to batches 90 to 100 can be supplemented based on the average value.

[0120] The DTS measured data corresponding to batches 90 to 100 can be normalized, and a DTS measured data sequence ordered by batch can be generated based on the normalized DTS measured data.

[0121] Furthermore, training samples can be input into the CNN-LSTM model to perform supervised training on the CNN-LSTM model until the coefficient of determination of the CNN-LSTM model reaches 0.8.

[0122] The training samples may include a sequence of historical DTS measured data generated based on the DTS measured data corresponding to batches 1 to 11, sorted by batch, and a label for the historical DTS measured data sequence, wherein the label may be the DTS measured data corresponding to batch 12.

[0123] Furthermore, the actual DTS data sequence can be input into the already trained CNN-LSTM model so that the CNN-LSTM model can predict the DTS prediction data corresponding to the 101st batch based on the actual DTS data sequence.

[0124] Assuming the vehicle includes measurement point 1 and measurement point 2, it can be determined whether the prediction data corresponding to measurement point 1 in the DTS prediction data of the 101st batch is within the preset range corresponding to measurement point 1, and whether the prediction data corresponding to measurement point 2 is within the preset range corresponding to measurement point 2. If the prediction data corresponding to measurement point 1 is not within the preset range corresponding to measurement point 1, or the prediction data corresponding to measurement point 2 is not within the preset range corresponding to measurement point 2, a warning prompt can be output for measurement point 1 or measurement point 2.

[0125] It is also possible to obtain the DTS measured data corresponding to the 101st batch of vehicles obtained by performing whole-vehicle DTS measurement, and add the DTS measured data corresponding to the 101st batch to the historical DTS measured data set. The coefficient of determination can be calculated based on the average value of the DTS predicted data corresponding to the 101st batch, the DTS measured data corresponding to the 101st batch, and the updated historical DTS measured data set. It can be determined whether the coefficient of determination reaches 0.8. If the coefficient of determination does not reach 0.8, the model parameters of the CNN-LSTM model can be adjusted, and the CNN-LSTM model can be retrained in supervised manner based on the adjusted CNN-LSTM model and the updated historical DTS measured data set.

[0126] Assuming the preset time period is from 2:00 AM to 2:00 AM the next day, if the coefficient of determination is determined to be 0.8, the CNN-LSTM model can be incrementally trained periodically from 0:00 AM to 2:00 AM the next day based on the DTS test data obtained from 2:00 AM to 2:00 AM the next day.

[0127] Corresponding to the embodiment of the whole vehicle DTS measurement data prediction method, this specification also provides an embodiment of a whole vehicle DTS measurement data prediction device.

[0128] Please see Figure 3 , Figure 3 This is a schematic diagram of the electronic device housing the whole-vehicle DTS measurement data prediction device in one embodiment of this specification. At the hardware level, the device includes a processor 302, an internal bus 304, a network interface 306, memory 308, and non-volatile memory 310, and may also include other hardware required for other services. One or more embodiments of this specification can be implemented in software, for example, the processor 303 can read the corresponding computer program from the non-volatile memory 310 into memory 308 and then run it. Of course, besides software implementation, one or more embodiments of this specification do not exclude other implementation methods, such as logic devices or a combination of hardware and software, etc. That is to say, the execution entity of the following processing flow is not limited to individual logic units, but can also be hardware or logic devices.

[0129] Please see Figure 4 , Figure 4 This is a block diagram of a vehicle DTS measurement data prediction device according to one embodiment of this specification. The vehicle DTS measurement data prediction device can be applied to, for example... Figure 3 The electronic device shown implements the technical solution described in this specification. The whole vehicle DTS measurement data prediction device includes:

[0130] The acquisition unit 402 is used to acquire the DTS measured data corresponding to the Nth batch of vehicles obtained by performing whole vehicle DTS measurement on the Nth batch of vehicles, and add the DTS measured data corresponding to the Nth batch to the historical DTS measured data set.

[0131] The generation unit 404 is used to obtain DTS measured data corresponding to batches NT to N-1 from the historical DTS measured data set, and generate a DTS measured data sequence sorted by batch based on the DTS measured data corresponding to batches NT to N.

[0132] The prediction unit 406 is used to input the DTS measured data sequence into the trained sequence model, so that the sequence model predicts the DTS prediction data corresponding to the (N+1)th batch based on the DTS measured data sequence; wherein, the sequence model is a machine learning model obtained by supervised training based on the historical DTS measured data set; N is a positive integer, T is a preset first threshold, and the value of N is greater than the value of T.

[0133] In this embodiment, the device may further include a supervised training unit for:

[0134] Obtain the DTS measured data corresponding to the N+1th batch of vehicles by performing whole-vehicle DTS measurement, and add the DTS measured data corresponding to the N+1th batch to the historical DTS measured data set.

[0135] Based on the DTS prediction data corresponding to the N+1th batch, the DTS measured data corresponding to the N+1th batch, and the average value of the DTS measured data in the updated historical DTS measured data set, the coefficient of determination is calculated, and it is determined whether the coefficient of determination reaches a preset second threshold.

[0136] If the coefficient of determination does not reach the second threshold, the sequence model is retrained in a supervised manner based on the updated historical DTS measured data set.

[0137] The device may further include an incremental training unit for:

[0138] If the coefficient of determination reaches the second threshold, the sequence model is incrementally trained periodically based on the DTS measured data obtained within the preset time period.

[0139] The prediction unit 406 can also be used to perform supervised training of the sequence model in the following ways:

[0140] Training samples are determined from the historical DTS measured data set; wherein, the training samples include a sequence of historical DTS measured data sorted by batch, generated based on the DTS measured data corresponding to batches MT to M respectively, and a label for the historical DTS measured data sequence; the label is the DTS measured data corresponding to batch M+1; M is a positive integer; the value of M is greater than the value of T;

[0141] The training samples are input into the sequence model to perform supervised training on the sequence model until the coefficient of determination corresponding to the sequence model reaches a preset second threshold.

[0142] The DTS measured data includes DTS measured sub-data corresponding to several DTS measurement points respectively;

[0143] The apparatus may further include a determining unit, configured to:

[0144] Determine whether the amount of missing data in the DTS measured data corresponding to any target DTS measurement point in the DTS measured data corresponding to batches NT to N reaches a preset third threshold.

[0145] If so, delete the DTS measured sub-data corresponding to the target DTS measurement point from the DTS measured data corresponding to batches NT to N respectively;

[0146] If not, based on the average value of the DTS measured sub-data corresponding to the target DTS measurement point, the DTS measured data corresponding to batches NT to N are supplemented.

[0147] The sequence model in the device may include an LSTM model or a CNN-LSTM model.

[0148] The device may further include an early warning unit, used for:

[0149] Determine whether all the prediction data included in the DTS prediction data corresponding to the N+1th batch are within the corresponding preset range;

[0150] If any of the predicted data in the DTS prediction data corresponding to the N+1th batch is not within the corresponding preset range, an early warning prompt will be output.

[0151] The specific implementation process of the functions and roles of each unit in the above device can be found in the implementation process of the corresponding steps in the above method, and will not be repeated here.

[0152] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of the solution in this specification according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0153] The apparatus, module, or unit described in the embodiments may be implemented by a computer chip or entity, or by a product with a certain function. A typical implementation device is a computer, which may be a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email sending and receiving device, game console, tablet computer, wearable device, or any combination of these devices.

[0154] In a typical configuration, a computer includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0155] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0156] Computer-readable media, including both permanent and non-permanent, removable and non-removable media, can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage, quantum memory, graphene-based storage media or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0157] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0158] The description focuses on specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than those shown in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0159] The terminology used in one or more embodiments of this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of one or more embodiments of this specification. The singular forms “a,” “described,” and “the” as used in one or more embodiments of this specification and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0160] It should be understood that although the terms first, second, third, etc., may be used to describe various information in one or more embodiments of this specification, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, first information may also be referred to as second information without departing from the scope of one or more embodiments of this specification, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "in response to a determination," or "when," or "in the event of a determination."

[0161] The above description is merely a preferred embodiment of one or more embodiments of this specification and is not intended to limit the scope of one or more embodiments of this specification. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of one or more embodiments of this specification should be included within the protection scope of one or more embodiments of this specification.

Claims

1. A method for predicting DTS measurement data of a whole vehicle, the method comprising: Obtain the DTS measured data corresponding to the Nth batch of vehicles by performing whole-vehicle DTS measurement, and add the DTS measured data corresponding to the Nth batch to the historical DTS measured data set. Obtain DTS measured data corresponding to batches NT to N-1 from the historical DTS measured data set, and generate a DTS measured data sequence sorted by batch based on the DTS measured data corresponding to batches NT to N. The DTS measured data sequence is input into the trained sequence model so that the sequence model can predict the DTS prediction data corresponding to the (N+1)th batch based on the DTS measured data sequence; wherein, the sequence model is a machine learning model obtained by supervised training based on the historical DTS measured data set; N is a positive integer, T is a preset first threshold, and the value of N is greater than the value of T. The method further includes: Obtain the DTS measured data corresponding to the N+1th batch of vehicles by performing whole-vehicle DTS measurement, and add the DTS measured data corresponding to the N+1th batch to the historical DTS measured data set. Based on the DTS prediction data corresponding to the N+1th batch, the DTS measured data corresponding to the N+1th batch, and the average value of the DTS measured data in the updated historical DTS measured data set, the coefficient of determination is calculated, and it is determined whether the coefficient of determination reaches a preset second threshold. If the coefficient of determination does not reach the second threshold, the sequence model is retrained in a supervised manner based on the updated historical DTS measured data set.

2. The method according to claim 1, further comprising: If the coefficient of determination reaches the second threshold, the sequence model is incrementally trained periodically based on the DTS measured data obtained within the preset time period.

3. The method according to claim 1, wherein the sequence model is subjected to supervised training in the following manner: Training samples were determined from the historical DTS measured data set; wherein... The training samples include a sequence of historical DTS measured data, sorted by batch, generated based on the measured DTS data corresponding to batches MT to M respectively, and labels labeled for the historical DTS measured data sequences; the labels are the measured DTS data corresponding to batch M+1; M is a positive integer; the value of M is greater than the value of T; The training samples are input into the sequence model to perform supervised training on the sequence model until the coefficient of determination corresponding to the sequence model reaches a preset second threshold.

4. The method according to claim 1, wherein the DTS measured data includes DTS measured sub-data corresponding to a plurality of DTS measurement points respectively; Before generating a DTS measured data sequence ordered by batch based on the DTS measured data corresponding to batches NT to N respectively, the method further includes: Determine whether the amount of missing data in the DTS measured data corresponding to any target DTS measurement point in the DTS measured data corresponding to batches NT to N reaches a preset third threshold. If so, delete the DTS measured sub-data corresponding to the target DTS measurement point from the DTS measured data corresponding to batches NT to N respectively; If not, based on the average value of the DTS measured sub-data corresponding to the target DTS measurement point, the DTS measured data corresponding to batches NT to N are supplemented.

5. The method according to claim 1, wherein the sequence model includes an LSTM model or a CNN-LSTM model.

6. The method according to claim 1, further comprising: Determine whether all the predicted data included in the DTS predicted data corresponding to the N+1th batch are within the corresponding preset range; If any of the predicted data in the DTS prediction data corresponding to the N+1th batch is not within the corresponding preset range, an early warning prompt will be output.

7. A whole vehicle DTS measurement data prediction device, the device comprising: The acquisition unit is used to acquire the DTS measured data corresponding to the Nth batch of vehicles obtained by performing whole-vehicle DTS measurement on the Nth batch of vehicles, and add the DTS measured data corresponding to the Nth batch to the historical DTS measured data set. The generation unit is used to obtain DTS measured data corresponding to batches NT to N-1 from the historical DTS measured data set, and generate a DTS measured data sequence sorted by batch based on the DTS measured data corresponding to batches NT to N. A prediction unit is used to input the DTS measured data sequence into a trained sequence model, so that the sequence model can predict the DTS prediction data corresponding to the (N+1)th batch based on the DTS measured data sequence; wherein, the sequence model is a machine learning model obtained by supervised training based on the historical DTS measured data set; N is a positive integer, T is a preset first threshold, and the value of N is greater than the value of T. The device may further include a supervised training unit for: Obtain the DTS measured data corresponding to the N+1th batch of vehicles by performing whole-vehicle DTS measurement, and add the DTS measured data corresponding to the N+1th batch to the historical DTS measured data set. Based on the DTS prediction data corresponding to the N+1th batch, the DTS measured data corresponding to the N+1th batch, and the average value of the DTS measured data in the updated historical DTS measured data set, the coefficient of determination is calculated, and it is determined whether the coefficient of determination reaches a preset second threshold. If the coefficient of determination does not reach the second threshold, the sequence model is retrained in a supervised manner based on the updated historical DTS measured data set.

8. An electronic device, comprising a communication interface, a processor, a memory, and a bus, wherein the communication interface, the processor, and the memory are interconnected via the bus; The memory stores machine-readable instructions, and the processor executes the method according to any one of claims 1 to 6 by invoking the machine-readable instructions.

9. A machine-readable storage medium storing machine-readable instructions that, when invoked and executed by a processor, implement the method of any one of claims 1 to 6.

Images