Comfort evaluation sample construction method and device, computer device, and storage medium

Abstract

The application relates to a comfort evaluation sample construction method and device, computer equipment, a storage medium and a computer program product. The method comprises the following steps: acquiring a test data set, determining a first correlation between first data mean values corresponding to any two transmission channels in a first data subset, and determining a second correlation between the first data mean value corresponding to any transmission channel in the first data subset and a second data mean value corresponding to a second data subset; filtering the data of each transmission channel contained in the first data subset according to the first correlation, the second correlation and a preset filtering condition, and determining the data of a target transmission channel; and constructing a comfort evaluation sample according to the data of the target transmission channel and the second data subset. The comfort evaluation sample is used for training a comfort evaluation model to obtain a comfort score. The method can obtain a comfort evaluation sample that accurately reflects passenger comfort.

Description

Technical Field

[0001] This application relates to the field of vehicle technology, and in particular to a method, apparatus, computer equipment, storage medium, and computer program product for constructing comfort evaluation samples. Background Technology

[0002] Currently, people have increasingly higher requirements for vehicle comfort. Based on the comfort evaluation sample construction method, the constructed comfort evaluation samples are used to train the comfort evaluation model, which can then be used to evaluate the comfort of vehicles.

[0003] Current methods for constructing comfort evaluation samples are typically based on the ISO-2631 standard (a standard for whole-body vibration evaluation). This involves obtaining a vehicle vibration acceleration dataset within a preset time period, then segmenting the vehicle vibration acceleration dataset using the preset time period as the time division point, and randomly selecting vehicle vibration acceleration data from the segmented vehicle vibration acceleration dataset to construct comfort evaluation samples.

[0004] However, current methods for constructing comfort evaluation samples may result in correlations or redundancies between the samples, and they primarily evaluate comfort based on vehicle vibration acceleration data. This results in limited evaluation information, making it difficult to accurately reflect passenger comfort and thus affecting the efficiency of the comfort evaluation model. Summary of the Invention

[0005] Therefore, it is necessary to provide a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for constructing comfort evaluation samples to address the aforementioned technical problems.

[0006] Firstly, this application provides a method for constructing comfort evaluation samples. The method includes:

[0007] Obtain a test dataset, which includes a first data subset and a second data subset; the first data subset contains data from multiple transmission channels;

[0008] A first correlation is determined between the first data means corresponding to any two transmission channels in the first data subset, and a second correlation is determined between the first data means corresponding to any transmission channel in the first data subset and the second data means corresponding to the second data subset; the first data means are obtained by performing root mean square value processing on the data of the transmission channels; the second data means are obtained by performing mean processing on each second data in the second data subset;

[0009] Based on the first correlation, the second correlation, and the preset filtering conditions, the data of each transmission channel contained in the first data subset are filtered to determine the data of the target transmission channel.

[0010] A comfort evaluation sample is constructed based on the data from the target transmission channel and the second data subset; the comfort evaluation sample is used to train the comfort evaluation model to obtain a comfort score.

[0011] In one embodiment, before determining the first correlation between the first data means corresponding to any two transmission channels in the first data subset, the method further includes:

[0012] The root mean square value is calculated on the data of each transmission channel in the first data subset to obtain the first data mean value corresponding to each transmission channel.

[0013] The mean of the second data in the second subset of data is calculated by performing a mean operation.

[0014] In one embodiment, the data of each transmission channel included in the first data subset is filtered according to the first correlation, the second correlation, and preset filtering conditions to determine the data of the target transmission channel, including:

[0015] In the second correlation between the first data mean corresponding to any transmission channel in the first data subset and the second data mean corresponding to the second data subset, determine the first target transmission channel corresponding to the largest second correlation;

[0016] Determine whether the number of the first target transmission channels meets the preset target number condition;

[0017] If the number of the first target transmission channels does not meet the target number condition, multiple average correlation ratios are determined based on the first target correlation and the second target correlation; the first target correlation is the sum of the first correlation corresponding to the target transmission channel and the first correlation corresponding to any one of the other transmission channels; the second target correlation is the sum of the second correlation corresponding to the target transmission channel and the second correlation corresponding to any one of the other transmission channels.

[0018] Among the plurality of average correlation ratios, the second target transmission channel corresponding to the largest average correlation ratio is determined;

[0019] If the number of the first target transmission channel and the second target transmission channel meets the target quantity condition, the data of the target transmission channel is determined.

[0020] In one embodiment, constructing a comfort evaluation sample based on the data from the target transmission channel and the second data subset includes:

[0021] According to the preset correspondence rules, determine the symmetrical transmission channel corresponding to each target transmission channel;

[0022] Data from each of the symmetrical transmission channels is acquired. Data from the first transmission channel and any one of the symmetrical transmission channels corresponding to the first transmission channel are combined with data from other target transmission channels besides the first transmission channel to obtain a first initial data subset. The first transmission channel is any one of the target transmission channels.

[0023] Based on a preset downsampling factor, the data of each target transmission channel in the first initial data subset is grouped to obtain a first target dataset of the number of downsampling factors.

[0024] Comfort evaluation samples are constructed based on multiple first target datasets and second data subsets.

[0025] In one embodiment, constructing comfort evaluation samples based on a plurality of first target datasets and second data subsets includes:

[0026] For each of the first target datasets, time-frequency features, time-domain features, and frequency-domain features are extracted from the first target dataset according to a preset feature extraction algorithm to obtain a feature dataset;

[0027] A comfort evaluation sample is constructed based on the feature dataset and the second data subset.

[0028] In one embodiment, after constructing a comfort evaluation sample based on the first data and the second data subset of the target transmission channel, the method includes:

[0029] The feature dataset from the comfort evaluation sample set is input into a preset time recurrent neural network for data processing to obtain a predicted comfort score.

[0030] The loss result is obtained by performing loss calculation based on the preset loss algorithm, the predicted comfort score, and the second data in the second data subset of the comfort evaluation sample;

[0031] When the loss result meets the preset loss condition, a trained recurrent neural network is obtained, and the trained recurrent neural network is used as the comfort evaluation model; the comfort evaluation model is used to output the real-time comfort score of the vehicle.

[0032] Secondly, this application also provides a comfort evaluation sample construction device. The device includes:

[0033] An acquisition module is used to acquire a test dataset, which includes a first data subset and a second data subset; the first data subset contains data from multiple transmission channels.

[0034] The determining module is used to determine a first correlation between the first data means corresponding to any two transmission channels in the first data subset, and to determine a second correlation between the first data means corresponding to any transmission channel in the first data subset and the second data means corresponding to the second data subset; the first data means are obtained by performing root mean square value processing on the data of the transmission channels; the second data means are obtained by performing mean processing on each second data in the second data subset;

[0035] The filtering module is used to filter the data of each transmission channel included in the first data subset according to the first correlation, the second correlation and the preset filtering conditions, and determine the data of the target transmission channel.

[0036] A construction module is used to construct comfort evaluation samples based on the data from the target transmission channel and the second data subset; the comfort evaluation samples are used to train a comfort evaluation model to obtain a comfort score.

[0037] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to perform the following steps:

[0038] Obtain a test dataset, which includes a first data subset and a second data subset; the first data subset contains data from multiple transmission channels;

[0039] A first correlation is determined between the first data means corresponding to any two transmission channels in the first data subset, and a second correlation is determined between the first data means corresponding to any transmission channel in the first data subset and the second data means corresponding to the second data subset; the first data means are obtained by performing root mean square value processing on the data of the transmission channels; the second data means are obtained by performing mean processing on each second data in the second data subset;

[0040] Based on the first correlation, the second correlation, and the preset filtering conditions, the data of each transmission channel contained in the first data subset are filtered to determine the data of the target transmission channel.

[0041] A comfort evaluation sample is constructed based on the data from the target transmission channel and the second data subset; the comfort evaluation sample is used to train the comfort evaluation model to obtain a comfort score.

[0042] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, performs the following steps:

[0043] Obtain a test dataset, which includes a first data subset and a second data subset; the first data subset contains data from multiple transmission channels;

[0044] A first correlation is determined between the first data means corresponding to any two transmission channels in the first data subset, and a second correlation is determined between the first data means corresponding to any transmission channel in the first data subset and the second data means corresponding to the second data subset; the first data means are obtained by performing root mean square value processing on the data of the transmission channels; the second data means are obtained by performing mean processing on each second data in the second data subset;

[0045] Based on the first correlation, the second correlation, and the preset filtering conditions, the data of each transmission channel contained in the first data subset are filtered to determine the data of the target transmission channel.

[0046] A comfort evaluation sample is constructed based on the data from the target transmission channel and the second data subset; the comfort evaluation sample is used to train the comfort evaluation model to obtain a comfort score.

[0047] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, performs the following steps:

[0048] Obtain a test dataset, which includes a first data subset and a second data subset; the first data subset contains data from multiple transmission channels;

[0049] A first correlation is determined between the first data means corresponding to any two transmission channels in the first data subset, and a second correlation is determined between the first data means corresponding to any transmission channel in the first data subset and the second data means corresponding to the second data subset; the first data means are obtained by performing root mean square value processing on the data of the transmission channels; the second data means are obtained by performing mean processing on each second data in the second data subset;

[0050] Based on the first correlation, the second correlation, and the preset filtering conditions, the data of each transmission channel contained in the first data subset are filtered to determine the data of the target transmission channel.

[0051] A comfort evaluation sample is constructed based on the data from the target transmission channel and the second data subset; the comfort evaluation sample is used to train the comfort evaluation model to obtain a comfort score.

[0052] The aforementioned comfort evaluation sample construction method, apparatus, computer equipment, storage medium, and computer program product acquire a test dataset, which includes a first data subset and a second data subset. The first data subset contains data from multiple transmission channels. A first correlation is determined between the first data means corresponding to any two transmission channels in the first data subset, and a second correlation is determined between the first data means corresponding to any transmission channel in the first data subset and the second data means corresponding to the second data subset. The first data means are obtained by performing root mean square value processing on the data of the transmission channels. The second data means are obtained by performing mean processing on each second data in the second data subset. Based on the first correlation, the second correlation, and preset filtering conditions, the data of each transmission channel included in the first data subset are filtered to determine the data of the target transmission channel. A comfort evaluation sample is constructed based on the data of the target transmission channel and the second data subset. The comfort evaluation sample is used to train a comfort evaluation model to obtain a comfort score. This method uses first and second correlations, along with preset screening conditions, to filter target transmission channels, determine the data corresponding to the target transmission channels, clarify the data of transmission channels that accurately reflect passenger comfort, and construct comfort evaluation samples based on the data of the target transmission channels and the second data subset, thereby obtaining comfort evaluation samples that accurately reflect passenger comfort and improving the efficiency of the comfort evaluation model. Attached Figure Description

[0053] Figure 1 This is a flowchart illustrating a method for constructing comfort evaluation samples in one embodiment;

[0054] Figure 2 This is a flowchart illustrating the steps involved in constructing the experimental dataset in one embodiment.

[0055] Figure 3 This is a flowchart illustrating the steps for determining the first and second data mean in one embodiment.

[0056] Figure 4 This is a flowchart illustrating the data steps for determining the target transmission channel in one embodiment;

[0057] Figure 5 This is a flowchart illustrating the steps for constructing the optimal channel subset in one embodiment;

[0058] Figure 6This is a flowchart illustrating the steps of determining a first target dataset and constructing comfort evaluation samples in one embodiment.

[0059] Figure 7 This is a flowchart illustrating the data augmentation steps in one embodiment;

[0060] Figure 8 This is a flowchart illustrating the steps involved in constructing a comfort evaluation sample in one embodiment.

[0061] Figure 9 This is a flowchart illustrating the steps of training a comfort evaluation model in one embodiment;

[0062] Figure 10 This is a flowchart illustrating a method for feature extraction and comfort score prediction in one embodiment;

[0063] Figure 11 This is a flowchart illustrating a vehicle comfort evaluation model in one embodiment.

[0064] Figure 12 This is a flowchart illustrating the real-time prediction of comfort steps in one embodiment;

[0065] Figure 13 This is a structural block diagram of a comfort evaluation sample construction device in one embodiment;

[0066] Figure 14 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0067] To make the objectives, technical solutions, and advantages of this application 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 application.

[0068] In one embodiment, such as Figure 1 As shown, a method for constructing comfort evaluation samples is provided. Taking the application of this method to a computer device as an example, this application embodiment does not limit the execution device for executing the comfort evaluation sample construction method, and includes the following steps:

[0069] Step 102: Obtain the experimental dataset.

[0070] The experimental dataset comprises a first subset and a second subset. The first subset contains data from multiple transmission channels. Each transmission channel corresponds to one sensor.

[0071] In implementation, the computer equipment acquires data from multiple sensors within a preset time period through multiple transmission channels. Simultaneously, the computer equipment acquires comfort rating data from multiple passengers within the same preset time period. Then, the computer equipment constructs a second dataset based on the comfort rating data and a first dataset based on the data from the multiple transmission channels. Next, the computer equipment divides the first and second datasets within the preset time period according to a preset time cycle, obtaining multiple first data subsets and multiple second data subsets. The computer equipment then constructs an experimental dataset for each first data subset and each second data subset.

[0072] For example, such as Figure 2 As shown, the computer equipment acquires data from each vibration acceleration channel, obtaining multiple objective data points. Based on these objective data points, the computer equipment constructs a first dataset. Each acceleration transmission channel corresponds to one acceleration sensor. Multiple acceleration sensors are located at the seats, roof, windows, and floor of the vehicle compartment. Then, the computer equipment acquires subjective evaluation scores from multiple individuals, obtaining multiple subjective data points. Based on these subjective data points, the computer equipment constructs a second data subset. Then, according to a preset time period, the computer equipment divides the first and second datasets, obtaining multiple first and second data subsets. For each first and second data subset, the computer equipment constructs an experimental dataset, as shown below. Figure 2 The comfort test dataset shown.

[0073] Optionally, the number and type of sensors are determined according to the data acquisition requirements. This application embodiment does not limit the number and type of sensors for data acquisition.

[0074] Optionally, the time period can be 1 second, 5 seconds, or 10 seconds, depending on the collection frequency of the acquired comfort score data and training requirements. This application does not limit the size of the time period.

[0075] Optionally, the number of the first and second data subsets obtained by the partitioning is greater than 300. This application does not limit the specific number of the first and second data subsets obtained by the partitioning.

[0076] Step 104: Determine the first correlation between the first data mean values ​​corresponding to any two transmission channels in the first data subset, and determine the second correlation between the first data mean value corresponding to any transmission channel in the first data subset and the second data mean value corresponding to the second data subset.

[0077] The first data mean is obtained by applying the root mean square value to the data from the transmission channels; the second data mean is obtained by applying the mean value to each second data point in the second data subset. The first correlation between each transmission channel and its corresponding first data mean is 1.

[0078] In implementation, the computer device stores a correlation algorithm. The computer device constructs a first mean matrix based on the first data mean corresponding to the transmission channel in each first subset of the first dataset. Using the correlation algorithm, the computer device performs a correlation operation on any two rows of the first data mean in the first mean matrix to obtain the first correlation between the first data mean corresponding to any two transmission channels in the first data subset. Then, using the correlation algorithm, the computer device performs a correlation operation on any row of the first data mean in the first mean matrix and the second data mean corresponding to the second data subset to obtain the second correlation between the first data mean corresponding to any transmission channel in the first data subset and the second data mean corresponding to the second data subset.

[0079] Specifically, the computer device constructs a first mean matrix based on the first data mean corresponding to the transmission channel in each first subset of the first dataset. The computer device stores a first correlation function as shown in formula (1).

[0080] Q = c(x) (i) ,x (j) ) i,j∈S (1)

[0081] In formula (1), Q represents the first correlation. c is the generalized correlation function. i is the i-th row of the first mean matrix, and j is the j-th row of the first mean matrix. S is the set of all transmission channels. x (i) x is the mean of the first data in the i-th row of the first mean matrix. (j) Let be the first data mean of the j-th row in the first mean matrix. The computer device performs a correlation operation on the first data mean of any two rows in the first mean matrix according to the first correlation function shown in formula (1), and obtains the first correlation between the first data mean of any two transmission channels in the first data subset.

[0082] The computer device stores a second correlation function as shown in formula (2).

[0083] R = c(x) (j) ,y)j∈S (2)

[0084] In formula (2), R represents the second correlation. c is the generalized correlation function. j is the j-th row in the first mean matrix. S is the set of all transmission channels. (j)Let y be the first data mean in the j-th row of the first mean matrix. Let y be the second data mean corresponding to the second data subset. The computer device performs a correlation operation on the first data mean in any row of the first mean matrix and the second data mean corresponding to the second data subset according to the second correlation function shown in formula (2), and obtains the second correlation between the first data mean corresponding to any transmission channel in the first data subset and the second data mean corresponding to the second data subset.

[0085] For example, the first dataset contains 10 first data subsets, and each first data subset contains three transmission channels. The target first data subset contains three transmission channels A, B, and C, where the mean value of the first data corresponding to transmission channel A is x. (A) The mean value of the first data corresponding to transmission channel B is x. (B) The mean value of the first data corresponding to transmission channel C is x. (C) The mean of the second data set corresponding to the second data subset is y. a .

[0086] The computer device constructs a first mean matrix based on the first data mean corresponding to the transmission channel in each first data subset of the first dataset. Using a correlation algorithm, the computer device performs correlation calculations on any two rows of the first data mean in the first mean matrix to obtain the first correlation Q corresponding to the AA transmission channel in the target first data subset. AA The first correlation Q corresponding to the AB transmission channel AB The first correlation Q corresponding to the BA transmission channel BA The first correlation Q corresponds to the BB transmission channel. BB The first correlation Q corresponding to the BC transmission channel BC The first correlation Q corresponding to the CB transmission channel CB The first correlation Q of the AC transmission channel AC The CA transmission channel corresponds to the first correlation Q. CA The first correlation Q corresponds to the CC transmission channel. CC Q AA Q BB Q CC The value is 1.

[0087] Then, the computer device performs a correlation calculation on the mean of the first data in any row of the first mean matrix and the mean of the second data in the corresponding second data subset, based on a correlation algorithm, to obtain the correlation between transmission channel A and y in the target first data subset. a The second correlation R between them Aa B transmission channel and y a The second correlation R between them Ba C transmission channel and y aThe second correlation R between them Ca .

[0088] Optionally, the generalized correlation function can be the Pearson correlation coefficient, the maximum information coefficient, etc. The specific function of the generalized correlation function is not limited in this embodiment.

[0089] Step 106: Based on the first correlation, the second correlation, and the preset filtering conditions, filter the data of each transmission channel contained in the first data subset to determine the data of the target transmission channel.

[0090] In practice, the computer equipment stores filtering criteria. Based on the first correlation, the second correlation, and the filtering criteria, the computer equipment filters the data of the transmission channels in the first dataset to obtain the data of the target transmission channel.

[0091] Step 108: Construct a comfort evaluation sample based on the data from the target transmission channel and the second data subset.

[0092] Among them, the comfort evaluation samples are used to train the comfort evaluation model to obtain the comfort score.

[0093] In implementation, the computer equipment determines a first target dataset based on data from the target transmission channel. Then, the computer equipment constructs a comfort evaluation sample based on the first target dataset and a second data subset.

[0094] In this embodiment, the target transmission channel is filtered by the first correlation and the second correlation, as well as the preset filtering conditions, to determine the data corresponding to the target transmission channel. The data of the transmission channel that accurately reflects passenger comfort is identified, and a comfort evaluation sample is constructed based on the data of the target transmission channel and the second data subset. This results in a comfort evaluation sample that accurately reflects passenger comfort, thereby improving the efficiency of the comfort evaluation model.

[0095] In one embodiment, such as Figure 3 As shown, before determining the first and second correlations, the data from the transmission channel needs to be processed by root mean square (RMS) to obtain the first data mean, and the second data in the second data subset needs to be processed by averaging to obtain the second data mean. Before step 104 is executed, this comfort evaluation sample construction method also includes:

[0096] Step 302: Perform root mean square (RMS) calculation on the data of each transmission channel in the first data subset to obtain the first data mean value corresponding to each transmission channel.

[0097] In practice, the computer equipment stores the root mean square value function as shown in formula (3).

[0098]

[0099] In formula (3), x (j) rms Let x be the average of the first data corresponding to the j-th transmission channel, and N represent the number of data points in the j-th transmission channel within the first data subset. i (j) This represents the i-th data in the j-th transmission channel. i represents the i-th data in that transmission channel. Let be the sum of squares of the data in the j-th transmission channel. The computer device performs root mean square (RMS) calculation on the data of each transmission channel in the first data subset according to the RMS function shown in formula (3) to obtain the first data mean value corresponding to each transmission channel.

[0100] Step 304: Perform a mean operation on the second data in the second data subset to obtain the mean of the second data.

[0101] In practice, the computer device stores the mean function as shown in formula (4).

[0102]

[0103] In formula (4), C i This represents the comfort rating data of the i-th person in the second data subset, where i represents the i-th person participating in the comfort rating. s This represents the number of people who participated in the comfort rating. (C) mean This represents the mean of the second data set corresponding to the second data subset. This represents the sum of all comfort score data in the second data subset. The computer device performs a mean operation on the second data in the second data subset using a mean function to obtain the mean of the second data corresponding to the second data subset.

[0104] In this embodiment, the root mean square (RMS) value of the data from each transmission channel is processed to obtain the first data mean for each transmission channel. The second data mean is obtained by averaging the second data in the second data subset, which facilitates subsequent processing of the first and second data means to obtain the first and second correlations.

[0105] In one embodiment, such as Figure 4 As shown, step 106 includes:

[0106] Step 402: Determine the first target transmission channel corresponding to the largest second correlation among the second correlation between the first data mean corresponding to any transmission channel in the first data subset and the second data mean corresponding to the second data subset.

[0107] In practice, the computer device determines the first target transmission channel corresponding to the largest second correlation value among the second correlation values ​​between the first data mean value corresponding to any transmission channel in the first data subset and the second data mean value corresponding to the second data subset.

[0108] For example, the first data subset contains three transmission channels A, B, and C, and the second correlation between transmission channel A in the first data subset and the mean of the second data is R. Aa The second correlation between transmission channel B and the second data mean is R. Ba The second correlation between the C transmission channel and the second data mean is R. Ca If R Aa When the value is the largest, the computer equipment will identify transmission channel A as the first target transmission channel.

[0109] Step 404: Determine whether the number of the first target transmission channels meets the preset target number condition.

[0110] In practice, the computer equipment stores the target quantity conditions. The computer equipment determines whether the number of the first target transmission channels meets the target quantity conditions.

[0111] Optionally, the target quantity condition is determined by the training requirements, and this embodiment of the application does not limit the target quantity condition.

[0112] Step 406: If the number of first target transmission channels does not meet the target number condition, determine multiple average correlation ratios based on the first target correlation and the second target correlation.

[0113] The target's first correlation is the sum of the first correlation corresponding to the target transmission channel and the first correlation corresponding to any one of the other transmission channels. The target's second correlation is the sum of the second correlation corresponding to the target transmission channel and the second correlation corresponding to any one of the other transmission channels.

[0114] In implementation, the computer equipment stores an average correlation ratio algorithm. Based on the average correlation ratio algorithm, the computer equipment calculates the first and second correlations of the target, obtaining the average correlation ratios for multiple transmission channels.

[0115] Specifically, the computer device stores the average first correlation formula (5).

[0116]

[0117] In the above formula (5), Q m S represents the average first correlation corresponding to the set of the target transmission channel and any one of the other transmission channels.m x is the set of the target transmission channel and any one of the other transmission channels. (i) x is the mean of the first data in the i-th row of the first mean matrix. (j) Let S be the mean of the first data point in the j-th row of the first mean matrix. m | 2 It represents the square of the absolute value of the number of transmission channels in the set. This represents the sum of the first correlations corresponding to all transmission channels in the set.

[0118] The computer device calculates the first correlation corresponding to the target transmission channel and the first data mean corresponding to any one of the other transmission channels according to formula (5), to obtain the average first correlation corresponding to the set of the target transmission channel and any one of the other transmission channels. The computer device stores the average second correlation formula (6).

[0119]

[0120] In the above formula (6), R m S represents the average second correlation corresponding to the set of the target transmission channel and any one of the other transmission channels. m x is the set of the target transmission channel and any one of the other transmission channels. (i) Let be the mean of the first data in the i-th row of the first mean matrix, and y represent the mean of the second data in the second subset. |S m | represents the absolute value of the number of transmission channels in this set. S m This represents the sum of the second correlations corresponding to all transmission channels in the set.

[0121] The computer device calculates the first correlation corresponding to the target transmission channel and the first data mean corresponding to any one of the other transmission channels and the second data mean corresponding to the second data subset according to formula (6) to obtain the average second correlation corresponding to the set of the target transmission channel and any one of the other transmission channels.

[0122] Specifically, the computer device stores the formula for the average correlation ratio (7).

[0123]

[0124] In formula (7) above, x is the average correlation ratio between the target transmission channel and any one of the other transmission channels. R is the average second correlation between the target transmission channel and any one of the other transmission channels. Q is the average first correlation between the target transmission channel and any one of the other transmission channels. S m Let be the set of any one of the target transmission channels and other transmission channels. Let i and j represent the i-th and j-th transmission channels in this set, respectively. The computer device calculates the ratio of the average first correlation and the average second correlation corresponding to the set of any one of the target transmission channels and other transmission channels according to formula (7), and obtains multiple average correlation ratios.

[0125] For example, the first data subset contains three transmission channels A, B, and C, and the first correlation corresponding to transmission channel AA in the first data subset is Q. AA The first correlation corresponding to the AB transmission channel is Q. AB The first correlation for the BA transmission channel is Q. BA The first correlation for the BB transmission channel is Q. BB The first correlation corresponding to the BC transmission channel is Q. BC The first correlation for the CB transmission channel is attribute Q. CB The first correlation for the AC transmission channel is Q. AC The first correlation between the CA transmission channel and the property Q is... CA The first correlation for the CC transmission channel is Q. CC The second correlation between transmission channel A in the first data subset and the mean of the second data is R. Aa The second correlation between transmission channel B and the second data mean is R. Ba The second correlation between the C transmission channel and the second data mean is R. Ca The first target transmission channel is transmission channel A. The computer equipment determines the first and second target correlations corresponding to the sets of transmission channels A and B, and the first and second target correlations corresponding to the set of transmission channels AC.

[0126] Specifically, the computer device will Q AA Q AB Q BA Q BB Perform an addition operation to obtain the target first correlation Q corresponding to the set of AB transmission channels. B Then, the computer device will R Aa B Ba By performing an additive operation, we obtain the target second correlation R corresponding to the set of AB transmission channels. BComputer equipment will Q AA Q AC Q CA Q CC Perform an addition operation to obtain the target first correlation Q corresponding to the set of AC transmission channels. C Then, the computer device will R Aa R Ca Perform an addition operation to obtain the target second correlation R corresponding to the set of AC transmission channels. C .

[0127] Then, the computer equipment calculates the correlation ratio of the target first correlation and the target second correlation corresponding to the set of AB transmission channels according to the average correlation ratio algorithm, and obtains the average correlation ratio corresponding to the set of AB transmission channels. The computer equipment calculates the correlation ratio of the target first correlation and the target second correlation corresponding to the set of AC transmission channels, and obtains the average correlation ratio corresponding to the set of AC transmission channels.

[0128] Specifically, the number of first correlations corresponding to the set of AB transmission channels is 4. The number of second correlations corresponding to the set of AB transmission channels is 2. The computer device's first correlation Q with the target B The ratio of 4 to 1 yields the average first correlation corresponding to the set of AB transmission channels. The second correlation R of the computer equipment to the target is then calculated. B The average second correlation is calculated by comparing the average second correlation with the average first correlation of the AB transmission channels. The computer device then calculates the average correlation ratio of the AB transmission channels by comparing the average second correlation with the average first correlation. The number of first correlations corresponding to the AC transmission channels is 4. The number of second correlations corresponding to the AC transmission channels is 2. The computer device calculates the target first correlation Q. C The ratio of 4 to the average first correlation of the set of AC transmission channels is calculated to obtain the second correlation R of the computer equipment to the target. C The average second correlation is calculated by comparing the average second correlation with the average first correlation of the AC transmission channels. The computer device then calculates the average correlation ratio of the set of AC transmission channels by comparing the average second correlation with the average first correlation.

[0129] Step 408: Among multiple average correlation ratios, determine the second target transmission channel corresponding to the largest average correlation ratio.

[0130] In practice, the computer equipment determines the second target transmission channel corresponding to the average correlation ratio with the largest value among multiple average correlation ratios.

[0131] For example, a computer device stores a first target transmission channel A. The computer device also stores the average correlation ratios corresponding to the set of transmission channels A and B, and the average correlation ratios corresponding to the set of transmission channels A and C. If the average correlation ratio corresponding to the set of transmission channels A and C is the largest, the computer device determines channel C as the second target transmission channel.

[0132] Step 410: If the number of the first target transmission channel and the second target transmission channel meets the target quantity condition, determine the data of the target transmission channel.

[0133] In practice, when the number of the first target transmission channel and the second target transmission channel meets the target quantity condition, the computer equipment determines the data of the target transmission channel.

[0134] For example, the data volume of two target transmission channels can be used as the target quantity condition. The first target transmission channel is channel A, and the second target transmission channel is channel C. The computer equipment uses the data from transmission channels A and C as the data for the target transmission channel.

[0135] Optionally, if the number of the first target transmission channel and the second target transmission channel does not meet the target quantity condition, the computer device uses the second target transmission channel and the first target transmission channel as the new first target transmission channel. The computer device executes steps 406 to 408 to obtain the new second target transmission channel until the number of the first target transmission channel and the second target transmission channel meets the target quantity condition.

[0136] In this embodiment, the data of the target transmission channel is determined by the first correlation, the second correlation, and the preset filtering conditions, thus clarifying the data of the transmission channel that accurately reflects passenger comfort.

[0137] like Figure 5 As shown, the computer device acquires data from each channel, obtaining multiple objective data points. Simultaneously, the computer device acquires multiple comfort score data points, obtaining multiple subjective data points. The computer constructs an experimental dataset based on the objective and subjective data. Then, the computer device performs dimensionality reduction on the data in the experimental dataset, obtaining the first mean data point corresponding to each transmission channel and the second mean data point corresponding to the second data point. The computer device calculates the correlation between each pair of objective channels, obtaining the first correlation between the first mean data points of any two transmission channels in the first data subset. Simultaneously, the computer device calculates the correlation between each objective channel and the subjective data, obtaining the second correlation between the first mean data point of any transmission channel in the first data subset and the second mean data point of the second data subset. Then, based on the feature selection method of maximum correlation and minimum redundancy, the computer device determines the optimal channel subset, obtaining the data for the target transmission channel.

[0138] In one embodiment, such as Figure 6 As shown, step 108 includes:

[0139] Step 602: Determine the symmetrical transmission channel corresponding to each target transmission channel according to the preset correspondence rules.

[0140] In practice, the computer equipment stores corresponding rules, and the computer equipment determines the symmetrical transmission channel corresponding to each target transmission channel according to the corresponding rules, thus obtaining multiple symmetrical transmission channels.

[0141] For example, target transmission channels are A and B. Transmission channel A and transmission channel A' are symmetrical about the vehicle's centerline. Transmission channel B and transmission channel B' are also symmetrical about the vehicle's centerline. The computer device determines the symmetrical transmission channel corresponding to transmission channel A as transmission channel A'. Then, the computer device determines the symmetrical transmission channel corresponding to transmission channel B as transmission channel B'.

[0142] Step 604: Obtain data from each symmetrical transmission channel, and combine the data from the first transmission channel and any one of the symmetrical transmission channels corresponding to the first transmission channel with the data from other target transmission channels besides the first transmission channel to obtain a first initial data subset.

[0143] The first transmission channel is any one of the target transmission channels;

[0144] In implementation, the computer device acquires data from each symmetrical transmission channel. Then, the computer device determines a first transmission channel within the target transmission channels. The computer device combines the data from the first transmission channel and any one of the corresponding symmetrical transmission channels with the data from the other target transmission channels besides the first transmission channel to obtain a first initial data subset.

[0145] For example, the target transmission channels are A and B. The symmetrical transmission channel corresponding to transmission channel A is transmission channel A'. The symmetrical transmission channel corresponding to transmission channel B is transmission channel B'. The computer device determines A as the first transmission channel. Then, the computer device combines the data from transmission channel A and transmission channel A' with the data from transmission channel B to obtain a first initial data subset composed of data from transmission channels A and B. Then, the computer device combines the data from transmission channels A and B' to obtain a first initial data subset composed of data from transmission channels A and B'. The computer device combines the data from transmission channel A' with the data from transmission channel B to obtain a first initial data subset composed of data from transmission channels A' and B. Then, the computer device combines the data from transmission channels A' and B' to obtain a first initial data subset composed of data from transmission channels A' and B'.

[0146] Step 606: Based on the preset downsampling factor, group the data of each target transmission channel in the first initial data subset to obtain the first target dataset with the downsampling factor.

[0147] In implementation, the computer equipment stores downsampling factors. For each target transmission channel in the first initial data subset, the computer equipment extracts data from that target transmission channel starting from the first data point and at downsampling factor intervals, obtaining a first set of target datasets. Then, starting from the second data point, the computer equipment extracts data from that target transmission channel at downsampling factor intervals, obtaining a second set of target datasets. This process continues until all data from the target transmission channel has been extracted, at which point the computer equipment obtains a first target dataset representing the number of downsampling factors.

[0148] For example, the sampling frequency of the data in the first initial data subset is 512 Hz. The specific value of the downsampling factor is 5. The computer device calculates the ratio of the sampling frequency and the downsampling factor, resulting in a downsampling frequency of 102.4 Hz. The first initial data subset in the computer device contains data from one target transmission channel, which includes data A1, A2, A3, A4, A5, A6, A7, A8, A9, and A10. The first set of data for the first target dataset includes A1 and A6. Then, computer device A2 begins to extract data from the target transmission channel at downsampling factor intervals, obtaining the second set of data for the first target dataset. The second set of data for the first target dataset includes A2 and A7. This process continues until all data from the target transmission channel has been extracted, resulting in five sets of 102.4 Hz first target datasets. The third set of data for the first target dataset includes A3 and A8. The fourth set of data for the first target dataset includes A4 and A9. The fifth set of data for the first target dataset includes A5 and A10.

[0149] Optionally, the specific value of the downsampling factor can be 5, and the frequency after downsampling must be greater than 50Hz. The specific value of the downsampling factor is determined by the training requirements. This application does not limit the specific value of the downsampling factor.

[0150] Step 608: Construct comfort evaluation samples based on multiple first target datasets and second data subsets.

[0151] In implementation, the computer equipment constructs comfort evaluation samples based on the second data in the second data subset and each target dataset, resulting in multiple comfort evaluation samples.

[0152] In this embodiment, multiple first target datasets are obtained by downsampling and channel replacement of the data in the target transmission channel. Comfort evaluation samples are constructed based on the second data in the first target dataset and the second data subset, which increases the number of samples and facilitates the construction of multiple comfort evaluation samples to train the comfort evaluation model.

[0153] like Figure 7 As shown, the computer device downsamples and combines channels on the original dataset to obtain an enhanced dataset, and constructs comfort evaluation samples based on the enhanced dataset and the second data subset to obtain multiple comfort evaluation samples.

[0154] In one embodiment, such as Figure 8 As shown, step 608 includes:

[0155] Step 802: For each first target dataset, extract the time-frequency features, time-domain features, and frequency-domain features from the first target dataset according to the preset feature extraction algorithm to obtain the feature dataset.

[0156] In implementation, the computer device stores a feature extraction algorithm. Based on this algorithm, the computer device extracts features from each first target dataset, obtaining the time-frequency features, time-domain features, and frequency-domain features corresponding to each first target dataset. The computer device then constructs a feature dataset based on these features.

[0157] Specifically, computer devices have a short-time Fourier transform, as shown in formula (8).

[0158]

[0159] In the above formula (8), t and τ represent time t and time τ, respectively, F(ω,τ) refers to the spectrum at time τ, ψ(t) refers to the window function centered at time t, and f(t) is the original signal. j is the imaginary unit, and ω represents the frequency. The computer device performs time-frequency analysis on the data of each first target dataset according to formula (8) to obtain the time-frequency data corresponding to each first target dataset. Then, the computer device stores a feature extraction algorithm. The computer device performs feature extraction on each first target dataset and its corresponding time-frequency data according to the feature extraction algorithm to obtain the time-frequency features, time-domain features, and frequency-domain features corresponding to each first target dataset. The computer device constructs a feature dataset based on the time-frequency features, time-domain features, and frequency-domain features corresponding to each first target dataset.

[0160] Optionally, the window function in the short-time Fourier transform can be a Hanning window, a rectangular window, etc. The specific function of the window function is not limited in the embodiments of this application.

[0161] Optionally, the feature extraction algorithms include time-domain feature extraction algorithms, time-frequency feature extraction algorithms, and frequency-domain feature extraction algorithms. Time-domain feature extraction algorithms extract the mean, variance, root mean square value, steepness, kurtosis, etc., of the spectral data. Time-frequency feature extraction algorithms include algorithms such as short-time Fourier transform and wavelet transform, used to extract features of the signal spectrum changing over time. Frequency-domain feature extraction algorithms extract spectral peak values, peak frequencies, root mean square values ​​of frequency bands, etc. The specific processing steps of the feature extraction algorithms are not limited in this embodiment.

[0162] Step 804: Construct comfort evaluation samples based on the feature dataset and the second data subset.

[0163] In implementation, the computer equipment constructs a set of comfort evaluation samples based on the second data in the second data subset and multiple feature datasets.

[0164] In this embodiment, the time-frequency features, time-domain features, and frequency-domain features of each target dataset are extracted to obtain a feature dataset. Based on the feature dataset and the second data subset, a comfort evaluation sample is constructed, which accurately reflects the comfort of passengers and improves the efficiency of the comfort evaluation model.

[0165] In one embodiment, such as Figure 9 As shown, after obtaining the comfort evaluation samples, these samples are used to train a time recurrent neural network to obtain a comfort evaluation model. After step 108 is completed, the construction of the comfort evaluation samples further includes:

[0166] Step 902: Input the feature dataset from the comfort evaluation sample set into a preset time recurrent neural network for data processing to obtain a predicted comfort score.

[0167] In implementation, the computer device stores a recurrent neural network. The computer device inputs the feature dataset from the comfort evaluation samples into the recurrent neural network. The recurrent neural network processes the comfort evaluation samples and outputs a predicted comfort score.

[0168] For example, such as Figure 10 As shown, the computer equipment processes each target dataset based on the short-time Fourier transform algorithm to obtain the time-frequency data corresponding to each target dataset. Then, the computer equipment extracts time-frequency features, time-domain features, and frequency-domain features from the objective vibration signal to obtain a feature subset. Figure 10 The feature subset in the dataset is the feature dataset. The computer device inputs the feature subset and the second data subset into a time-recurrent neural network to obtain a comfort prediction score. Figure 10 The comfort prediction score is the predicted comfort score.

[0169] Step 904: Perform loss calculation based on the preset loss algorithm, the predicted comfort score, and the second data in the second data subset of the comfort evaluation sample to obtain the loss result.

[0170] In implementation, the computer equipment stores a loss algorithm. Based on the loss algorithm, the computer equipment performs calculations on the second data in the second data subset of multiple comfort evaluation samples within a preset time period, as well as multiple predicted comfort scores corresponding to the multiple comfort evaluation samples, to obtain the loss result.

[0171] Specifically, the computer device stores a loss function as shown in formula (9).

[0172]

[0173] In the above formula (9), y represents the predicted comfort score for the i-th comfort evaluation sample among multiple sets of comfort evaluation samples. i This represents the mean of the i-th second data point. Loss MAE denoted as mean absolute error. N represents the number of comfort evaluation samples within a preset time period. The computer device calculates the loss value by performing calculations on the second data in the second data subset of multiple comfort evaluation samples within the preset event period and on the multiple predicted comfort scores corresponding to the multiple comfort evaluation samples, based on the loss function shown in formula (9).

[0174] Optionally, the loss algorithm can be the mean absolute error. The loss algorithm is determined by the training requirements. This application embodiment does not limit the specific processing procedure of the loss algorithm.

[0175] Step 906: When the loss result meets the preset loss condition, the trained recurrent neural network is obtained, and the trained recurrent neural network is used as a comfort evaluation model.

[0176] The comfort evaluation model is used to output the real-time comfort score of the vehicle.

[0177] In implementation, the computer device stores the loss conditions. When the loss result meets the loss conditions, the computer device obtains a trained recurrent neural network. The computer device uses the trained recurrent neural network as a comfort evaluation model.

[0178] For example, the loss condition stored in the computer device is such that when the loss value begins to converge and stabilize, the computer device obtains a trained recurrent neural network. When the loss value is less than or equal to the product of 0.05 and the mean of the second data corresponding to the second subset, the computer device uses the trained recurrent neural network as a comfort evaluation model.

[0179] In this embodiment, the computer device inputs the constructed comfort samples into the recurrent time neural network. When the loss result meets the preset loss condition, the trained recurrent time neural network is obtained. The trained recurrent time neural network is then used as a comfort evaluation model that accurately outputs the real-time comfort score of the vehicle, thereby improving the efficiency of the comfort evaluation model.

[0180] like Figure 11 As shown, after selecting the target transmission channel, the computer acquires data collected by the sensors corresponding to that channel. Then, the computer constructs a comfort evaluation model. Based on this model, the computer predicts comfort levels and obtains a real-time comfort score.

[0181] like Figure 12As shown, the computer equipment acquires objective vibration acceleration signals and subjective comfort scores from sensors located at several positions on the vehicle, resulting in a first dataset and a second dataset. The computer equipment then preprocesses the data, segmenting the objective data and combining it with the subjective comfort scores to construct a comfort dataset, thus obtaining the experimental dataset. Based on a feature selection method of maximum correlation and minimum redundancy, the computer equipment identifies several sensor channels that have the most critical impact on comfort, determining the data for the target transmission channels. Then, the computer equipment uses data augmentation methods to increase the number of training samples in the dataset, obtaining multiple first target datasets. The computer equipment extracts the time-frequency features of the objective data, trains and constructs a comfort scoring model, obtaining a comfort evaluation model. The computer equipment then applies this comfort evaluation model to a real vehicle, predicting comfort in real time and obtaining a comfort score.

[0182] 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.

[0183] Based on the same inventive concept, this application also provides a comfort evaluation sample construction apparatus for implementing the comfort evaluation sample construction method described above. The solution provided by this apparatus is similar to the solution described in the above method; therefore, the specific limitations in one or more comfort evaluation sample construction apparatus embodiments provided below can be found in the limitations of the comfort evaluation sample construction method described above, and will not be repeated here.

[0184] In one embodiment, such as Figure 13 As shown, a comfort evaluation sample construction device is provided, including: an acquisition module 1301, a determination module 1302, a filtering module 1303, and a construction module 1304, wherein:

[0185] The acquisition module 1301 is used to acquire a test dataset, which includes a first data subset and a second data subset; the first data subset includes data from multiple transmission channels.

[0186] The determining module 1302 is used to determine the first correlation between the first data means corresponding to any two transmission channels in the first data subset, and to determine the second correlation between the first data means corresponding to any transmission channel in the first data subset and the second data means corresponding to the second data subset; the first data means is obtained by performing root mean square value processing on the data of the transmission channels; the second data means is obtained by performing mean processing on each second data in the second data subset.

[0187] The filtering module 1303 is used to filter the data of each transmission channel included in the first data subset according to the first correlation, the second correlation and the preset filtering conditions, and to determine the data of the target transmission channel.

[0188] The construction module 1304 is used to construct a comfort evaluation sample based on the data of the target transmission channel and the second data subset; the comfort evaluation sample is used to train the comfort evaluation model to obtain a comfort score.

[0189] In one exemplary embodiment, before the determining module 1302 performs its operation, the comfort evaluation sample construction apparatus further includes:

[0190] The first calculation submodule is used to perform root mean square value calculation on the data of each transmission channel in the first data subset to obtain the first data mean value corresponding to each transmission channel.

[0191] The second operation submodule is used to perform a mean operation on the second data in the second data subset to obtain the mean of the second data.

[0192] In one exemplary embodiment, the filtering module 1303 includes:

[0193] The first determining submodule is used to determine the first target transmission channel corresponding to the maximum second correlation among the second correlation between the first data mean corresponding to any transmission channel in the first data subset and the second data mean corresponding to the second data subset.

[0194] The judgment submodule is used to determine whether the number of the first target transmission channels meets the preset target number condition.

[0195] The second determining submodule is used to determine multiple average correlation ratios based on the first target correlation and the second target correlation when the number of the first target transmission channels does not meet the target number condition; the first target correlation is the sum of the first correlation corresponding to the target transmission channel and the first correlation corresponding to any one of the other transmission channels; the second target correlation is the sum of the second correlation corresponding to the target transmission channel and the second correlation corresponding to any one of the other transmission channels.

[0196] The third determining submodule is used to determine the second target transmission channel corresponding to the largest average correlation ratio among the plurality of average correlation ratios.

[0197] The fourth determining submodule is used to determine the data of the target transmission channel when the number of the first target transmission channel and the second target transmission channel meets the target quantity condition.

[0198] In one exemplary embodiment, the construction module 1304 includes:

[0199] The fifth determining submodule is used to determine the symmetrical transmission channel corresponding to each target transmission channel according to the preset corresponding rules.

[0200] The combination submodule is used to acquire data from each of the symmetrical transmission channels, and combine the data from the first transmission channel and any one of the symmetrical transmission channels corresponding to the first transmission channel with the data from other target transmission channels besides the first transmission channel to obtain a first initial data subset; the first transmission channel is any one of the target transmission channels.

[0201] The grouping submodule is used to group the data of each target transmission channel in the first initial data subset according to a preset downsampling factor, so as to obtain a first target dataset of the number of downsampling factors.

[0202] The first construction submodule is used to construct comfort evaluation samples based on multiple first target datasets and the second data subset.

[0203] In one exemplary embodiment, the first construction submodule includes:

[0204] The extraction submodule is used to extract time-frequency features, time-domain features, and frequency-domain features from each of the first target datasets according to a preset feature extraction algorithm, so as to obtain a feature dataset.

[0205] The second construction submodule is used to construct comfort evaluation samples based on the feature dataset and the second data subset.

[0206] In one exemplary embodiment, after the construction module 1304 performs its operation, the comfort evaluation sample construction apparatus further includes:

[0207] The processing submodule is used to input the feature dataset from the comfort evaluation sample set into a preset time recurrent neural network for data processing to obtain a predicted comfort score.

[0208] The third calculation submodule is used to perform loss calculation based on the preset loss algorithm, the predicted comfort score and the second data in the second data subset of the comfort evaluation sample, to obtain the loss result.

[0209] A submodule is obtained to obtain a trained recurrent neural network when the loss result meets a preset loss condition, and the trained recurrent neural network is used as the comfort evaluation model; the comfort evaluation model is used to output the real-time comfort score of the vehicle.

[0210] Each module in the aforementioned comfort evaluation sample construction 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 a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0211] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 14 As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database stores data from the experimental dataset. The network interface communicates with external terminals via a network connection. When executed by the processor, the computer program implements a method for constructing comfort evaluation samples.

[0212] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 14As shown, the computer device includes a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When executed by the processor, the computer program implements a method for constructing comfort evaluation samples. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the computer device casing, or an external keyboard, touchpad, or mouse.

[0213] Those skilled in the art will understand that Figure 14 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0214] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0215] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0216] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0217] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0218] Those skilled in the art will understand that all or part of the processes in the methods of 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 application 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 application 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 application 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.

[0219] 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.

[0220] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for constructing comfort evaluation samples, characterized in that, The method includes: A test dataset is obtained, which includes a first data subset and a second data subset; the first data subset includes data from multiple transmission channels; each transmission channel is used to transmit objective data collected by a corresponding acceleration sensor, which is located at various locations in the vehicle compartment; A first correlation is determined between the first data means corresponding to any two transmission channels in the first data subset, and a second correlation is determined between the first data means corresponding to any transmission channel in the first data subset and the second data means corresponding to the second data subset; the first data means are obtained by performing root mean square value processing on the data of the transmission channels; the second data means are obtained by performing mean processing on each second data in the second data subset; In the second correlation between the first data mean corresponding to any transmission channel in the first data subset and the second data mean corresponding to the second data subset, determine the first target transmission channel corresponding to the largest second correlation; Determine whether the number of the first target transmission channels meets the preset target number condition; If the number of the first target transmission channels does not meet the target number condition, multiple average correlation ratios are determined based on the first target correlation and the second target correlation; the first target correlation is the sum of the first correlation corresponding to the target transmission channel and the first correlation corresponding to any one of the other transmission channels; the second target correlation is the sum of the second correlation corresponding to the target transmission channel and the second correlation corresponding to any one of the other transmission channels. Among the plurality of average correlation ratios, the second target transmission channel corresponding to the largest average correlation ratio is determined; If the number of the first target transmission channel and the second target transmission channel meets the target number condition, the data of the target transmission channel is determined; According to the preset correspondence rules, determine the symmetrical transmission channel corresponding to each target transmission channel; Data from each of the symmetrical transmission channels is acquired. Data from the first transmission channel and any one of the symmetrical transmission channels corresponding to the first transmission channel are combined with data from other target transmission channels besides the first transmission channel to obtain a first initial data subset. The first transmission channel is any one of the target transmission channels. Based on a preset downsampling factor, the data of each target transmission channel in the first initial data subset is grouped to obtain a first target dataset of the number of downsampling factors. For each of the first target datasets, time-frequency features, time-domain features, and frequency-domain features are extracted from the first target dataset according to a preset feature extraction algorithm to obtain a feature dataset; Comfort evaluation samples are constructed based on the feature dataset and the second data subset; the comfort evaluation samples are used to train the comfort evaluation model to obtain a comfort score.

2. The method according to claim 1, characterized in that, Before determining the first correlation between the first data means corresponding to any two transmission channels in the first data subset, the method further includes: The root mean square value is calculated on the data of each transmission channel in the first data subset to obtain the first data mean value corresponding to each transmission channel. The mean of the second data in the second subset of data is calculated by performing a mean operation.

3. The method according to claim 1, characterized in that, The step of determining the first correlation between the first data means corresponding to any two transmission channels in the first data subset, and determining the second correlation between the first data means corresponding to any transmission channel in the first data subset and the second data means corresponding to the second data subset, includes: Construct a first mean matrix based on the first data mean corresponding to the transmission channel in each first data subset of the first dataset; Based on a preset correlation algorithm, a correlation operation is performed on the first data mean values ​​of any two rows in the first mean matrix to obtain the first correlation between the first data mean values ​​of any two transmission channels in the first data subset. According to the correlation algorithm, a correlation operation is performed on the first data mean of any row in the first mean matrix and the second data mean corresponding to the second data subset to obtain a second correlation between the first data mean corresponding to any transmission channel in the first data subset and the second data mean corresponding to the second data subset.

4. The method according to claim 1, characterized in that, The step of grouping the data of each target transmission channel in the first initial data subset according to a preset downsampling factor to obtain a first target dataset of the number of downsampling factors includes: For the data of each target transmission channel in the first initial data subset, the data of the target transmission channel is extracted at intervals of a preset downsampling factor, starting from the first data of the target transmission channel, to obtain the first set of the first target dataset; Starting from the second data point of the target transmission channel, data from the target transmission channel is extracted at intervals according to the downsampling factor to obtain a second set of first target datasets, until the extraction of all data from the target transmission channel is completed, resulting in a first target dataset with the downsampling factor.

5. The method according to claim 1, characterized in that, For each of the first target datasets, time-frequency features, time-domain features, and frequency-domain features are extracted from the first target dataset according to a preset feature extraction algorithm to obtain a feature dataset, including... According to the preset feature extraction algorithm, feature extraction is performed on each of the first target datasets to obtain the time-frequency features, time-domain features and frequency-domain features corresponding to each of the first target datasets; A feature dataset is constructed based on the time-frequency features, time-domain features, and frequency-domain features corresponding to each of the first target datasets.

6. The method according to claim 1, characterized in that, After constructing the comfort evaluation samples based on the feature dataset and the second data subset, the method includes: The feature dataset from the comfort evaluation samples is input into a preset time recurrent neural network for data processing to obtain a predicted comfort score. The loss result is obtained by performing loss calculation based on the preset loss algorithm, the predicted comfort score, and the second data in the second data subset of the comfort evaluation sample; When the loss result meets the preset loss condition, a trained recurrent neural network is obtained, and the trained recurrent neural network is used as the comfort evaluation model; the comfort evaluation model is used to output the real-time comfort score of the vehicle.

7. A comfort evaluation sample construction device, characterized in that, The device includes: An acquisition module is used to acquire a test dataset, which includes a first data subset and a second data subset; the first data subset includes data from multiple transmission channels; each transmission channel is used to transmit objective data collected by a corresponding acceleration sensor, which is located at various locations in the vehicle compartment; The determining module is used to determine a first correlation between the first data means corresponding to any two transmission channels in the first data subset, and to determine a second correlation between the first data means corresponding to any transmission channel in the first data subset and the second data means corresponding to the second data subset; the first data means are obtained by performing root mean square value processing on the data of the transmission channels; the second data means are obtained by performing mean processing on each second data in the second data subset; The first determining submodule is used to determine the first target transmission channel corresponding to the maximum second correlation among the second correlation between the first data mean corresponding to any transmission channel in the first data subset and the second data mean corresponding to the second data subset. The judgment submodule is used to determine whether the number of the first target transmission channels meets the preset target number condition; The second determining submodule is used to determine multiple average correlation ratios based on the first target correlation and the second target correlation when the number of the first target transmission channels does not meet the target number condition; the first target correlation is the sum of the first correlation corresponding to the target transmission channel and the first correlation corresponding to any one of the other transmission channels; the second target correlation is the sum of the second correlation corresponding to the target transmission channel and the second correlation corresponding to any one of the other transmission channels. The third determining submodule is used to determine the second target transmission channel corresponding to the maximum average correlation ratio among the plurality of average correlation ratios; The fourth determining submodule is used to determine the data of the target transmission channel when the number of the first target transmission channel and the second target transmission channel meets the target number condition; The fifth determining submodule determines the symmetrical transmission channel corresponding to each target transmission channel according to the preset correspondence rules; The combination submodule is used to acquire data from each of the symmetrical transmission channels, and combine the data from the first transmission channel and any one of the symmetrical transmission channels corresponding to the first transmission channel with the data from other target transmission channels besides the first transmission channel to obtain a first initial data subset; the first transmission channel is any one of the target channels. The grouping submodule is used to group the data of each target transmission channel in the first initial data subset according to a preset downsampling factor, so as to obtain a first target dataset of the number of downsampling factors. The extraction submodule is used to extract time-frequency features, time-domain features, and frequency-domain features from each of the first target datasets according to a preset feature extraction algorithm to obtain a feature dataset. The second construction submodule is used to construct comfort evaluation samples based on the feature dataset and the second data subset; the comfort evaluation samples are used to train the comfort evaluation model to obtain a comfort score.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, 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.

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