Intelligent electric vehicle motion sickness resistance cruise planning method based on subjective and objective comprehensive indexes

Abstract

The invention discloses an anti-motion sickness cruise planning method for an intelligent electric vehicle based on subjective and objective comprehensive indexes. The method comprises the following steps: collecting vehicle motion track data and subjective and objective data of human physiological signals; according to the vehicle motion trajectory data, motion sickness indexes under frequency domain weighting and evaluation indexes under a time domain are calculated through the Parseval theorem and a band-pass filter, the motion sickness indexes and the evaluation indexes are fused through a state-space equation, and time-frequency domain objective motion sickness indexes under vehicle variable motion are formed. Subjective and objective comprehensive indexes of motion sickness are obtained by combining subjective and objective data of human physiological signals; and constructing a cruise planning model by using the obtained subjective and objective comprehensive indexes of the motion sickness and the self-adaptive weighting factor set, and solving optimal anti-motion sickness cruise parameters of the model by combining a dynamic planning method, so that the subjective and objective comprehensive indexes of the motion sickness and the passing time are minimized in the driving process of the vehicle. The method aims at improving the comfort degree of a driver and passengers under the cruise working condition of the intelligent electric vehicle.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent automobile cruise control, and in particular relates to an anti-motion sickness cruise control planning method for an intelligent electric vehicle based on subjective and objective comprehensive indicators. Background Art

[0002] With the rapid development of smart electric vehicle technology, people are freed from driving tasks, and the role of the driver is transformed into a "free-moving" passenger who can read, hold meetings or rest, etc., which requires more stringent riding comfort requirements. At the same time, the kinetic energy recovery system and faster power response speed make the speed of electric vehicles change dramatically when accelerating and braking, especially in crowded urban traffic, electric vehicles are more likely to induce motion sickness in passengers. When smart electric vehicles are used for transportation passenger services, comfort is one of the main concerns. Therefore, it is urgent to further improve the riding comfort of smart electric vehicles and reduce passenger motion sickness to ensure the social benefits and social acceptance brought by electric vehicles and smart driving.

[0003] In terms of reducing vibration stimulation and improving ride comfort, the main methods at this stage are passive methods such as optimizing the layout of the cabin environment and optimizing the user interface. For example, by adjusting the seat position to obtain a wider external road view, reducing the temperature in the car, and optimizing the seat structure. The essence of this passive method is that after the uncomfortable vibration excitation is generated, its impact on the human body is reduced by various means, and the excitation source cannot be actively and directly eliminated. Another way is to optimize vehicle behavior through motion planning control and directly eliminate the vibration excitation source to improve passenger comfort. The motion planning and control of the current high-level intelligent car adaptive cruise technology are fully automated, so this method is feasible in practical applications. The core of this improvement method is to construct a comprehensive and accurate quantitative indicator for motion sickness assessment to cover a variety of comfort influencing factors. Because the subjective evaluation method has the advantages of low cost, easy implementation and high effectiveness, it is widely used in early motion sickness research or treatment. However, the questionnaire can only record the severity of motion sickness of the subjects before and after the start of the experiment, and cannot show the nonlinear changes that may exist in their motion sickness level during driving over time, and it is difficult to directly use it in anti-motion sickness planning strategies. Therefore, we should use the objective data that can be collected in real time and explore the mapping relationship between it and the subjective feelings of the human body. On this basis, we can build a comprehensive subjective and objective index of motion sickness, calculate the current motion sickness level of the occupants in real time, and use it as the objective function of the planning model to minimize the discomfort of motion sickness of the occupants.

[0004] Based on the motion sickness index, constructing a reasonable multi-objective optimization model and incorporating it into the planning and control process is currently a more novel and effective comfort improvement strategy suitable for intelligent driving cars. Kamiji et al. obtained a more comprehensive 6-DOF evaluation model by expanding the traditional SVC model (Subjective vertical conflict model), and developed from a qualitative model to a multi-DOF quantitative model. Subsequently, on this basis, researchers proposed a variety of optimization methods for intelligent driving cars to improve vibration comfort. When using adaptive cruise technology to assist driving under congested conditions, the driving behavior of the vehicle in front has a strong impact on the movement of the vehicle. In order to maintain a reasonable headway, some studies use the acceleration change rate, that is, the impact degree, to analyze vehicle vibration in real time and optimize speed and path. This type of method that only considers the longitudinal impact degree reduces the vehicle speed to too low a level in some scenarios, affecting the travel time. Although it can improve comfort well, it is not comprehensive for improving the driving travel experience (Study on the Impact of Lane Changing Operation on Passenger Comfort of Intelligent Vehicles Guo Yingshi, Su Yanqi, Fu Rui, Yuan Wei, School of Automotive Engineering, Chang'an University, Xi'an 710064). Therefore, the anti-motion sickness planning strategy for smart cars should fully consider the physiological characteristics of the human body while balancing comfort, risk and travel time. The primary criterion target can be adaptively adjusted during driving to effectively respond to various driving scenarios. Summary of the invention

[0005] The purpose of the present invention is to provide an anti-motion sickness cruise planning method for an intelligent electric vehicle based on subjective and objective comprehensive indicators.

[0006] The present invention is achieved by at least one of the following technical solutions.

[0007] The anti-motion sickness cruise planning method for intelligent electric vehicles based on subjective and objective comprehensive indicators includes the following steps: (1) Collect vehicle motion trajectory data and subjective and objective data of human physiological signals; (2) The collected vehicle motion trajectory data is used to calculate the motion sickness index in the frequency domain weighted and the evaluation index in the time domain through Parseval theorem and bandpass filter. The motion sickness index in the frequency domain weighted and the evaluation index in the time domain are fused using the state space equation to form the objective motion sickness index in the time and frequency domain under the vehicle speed change motion. The subjective and objective comprehensive index of motion sickness is obtained by combining the subjective and objective data of human physiological signals. (3) A cruise planning model is constructed using the obtained subjective and objective comprehensive indicators of motion sickness and a set of adaptive weighted factors. The optimal anti-motion sickness cruise parameters of the cruise planning model are solved using a combined dynamic programming method to minimize the subjective and objective comprehensive indicators of motion sickness and the travel time during the vehicle's driving process.

[0008] Furthermore, in step (2), the calculation of the motion sickness index under frequency domain weighting by Parseval theorem and bandpass filter includes: using Parseval theorem to convert the vehicle acceleration of step (1) into frequency domain form, inputting the frequency domain form into the bandpass filter to extract data within the human body sensitive frequency band, and simultaneously adding the weighting function specified in standard ISO2631 to the extracted data within the human body sensitive frequency band, thereby obtaining frequency domain weighted acceleration, and then integrating the frequency domain weighted acceleration to obtain the frequency domain weighted motion sickness measurement value.

[0009] Furthermore, in step (2), the evaluation index in the time domain is calculated as follows: Based on the frequency domain weighted acceleration, the time domain weighted motion sickness measurement value is constructed by combining the inverse Fourier transform to consider the influence of vibration persistence. After the transformation, the real part of the transformation result is taken to obtain the time-domain weighted acceleration, and then the time-domain weighted motion sickness measurement value is calculated using the time-domain weighted acceleration; Two additional time-related random disturbances are added to produce disturbance effects. The time-domain weighted motion sickness measurement value and the disturbance effect are combined to form the evaluation index in the time domain.

[0010] Furthermore, in step (2), the bandpass filter is converted into a continuous time form and discretized, the input of the bandpass filter is a constant value within the discrete time step, and when the vehicle travels between the planning points, the matrix index in the discretization equivalent expression is different at each time step; the matrix index is transformed based on the diagonalization method, and the dynamic characteristic coefficient matrix is ​​diagonalized through the eigenvector matrix, and the dynamic characteristic coefficient matrix is ​​a time constant corresponding to the expected cutoff frequency of the bandpass filter; the eigenvector matrix used in the diagonalization process is an invariant of the time step, and before planning, the value of the eigenvector matrix is ​​predefined according to the actual motion parameters of the vehicle.

[0011] Furthermore, in step (2), the step of combining the subjective and objective data of human physiological signals to obtain a subjective and objective comprehensive index of motion sickness specifically includes: performing time synchronization processing on the subjective and objective data of human physiological signals, and labeling the physiological signal acquisition data according to the subjective feeling quantification data; matching the physiological signal data of each occupant with a reference range to obtain a first parameter deviation of each signal, which is used to characterize whether each physiological signal deviates from the reference range; and calculating a second parameter deviation based on the deviation between the data of each physiological signal and the reference range to characterize the degree of deviation of each physiological signal; and assigning a weight to each physiological signal acquisition data according to a decreasing principle; and then matching the total amount of the first parameter deviation and the total amount of the second parameter deviation with the objective motion sickness index in the time-frequency domain under the vehicle speed change motion to obtain a subjective and objective comprehensive index of motion sickness.

[0012] Furthermore, in step (3), the cruise planning model takes the minimization of the subjective and objective comprehensive index of motion sickness and the travel time as the objective function, and uses the adaptive weighting factor set [ , ] and weighted method to control the weight of each target value in the objective function. The cruise planning model is as follows:

[0013] In the formula and are the motion sickness indicator weight and travel time weight, is the current state of the vehicle system, is the objective function value in the current state, for The vehicle system state at the initial moment, For the total planning duration, The cut-off time point for the planning domain The vehicle system status under is the state vector at the initial moment, is obtained based on the target position and speed The state vector at the end time, Indicates the vehicle pitch angle is related to the longitudinal acceleration of the vehicle The relevant parameters are is the current vehicle longitudinal acceleration The resulting vehicle pitch angle, Represented as the current vehicle longitudinal acceleration Vehicle driving risk indicators caused by , Represents the threshold of heterogeneous vehicle models; the safety constraint based on the vehicle driving risk index and the comfort constraint of the vehicle pitch angle are used to ensure the most basic safety criteria and comfort criteria for vehicle driving.

[0014] Furthermore, the vehicle driving risk index The risk heterogeneity of different models during longitudinal cruising is quantified, and the heterogeneous model thresholds of different models are identified by using the POT extreme value method to obtain the heterogeneous model thresholds. , Represents different vehicle models. When the current vehicle dynamically adjusts its driving action, the threshold of heterogeneous vehicle models is used to determine whether there is a potential collision risk, that is, to determine the vehicle driving risk index. and When the vehicle driving risk index When the vehicle driving risk index is greater than the threshold of heterogeneous vehicle models, there is no potential collision risk in the workshop. When it is less than the threshold of heterogeneous vehicle types, there is a potential risk of collision between vehicles. The headway between vehicles needs to be constrained to be outside the threshold of heterogeneous vehicle types to ensure that the basic principles of driving safety are met.

[0015] Furthermore, the comfort constraint of the vehicle pitch angle is formed based on the pitch and vertical dynamics, the vehicle pitch control range and the vehicle single track pitch dynamics model to form a comfort boundary constraint, and the vehicle pitch angle needs to be constrained within within the scope; The vehicle single-track pitch dynamics model is a vehicle single-track pitch dynamics model that includes suspension and sprung mass, in which the suspension is expressed equivalently by springs and dampers.

[0016] Furthermore, in step (3), solving the optimal anti-motion sickness cruise parameters of the cruise planning model by combining the dynamic programming method includes: using the dynamic programming method to find the optimal solution of the cruise planning model; The dynamic programming method discretizes the optimization problem in a distance segmented manner, uses the dynamic programming method to solve an optimal control solution in each distance domain, and applies the optimal control solution to the cruise control system; the dynamic programming method includes a loss function, an action network, and an evaluation network.

[0017] A computer device of the present invention comprises: a memory and a processor and a computer program stored in the memory. When the computer program is executed on the processor, the anti-motion sickness cruise planning method for an intelligent electric vehicle based on subjective and objective comprehensive indicators is implemented.

[0018] Compared with the prior art, the present invention has the following beneficial effects: First, we obtained evaluation data that met the physiological and psychological characteristics of Chinese drivers and passengers. We conducted multi-condition motion sickness tests, collected driving trajectory data, objective data on passenger physiological signals, and subjective feelings of motion sickness, identified the correlation mapping relationship between subjective and objective data, extracted the main feature quantities to reduce the dimension of the collected parameters, and based on this, we constructed a complete and streamlined subjective and objective data set for motion sickness, laying a data foundation for the subsequent proposal of subjective and objective evaluation indicators for motion sickness and the development of anti-motion sickness technology that meets the needs of the Chinese market.

[0019] Second, it meets the requirements of systematic, accurate, and subjective-objective consistency in motion sickness assessment. By introducing methods such as Parseval's theorem and state-space equations, the three main factors that affect human motion sickness in the time domain and frequency domain are fully considered. At the same time, by considering the subjective feelings of the human body, subjective feelings are added on the basis of objective indicators, and a comprehensive subjective and objective assessment method for motion sickness is proposed that integrates the amplitude of speed change impact, time domain continuity, frequency domain sensitivity, and human subjective feelings, further improving the motion sickness assessment method for drivers and passengers.

[0020] Third, ensure the optimal balance of safety, comfort and efficiency of the adaptive cruise control system under the influence of heterogeneous vehicle models. By identifying the sensitivity of drivers of heterogeneous vehicle models to driving risks in natural trajectory data, a safety constraint boundary that is more in line with the driver's behavior characteristics is proposed. At the same time, by considering different driving goals under conventional risk-free conditions, conflict conditions and dangerous conditions, a corresponding set of adaptive adjustment weighted factors is constructed, and combined with the two goals of motion sickness subjective and objective comprehensive indicators and travel time, it can effectively ensure the first criterion of the main goals under different conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 The present invention is a flowchart of an embodiment of an anti-motion sickness cruise planning method for an intelligent electric vehicle based on subjective and objective comprehensive indicators.

[0022] Figure 2 It is a schematic diagram of the slalom test working condition scenario of the embodiment.

[0023] Figure 3 It is a schematic diagram of the acceleration and deceleration alternating test scenario of the embodiment.

[0024] Figure 4 It is a graph of the sensitive frequency domain weighting function of the embodiment.

[0025] Figure 5 This is a comparison chart of acceleration before and after anti-motion sickness planning in the embodiment.

[0026] Figure 6 It is a comparison chart of comprehensive indicators of motion sickness before and after the anti-motion sickness planning of the embodiment.

[0027] Figure 7 This is a statistical chart of the comprehensive index value of motion sickness in the AD4CHE data set of the embodiment.

[0028] Figure 8 It is a comparison chart of the travel time before and after the anti-motion sickness planning of the embodiment. DETAILED DESCRIPTION

[0029] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0030] like Figure 1 As shown, the anti-motion sickness cruise planning method for an intelligent electric vehicle based on subjective and objective comprehensive indicators of this embodiment includes the following steps: (1) Through the multi-condition test of motion sickness real vehicle test, the vehicle motion trajectory data and the subjective and objective data of human physiological signals are collected to form the subjective and objective data set of motion sickness, which specifically includes the following steps: The first step is to design multi-condition test scenarios for motion sickness real-vehicle tests.

[0031] In order to explore the sensitive factors of motion sickness and propose subjective and objective comprehensive evaluation indicators, a real-car test for motion sickness was carried out, and the subjects' motion sickness was stimulated by visual and balance information. The real-car test was chosen mainly to avoid the test interference generally existing in driving simulators. Compared with the driving experience of real vehicles, driving simulators are insufficient in creating a sense of speed, which interferes with the input of balance. At the same time, in order to ensure the safety of the test, professional test drivers drove the test vehicles in a closed test area according to the preset working conditions. First, the test objectives were analyzed to determine the type of collected signals and the type of equipment required. The state of motion sickness is a stress response, so the subjects' heart rate, skin, eyes and other organs all show reaction behaviors. It was determined that the equipment used included gyroscopes, three-axis acceleration sensors, skin galvanometers, electrocardiographs and eye trackers, and the total number of subjects reached about 30. The positions of the subjects and equipment are as follows: the subjects sit on the rear passenger side; the gyroscope is installed on the rear passenger side; the three-way acceleration sensor is installed on the seat cushion and backrest of the seat where the subjects in the rear row are sitting; the subjects in the rear row have skin conductivity and electrocardiogram integrated testing equipment attached and carry eye movement equipment.

[0032] As a specific embodiment, the subject of this embodiment needs to rest in the car for 3 minutes before each test. After the single working condition test, the subject needs to rest for at least 30 minutes to restore the calm level. Each subject tests two working conditions, and each working condition is tested at least four times. The first time is the control group, which is used to unify the expectations of different subjects for the test and avoid obvious differences in subjective feelings caused by different expectations of different subjects for the test content. Working condition one is a slalom test, and working condition two is a straight-line acceleration and deceleration alternating test. The specific test scenarios of each working condition are as follows: Figure 2 and Figure 3As shown, the test was conducted on a straight two-way lane in a closed test area. In the working condition 1 slalom test, a cone barrel was placed every 20m on the straight test site, with a total of 8 cone barrels. The vehicle speed was kept at 30km / h when changing lines around the slalom, and each subject conducted 4 round-trip tests. In the working condition 2 straight-line acceleration and deceleration alternating test, the vehicle accelerated and decelerated rapidly from 0-30km / h-80km / h-30km / h-0, and each subject conducted 4 round-trip tests. During the test, objective data is collected by the acquisition equipment, and the acquisition frequency is the sampling frequency of the equipment. The subjective feelings are manually recorded by the subjects, and the recording frequency is once every 2 minutes, and it needs to be recorded again after completing a single test. If you cannot bear to get off the vehicle during the test, you must also make a record, and it is considered that the test is over.

[0033] The second step is to collect vehicle motion trajectory data and subjective and objective data of human physiological signals.

[0034] As an embodiment, subjective data is recorded using a pain scale widely used in motion sickness testing or treatment, as shown in Table 1. The scale measures the severity of motion sickness symptoms on an 11-point scale, where 0 represents no symptoms and 10 represents vomiting symptoms. The subjects score from 0 to 10 based on whether they have motion sickness symptoms and the severity of the motion sickness symptoms, so as to reflect the degree of tension and discomfort felt by the subjects during different actions taken by the test vehicle. In the pain scale, when the subjects rate the severity of their motion sickness symptoms as 2 to 5 points, the symptoms are not limited to those listed in the questionnaire, but any symptoms that are not listed but that the subjects consider to be caused by motion sickness based on their own experience in daily life are also included as the basis for scoring and the changes in the subjects' motion sickness levels.

[0035] Table 1 Pain level scale

[0036] In terms of collecting objective physiological indicators of respondents, a physiological multi-channel instrument was used to collect ECG and skin electrical signals. The above signals were measured by electrodes attached to the chest and the inside of the palm, and data such as heart rate and skin conductivity level can be obtained from these measurement signals. In order to collect the physiological activities of the respondents' eyes, an eye tracker was used to record eye movement information, including blinking frequency, pupil diameter, line of sight focus, eyelid opening and closing, and other data. In addition, a gyroscope and a three-way acceleration sensor were used to synchronously record vehicle motion parameters and posture data. Since the test equipment came from different manufacturers, the data was collected and recorded independently during the test, and the world time was used to synchronize the time between different data when analyzing the results. At the same time, the Z-score normalization method was used to map the data within a certain range, and the calculation formula is shown as follows:

[0037] In the formula is the original data, is the average value, is the standard deviation, is the standardized dimensionless value after calculation. Standardization solves the problem of large differences in the order of magnitude of each objective indicator and individual differences to ensure the reliability of the results. In addition, due to the inconsistent sampling frequency of the equipment and the excessive amount of data, in order to better obtain the objective data group corresponding to the subjective, the data is downsampled and segmented so that one set of data corresponds to one degree of motion sickness.

[0038] The third step is to construct a subjective and objective dataset of motion sickness based on the correlation of perceptions.

[0039] After completing all the test conditions, the number of obtained parameter items is as high as 18. To facilitate the subsequent work, the parameter items most relevant to human feelings are first extracted based on the task objectives and correlation. To facilitate the subsequent work, the parameter items most relevant to human feelings are first extracted based on the task objectives and correlation, so as to achieve effective data dimension reduction and form a streamlined subjective and objective data set for motion sickness. Motion sickness can be regarded as a stress response, and some physiological parameters of the subjects change significantly. By solving the correlation coefficient To determine the correlation between parameter items, the solution process is a two-by-two combination solution, that is, after selecting a parameter item, the correlation coefficient between the parameter item and each other parameter item is calculated. When calculating the correlation coefficient between the first parameter item and the second parameter item It is recorded as , when finding the correlation coefficient between the first parameter term and the third parameter term It is recorded as , and so on when solving each parameter item pairwise. and are the serial numbers of the two parameter items respectively. Correlation coefficient The formula is as follows:

[0040] in:

[0041] In the formula is the total number of samples, and is the standard deviation of the two parameters involved in calculating the correlation coefficient. and is the standardized dimensionless value of the two parameters involved in calculating the correlation coefficient. and is the average value of the two parameters involved in calculating the correlation coefficient. Based on this, the characteristic roots of the correlation coefficient matrix are calculated using the standard characteristic equation. , and define , that is, after solving, the characteristic roots are sorted in order of size, and the largest characteristic root is recorded as , and the second is recorded as , and so on, the smallest characteristic root is recorded as , 30 groups of data were randomly extracted to identify the relationship between the collected physiological parameters and vehicle motion parameters and the feeling of motion sickness, and the characteristic root proportion was used to obtain the feature quantity with the cumulative variance contribution rate of the common factor greater than 80%, so as to achieve the purpose of including most of the information of each parameter item and reducing the dimension to simplify the original collected data set.

[0042] Through hypothesis testing to test the significance of parameter changes, the indicators in the frequency domain analysis of heart rate variability and variability can be included in the subjective and objective data set of motion sickness. Among them, low-frequency power can represent the combined effect of the sympathetic nerve and the vagus nerve, high-frequency power can reflect the activity of the vagus nerve, and the total power represents the total variability of the signal. When the human body is in an uncomfortable state, measuring skin electricity is an important parameter in the field of physiological psychology and motion sickness diagnosis and treatment. Generally speaking, the skin conductance level can reflect the degree of awakening, alertness, mental stress, mental workload and other information of the human body. After testing, the skin conductance level of the subject is significant and can be included in the subjective and objective data set of motion sickness. In addition, the human eye is a visual perception organ, and various indicators of this organ are believed to be able to represent cognitive load, stress level, fatigue level, emotional state, attention and other information. Among them, the pupil diameter increases with the increase of task difficulty, psychological load, tension and so on. Therefore, the pupil diameter among the various indicators collected by the eye tracker is included in the subjective and objective data set of motion sickness. During vehicle driving, road stimulation and the operation of the driver or automatic driving system both affect ride comfort. Therefore, the rest of the motion sickness subjective and objective dataset consists of vehicle motion parameters. This example focuses on the anti-motion sickness effect of longitudinal cruising, so the vehicle motion parameters are selected as the vehicle longitudinal acceleration and pitch angle.

[0043] (2) Design a bandpass filter by combining the transfer function and the zero-order hold hypothesis, introduce the Parseval theorem to calculate the frequency-domain weighted motion sickness measurement value, establish the objective motion sickness index in the time-frequency domain based on the state-space equation, and propose a subjective and objective comprehensive motion sickness index by integrating the subjective and objective motion sickness data sets, including the following steps: (21) A bandpass filter is designed by combining the transfer function with the zero-order hold assumption. The vehicle acceleration in step (1) is converted into frequency domain form using Parseval’s theorem and then input into the bandpass filter to extract data within the human body sensitive frequency band.

[0044] The most sensitive frequency band for motion sickness in humans is 0.0315~0.25Hz. To minimize the stimulation level in this frequency band, a bandpass filter is used to extract the acceleration in this frequency band. The conventional bandpass filter is shown below:

[0045] In the formula is the bandpass filter system transfer function, is the system output value, Enter a value for the system. is a complex frequency variable, representing the frequency component of the signal, and is the time constant corresponding to the desired cutoff frequency of the bandpass filter. The bandpass filter is converted into a continuous time form and discretized. The discretization method is based on the zero-order hold assumption theorem, that is, the input acceleration remains constant within the discrete time step. The discretization equivalent expression is shown as follows:

[0046]

[0047] In the formula It is at the moment The discrete state vector of and Separately for the moment system input and output, is the discretized state transfer matrix, is the discretized input matrix, is the discretized output matrix, , and The expressions are: ;

[0048]

[0049] ;

[0050] In the formula is the identity matrix, the dynamic characteristic coefficient matrix and the discrete state coefficient matrix The expressions are: ;

[0051]

[0052] When the vehicle travels between planning points, due to the time step changes with the time point, so the matrix index Different at each time step. Based on the diagonalization method, the matrix exponent is simplified and the matrix is ​​changed at each time step. Perform diagonalization:

[0053] In the formula is a matrix The diagonal matrix of It is used to matrix The eigenvector matrix is ​​diagonalized, and the matrix index is equal to:

[0054] The diagonal matrix exponential is expressed as:

[0055] In the formula and is a diagonal matrix Diagonal elements, and yes and The exponential operation indicates that the time step The dynamic evolution under . Since the eigenvector matrix yes The invariant of , so the eigenvector matrix can be pre-calculated according to the actual motion parameters of the vehicle The predefined method can reduce the computational complexity from matrix exponential calculation to calculation based on diagonalized matrix, which can effectively improve the computational efficiency.

[0056] (22) Use Parseval's theorem to calculate the frequency-domain weighted motion sickness measure.

[0057] Accelerations in different frequency bands have different effects on passenger comfort, that is, acceleration shocks in certain frequency bands are more likely to induce motion sickness in passengers. Therefore, on the basis of ensuring vehicle maneuverability, safety, and stability, acceleration is reshaped in the frequency domain to minimize the stimulation to comfort and the incidence of motion sickness. The frequency domain form of acceleration is defined using Parseval's theorem:

[0058] In the formula is the planning domain cut-off time point, for The longitudinal acceleration of the vehicle at time The Fourier transform of for The complex conjugate of is an imaginary unit, is the frequency variable in Fourier transform. Input the bandpass filter to extract the data in the human body sensitive frequency band, and add the weighting function to the extracted data in the human body sensitive frequency band according to the ISO2631 standard , the weighted function curve is as follows Figure 4 As shown, the frequency domain weighted acceleration is obtained As shown below:

[0059] Then, the frequency domain weighted acceleration Integrate to obtain the frequency-domain weighted motion sickness measure , as shown below:

[0060] In the formula for The complex conjugate of .

[0061] (23) Based on the state-space equation, an objective motion sickness index in the time-frequency domain is proposed.

[0062] In addition to the sensitive frequency factor, the international standard ISO2631 points out that vibration duration also has an important impact on human motion sickness. Therefore, in order to more comprehensively evaluate the vehicle motion comfort and reduce the motion sickness induced by vehicle speed change behavior, the frequency domain weighted acceleration , combined with The time domain index is constructed by the method to consider the influence of vibration persistence. In the calculation process, the frequency domain weighted acceleration Take the conjugate complex number, negate the imaginary part and perform a secondary transform. The result of the transformation is a complex number, but the imaginary part of this complex number is small, so we only need to take the real part of the complex number to get the time-domain weighted acceleration. , and then calculate the time-domain weighted motion sickness measurement value by the following formula :

[0063] In the formula is the time-domain weighted motion sickness measurement value, is the planning domain cutoff time point, i.e., the total planning duration. In addition, in order to maximize the stability of the input, two additional time-related random perturbations are added: and The disturbance effect ,Right now:

[0064] and The combination constitutes the evaluation index in the time domain, further improving the comfort evaluation method. On this basis, the state space equation is introduced to integrate the disturbance effect , time-domain weighted motion sickness measurement value and frequency-domain weighted motion sickness measure , forming the objective motion sickness index TFI (Time & Frequency Index) in the time and frequency domain under the vehicle speed change motion, as shown in the following formula:

[0065] in:

[0066] In the formula - is the weight coefficient of each index component. In the actual planning process, the acceleration sensitivity frequency and continuity have the greatest impact on comfort, so : : The weight ratio between them is 0.4:0.4:0.2.

[0067] (24) By integrating human perception, a comprehensive index of subjective and objective motion sickness is proposed.

[0068] Based on the motion sickness test in a real car, multiple physiological signal acquisition data and subjective feeling quantification data of the occupants were obtained, and thus a subjective and objective data set that directly describes the human feeling was formed. The physiological signal acquisition data includes electrocardiogram (ECG), electrodermal activity (EDA) and pupil diameter (PD). The subjective feeling quantification data includes multiple recording moments recorded by the occupants and the human feeling corresponding to each recording moment (the subjective feeling quantification data in this example is a score value selected according to the pain degree scale). The multiple physiological signal acquisition data are time-synchronized and the data of each physiological signal is segmented. The physiological signal acquisition segmented data is labeled according to the subjective data to extract the human feeling and quantification range in normal or motion sickness state. The data of each state of the occupant is matched with the reference range to obtain the first parameter deviation. ( ) is used to characterize whether the physiological signal deviates from the reference range. At the same time, the deviation of the second parameter is calculated based on the deviation between the state data and the reference range. It is used to characterize the degree of deviation of physiological signals. According to the decreasing principle, the weight of each physiological signal collection data is assigned as follows: > > , so the first parameter deviates from the total and the second parameter deviation total The calculation formula is shown in formula (17):

[0069] In the formula , , respectively represent the first parameter deviations of the electrocardiogram signal ECG, the electrodermal signal EDA and the pupil diameter PD, , , They represent the second parameter deviations of the electrocardiogram signal ECG, the electrodermal signal EDA and the pupil diameter PD respectively, where:

[0070]

[0071] In the formula are the actually measured ECG, EDA and PD, is the reference range of the state data. Then, the total deviation of the first parameter and the total deviation of the second parameter are matched with the objective motion sickness index in the time-frequency domain under the vehicle speed change motion to obtain the Motion Sickness Subjective-Objective Index, ), the calculation formula is shown in formula (20):

[0072] In the formula is the objective motion sickness index in the time-frequency domain. In formula (20), the parameter It has a discriminating function. When the human body's various conditions are within the reference range, is 0, that is When it is 0, the occupant is in normal condition and does not feel motion sickness. When it is 1, the occupant's state becomes abnormal, and as the calculated value of the second item increases, the occupant's motion sickness becomes more and more intense.

[0073] (3) Safety constraints are defined based on the heterogeneous sensitivity of the driver, and comfort constraints are defined through the pitch dynamic characteristics of the vehicle body to construct an adaptive weighted factor set and a cruise planning model. The optimal anti-motion sickness cruise parameters are solved by the combined dynamic programming method, including the following steps: (31) Define safety constraints based on the heterogeneous sensitivity of drivers.

[0074] For following behavior, the acceleration of the front vehicle is the disturbance signal, and the driver controls the longitudinal acceleration of the vehicle by considering the disturbance , achieving smooth and safe following. By statistically analyzing natural driving data, it is found that when facing the dynamic adjustment of the speed of the vehicle in front, the drivers of different models have different reactions, that is, there is a heterogeneous phenomenon in the sensitivity to driving risks. Introducing risk indicators MTTC Modified Time to Collision quantifies the risk heterogeneity of different vehicle types. MTTC Indicators make up for the traditional TTC The disadvantages of the indicator can be considered by considering the relative speed and acceleration of the front and rear vehicles. It is widely used in traffic conflict research. Its calculation formula is shown in (21):

[0075] In the formula are the roots of the quadratic equation, , , for:

[0076] In the formula 、 、 、 are the length, position, speed and acceleration of the front vehicle respectively. 、 、 are the position, velocity and acceleration of the vehicle respectively.

[0077]

[0078] Risk indicator The value of is equal to the smallest positive root. By additionally considering the vehicle acceleration and velocity, Can be In addition to the collision scenarios covered, more working conditions can be explained, such as the front car is faster than the rear car, the rear car has braking behavior, etc. The POT extreme value method is used to identify the heterogeneous thresholds of different models, and the obtained model heterogeneous thresholds are recorded as , Represents different types of vehicles, including cars and trucks. When the front vehicle dynamically adjusts its driving action, the heterogeneous risk threshold is used to determine whether there is a potential collision risk, that is, to determine and The safety constraint boundary is defined based on the heterogeneous sensitivity of the driver:

[0079] when When it is greater than the threshold, there is no potential collision risk in the workshop. When it is less than the vehicle type heterogeneity threshold, there is a potential risk of collision between vehicles. Therefore, the headway between vehicles needs to be constrained to be outside the vehicle type heterogeneity threshold to ensure that the basic principle of driving safety is met.

[0080] (32) Comfort constraints are formed based on the pitch dynamic characteristics of the vehicle body.

[0081] The vehicle body pitch angle and pitch rate are mainly affected by the longitudinal acceleration, so a vehicle single track pitch dynamics model including suspension and sprung mass is constructed. In the vehicle single track pitch dynamics model, the suspension uses springs and dampers. According to the pitch and vertical dynamics, the calculation formula of the vertical displacement and rate of the front suspension is shown in formula (25). Since the front and rear suspensions have opposite symmetry, in order to keep the content concise, only the calculation formula of the variables related to the front suspension is shown here. The formula related to the rear suspension is similar and will not be repeated.

[0082]

[0083]

[0084] In the formula is the vertical displacement of the front suspension, is the distance from the front axle to the vehicle's center of mass, and are the vehicle body pitch angle and angular velocity, and are the vertical displacement and velocity of the vehicle body, Its angular velocity It can be expressed as:

[0085]

[0086] In the formula It is the horizontal distance between the front suspension and the rotation center. The vehicle body pitch angle during normal driving and braking The value is small, and its influence on the horizontal displacement of the front and rear suspension can be ignored. Then the torque of the front and rear suspension acting on the vehicle body is and for:

[0087]

[0088] In the formula is the distance from the rear axle to the vehicle's center of mass, , are the rear suspension rotation angle and angular velocity, is the acceleration due to gravity, , are the front suspension rotation angle and angular velocity, is the sprung mass, and are the angular stiffness of the front and rear suspensions, and They are the rotational damping of the front and rear suspensions, the body pitch angular acceleration It can be expressed as:

[0089] In the formula is the height of the vehicle's center of mass, The mass of the spring is around The moment of inertia of the shaft rotation, is the vehicle curb weight, represents the longitudinal acceleration of the vehicle. To achieve good ride comfort, the vehicle pitch angle should be controlled within the range of ±1°. Therefore, based on the vehicle pitch control range and the vehicle dynamics model, a comfort boundary constraint is set, namely:

[0090] In the formula Indicates the vehicle pitch angle is related to the longitudinal acceleration of the vehicle The relevant parameters are For the current The resulting vehicle pitch angle.

[0091] (33) Construct an adaptive weighting factor set and cruise planning model.

[0092] The core of the anti-motion sickness planning strategy is to determine the appropriate vehicle control input so that the vehicle can control the subjective and objective comprehensive indicators of motion sickness during driving. and travel time By setting the objective function, a set of equality and inequality constraints, the anti-motion sickness planning problem is solved as a multi-objective optimal control problem, and the subjective and objective comprehensive indicators of motion sickness are minimized. and travel time Together they form the overall objective function. , ] and weighted method to control the target values ​​in the overall objective function The weight in is as shown in formula (30):

[0093] In the formula and are the motion sickness indicator weight and travel time weight, defined as:

[0094]

[0095] The essence of the adaptive weighting factor set is: when the vehicle is driving, , that is, when the vehicle is within the safety boundary without potential collision risk, the vehicle takes comfort as the main goal, and the model outputs safe and comfortable motion planning; but when Decreases and gradually approaches the vehicle model heterogeneity threshold , the driving risk under such conditions becomes more urgent. In order to avoid accidents, the weight of motion sickness indicators should be reduced as soon as possible, and the focus should no longer be on ride comfort, but safety should be the main criterion. By applying the weighted factor set, the proposed cruise planning model can adaptively adjust the driving criterion target according to whether there is a collision risk in the current working condition, which greatly improves the anthropomorphism of the cruise strategy. Finally, the cruise planning model is established based on the above objective function and constraints, as shown below:

[0096] Where st is the constraint, is the current state of the vehicle system, is the overall objective function value under the current state, is the vehicle system state at the initial time t=0, is the planning domain cut-off time point, i.e. the total planning duration. For t= The vehicle system status at the planning end time, is the state vector at the initial time t=0, is t= obtained based on the target position and speed The state vector at the end time. The equality constraints include the initial point constraints and the end point constraints. Indicates the vehicle pitch angle is related to the longitudinal acceleration of the vehicle The relevant parameters are The longitudinal acceleration of the current vehicle The resulting vehicle pitch angle, Similarly, it is specifically expressed as the current vehicle longitudinal acceleration Vehicle driving risk indicators caused by MTTC The safety constraints based on the vehicle driving risk index and the comfort constraints based on the vehicle pitch angle can ensure the most basic safety criteria and comfort criteria for vehicle driving.

[0097] (34) Combined dynamic programming method is used to solve the optimal anti-motion sickness cruise parameters.

[0098] In order to obtain the desired motion parameters of anti-motion sickness cruise in real time, the dynamic programming method is used to find the optimal solution for the cruise planning model. The basic idea of ​​this method is to discretize the optimization problem in a distance segmented manner, use the dynamic programming method to solve an optimal control solution in each distance domain, and apply the control solution to the cruise control system. The dynamic programming method includes a loss function, an action network, and an evaluation network. Here, the loss function is the objective function in the cruise planning model. The role of the action network is to calculate the output optimal control solution by inputting the vehicle state, that is, the desired motion parameters that can be achieved based on the current vehicle state, and it is a control quantity that can be directly used in the vehicle control system. The structure of the action network is expressed as:

[0099] In the formula For the The vehicle driving status input during step planning, , (·) is the activation function, For the The weight matrix of the action network during step planning. However, the action network cannot output the optimal control solution after one iteration, so it is necessary to calculate the current output And the numerical optimal solution The error between , and update the action network weight matrix by minimizing the error , which makes the action network output close to the optimal solution. Error The calculation formula is as follows:

[0100] Update the action network weight matrix by minimizing the error The iteration formula is:

[0101] In the formula is the action network learning rate, is the gradient scaling normalization factor. The numerical optimal solution used when calculating the error It is calculated using the evaluation network and the loss function. By applying numerical optimization theory in the evaluation network to minimize the loss function, a numerical optimal solution can be obtained. However, this optimal solution only meets mathematical constraints, and there may be situations where it cannot satisfy the vehicle control system or the optimal value cannot be achieved through the current vehicle state. Therefore, the numerical optimal solution is applied to the action network so that the action network outputs the optimal control solution that is closest to the optimal solution and can be used for the control system based on the current vehicle state. In addition, the updated action network calculation weight matrix can be inherited on each distance domain. This method is more suitable for the real world because it reduces the complexity of the optimization problem to be solved in each iteration, and there is no need to update the weight matrix every time, which further improves real-time performance.

[0102] To verify the effectiveness of the proposed anti-motion sickness driving strategy, vehicles with drastic speed changes were extracted from the AD4CHE dataset. Vehicle ID1407 experienced rapid deceleration and acceleration due to the front vehicle cutting in, resulting in a high comprehensive index value of motion sickness during the movement, which means that the comfort was seriously insufficient. The proposed anti-motion sickness planning method was used to replan its movement. The results are shown in the figure below. Figure 5 and Figure 6 As shown in the figure, it can be seen that the acceleration curve after planning is smoother, the subjective and objective comprehensive index value of motion sickness is reduced by 58.92%, and the comfort is effectively improved. Figure 7 The figure shows the statistical description of the subjective and objective comprehensive index values ​​of motion sickness in the AD4CHE data set. It can be found that the vehicles with the subjective and objective comprehensive index values ​​of motion sickness in the interval [0.5, 0.6] are the most, that is, this interval is the range where the human body feels more comfortable. The proposed anti-motion sickness planning method reduces the subjective and objective comprehensive index value of motion sickness to 0.56, which is within the comfort range, verifying the effectiveness of the proposed anti-motion sickness cruise planning method. The travel time before and after the anti-motion sickness planning is compared within a fixed driving distance (190m). The results are shown in Figure 8 As shown in the figure, the travel time before and after planning is 27.08s and 26.92s respectively, with a difference of only 0.6%. This shows that the subjective and objective comprehensive index values ​​of motion sickness are reduced before and after planning, the comfort is improved, and the travel time is not affected, which verifies the timeliness of the proposed anti-motion sickness cruise planning method.

[0103] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well.

Claims

1. An anti-motion sickness cruise planning method for intelligent electric vehicles based on subjective and objective comprehensive indicators, characterized in that: The following steps are involved: (1) Collect vehicle motion trajectory data and subjective and objective data of human physiological signals; (2) The collected vehicle motion trajectory data is used to calculate the motion sickness index in the frequency domain weighted and the evaluation index in the time domain through Parseval theorem and bandpass filter. The motion sickness index in the frequency domain weighted and the evaluation index in the time domain are fused using the state space equation to form the objective motion sickness index in the time and frequency domain under the vehicle speed change motion. The subjective and objective comprehensive index of motion sickness is obtained by combining the subjective and objective data of human physiological signals. (3) A cruise planning model is constructed using the obtained subjective and objective comprehensive indicators of motion sickness and a set of adaptive weighted factors. The optimal anti-motion sickness cruise parameters of the cruise planning model are solved using a combined dynamic programming method to minimize the subjective and objective comprehensive indicators of motion sickness and the travel time during the vehicle's driving process.

2. The method for anti-motion sickness cruise planning of intelligent electric vehicles based on subjective and objective comprehensive indicators according to claim 1 is characterized in that: In step (2), the calculation of the motion sickness index under frequency domain weighting by using Parseval theorem and bandpass filter includes: using Parseval theorem to convert the vehicle acceleration of step (1) into frequency domain form, inputting the frequency domain form into the bandpass filter to extract data within the human body sensitive frequency band, and simultaneously adding the weighting function specified in standard ISO2631 to the extracted data within the human body sensitive frequency band, thereby obtaining frequency domain weighted acceleration, and then integrating the frequency domain weighted acceleration to obtain the frequency domain weighted motion sickness measurement value.

3. The anti-motion sickness cruise planning method for intelligent electric vehicles based on subjective and objective comprehensive indicators according to claim 2 is characterized in that: In step (2), the evaluation index in the time domain is calculated as follows: Based on the frequency domain weighted acceleration, the time domain weighted motion sickness measurement value is constructed by combining the inverse Fourier transform to consider the influence of vibration persistence. After the transformation, the real part of the transformation result is taken to obtain the time-domain weighted acceleration, and then the time-domain weighted motion sickness measurement value is calculated using the time-domain weighted acceleration; Two additional time-related random disturbances are added to produce disturbance effects. The time-domain weighted motion sickness measurement value and the disturbance effect are combined to form the evaluation index in the time domain.

4. The method for anti-motion sickness cruise planning of an intelligent electric vehicle based on subjective and objective comprehensive indicators according to claim 2 is characterized in that: In step (2), the bandpass filter is converted into a continuous time form and discretized. The input of the bandpass filter is a constant value within the discrete time step. When the vehicle travels between the planning points, the matrix index in the discretized equivalent expression is different at each time step. The matrix index is transformed based on the diagonalization method, and the dynamic characteristic coefficient matrix is ​​diagonalized through the eigenvector matrix. The dynamic characteristic coefficient matrix is ​​a time constant corresponding to the expected cutoff frequency of the bandpass filter. The eigenvector matrix used in the diagonalization process is an invariant of the time step. Before planning, the value of the eigenvector matrix is ​​predefined according to the actual motion parameters of the vehicle.

5. The method for anti-motion sickness cruise planning of intelligent electric vehicles based on subjective and objective comprehensive indicators according to claim 1 is characterized in that: In step (2), the step of combining the subjective and objective data of human physiological signals to obtain the subjective and objective comprehensive index of motion sickness specifically includes: performing time synchronization processing on the subjective and objective data of human physiological signals, and labeling the physiological signal acquisition data according to the subjective feeling quantification data; matching the physiological signal data of each occupant with the reference range, obtaining the first parameter deviation of each signal, which is used to characterize whether each physiological signal deviates from the reference range, and calculating the second parameter deviation based on the deviation between the data of each physiological signal and the reference range, which is used to characterize the degree of deviation of each physiological signal, and assigning weights to each physiological signal acquisition data according to the decreasing principle, and then matching the total deviation of the first parameter and the total deviation of the second parameter with the objective motion sickness index in the time-frequency domain under the vehicle speed change motion, to obtain the subjective and objective comprehensive index of motion sickness.

6. The method for anti-motion sickness cruise planning of intelligent electric vehicles based on subjective and objective comprehensive indicators according to claim 1 is characterized in that: In step (3), the cruise planning model takes the minimization of the subjective and objective comprehensive index of motion sickness and the travel time as the objective function, and uses the adaptive weighting factor set [ , ] and weighted method to control the weight of each target value in the objective function. The cruise planning model is as follows: In the formula and are the motion sickness indicator weight and travel time weight, is the current state of the vehicle system, is the objective function value in the current state, for The vehicle system state at the initial moment, For the total planning duration, The cut-off time point for the planning domain The vehicle system status under is the state vector at the initial moment, is obtained based on the target position and speed The state vector at the end time, Indicates the vehicle pitch angle is related to the longitudinal acceleration of the vehicle The relevant parameters are is the current vehicle longitudinal acceleration The resulting vehicle pitch angle, Represented as the current vehicle longitudinal acceleration Vehicle driving risk indicators caused by , Represents the threshold of heterogeneous vehicle models; the safety constraint based on the vehicle driving risk index and the comfort constraint of the vehicle pitch angle are used to ensure the most basic safety criteria and comfort criteria for vehicle driving.

7. The method for anti-motion sickness cruise planning of an intelligent electric vehicle based on subjective and objective comprehensive indicators according to claim 6 is characterized in that: The vehicle driving risk index The risk heterogeneity of different models during longitudinal cruising is quantified, and the heterogeneous model thresholds of different models are identified by using the POT extreme value method to obtain the heterogeneous model thresholds. , Represents different vehicle models. When the current vehicle dynamically adjusts its driving action, the threshold of heterogeneous vehicle models is used to determine whether there is a potential collision risk, that is, to determine the vehicle driving risk index. and When the vehicle driving risk index When the vehicle driving risk index is greater than the threshold of heterogeneous vehicle models, there is no potential collision risk in the workshop. When it is less than the threshold of heterogeneous vehicle types, there is a potential risk of collision between vehicles. The headway between vehicles needs to be constrained to be outside the threshold of heterogeneous vehicle types to ensure that the basic principles of driving safety are met.

8. The method for anti-motion sickness cruise planning of an intelligent electric vehicle based on subjective and objective comprehensive indicators according to claim 6 is characterized in that: The comfort constraint of the vehicle pitch angle is formed based on the pitch and vertical dynamics, the vehicle pitch control range and the vehicle single track pitch dynamics model to form a comfort boundary constraint, and the vehicle pitch angle needs to be constrained within within the scope; The vehicle single-track pitch dynamics model is a vehicle single-track pitch dynamics model that includes suspension and sprung mass, in which the suspension is expressed equivalently by springs and dampers.

9. The method for anti-motion sickness cruise planning of intelligent electric vehicles based on subjective and objective comprehensive indicators according to claim 1, characterized in that: In step (3), solving the optimal anti-motion sickness cruise parameters of the cruise planning model by the combined dynamic programming method includes: using the dynamic programming method to find the optimal solution of the cruise planning model; The dynamic programming method discretizes the optimization problem in a distance segmented manner, uses the dynamic programming method to solve an optimal control solution in each distance domain, and applies the optimal control solution to the cruise control system; the dynamic programming method includes a loss function, an action network, and an evaluation network.

10. A computer device, characterized in that: include: A memory and a processor and a computer program stored in the memory, when the computer program is executed on the processor, implements the anti-motion sickness cruise planning method for an intelligent electric vehicle based on subjective and objective comprehensive indicators as described in any one of claims 1 to 9.

Images