A vehicle state estimation method based on parameter identification

Abstract

The present invention discloses a vehicle state estimation method based on parameter identification, including: establishing an unsprung mass parameter observer; establishing a vehicle state observer; initializing the parameter observer and the state observer; calculating the suspension resultant force H at time k k ; calculating the parameter predicted value #imgabs0# and the parameter prediction error covariance #imgabs1# at time k, calculating the parameter estimated value #imgabs2# and the parameter estimation error covariance #imgabs3# at time k, updating the vehicle state observer at time k; calculating the state predicted value #imgabs4# and the state prediction error covariance #imgabs5# at time k, calculating the state estimated value #imgabs6# and the state estimation error covariance #imgabs7# at time k; outputting the unsprung mass estimated value and the vehicle state estimation result at time k; repeating the above steps to obtain the unsprung mass estimated value and the vehicle state estimation result at the next moment. The present invention can estimate and dynamically correct and adjust the parameter value of the unsprung mass in real time, effectively reduce the model error of the state observer caused by parameter errors, and further improve the estimation accuracy of the vehicle operating state.

Description

Technical Field

[0001] The present invention relates to the field of vehicle parameter identification and vehicle state estimation, and particularly relates to a method for estimating the state of a semi-active suspension vehicle based on parameter identification. Background Art

[0002] The control force of a semi-active suspension system is mostly calculated by feedback based on vehicle operating state information. Therefore, it is required to obtain as much vehicle operating state information as possible to achieve better control performance. However, in engineering applications, not all state variables can be directly and accurately measured by on-vehicle sensors. On the one hand, it is difficult for sensors to directly measure some state variables. On the other hand, problems such as calibration errors, zero drift, and measurement noise existing in the sensors themselves will cause the measurement accuracy to decrease as the number of sensors increases. In addition, the more sensors used, the higher the cost of the semi-active suspension system. These factors will seriously restrict the full play of the control effect of the semi-active suspension system. Therefore, it is of great value to estimate the complete and accurate operating state of a semi-active suspension vehicle through known system information and observable signals.

[0003] Currently, the common method for obtaining the state of a semi-active suspension vehicle is to design a state observation algorithm based on a vehicle dynamics reference model. According to the known system input quantity and the vehicle state quantity that is easy to measure by sensors, other unknown state quantities are predicted, and then the predicted state quantities are corrected by introducing the deviation feedback of the measurement variable. Finally, a relatively reliable estimated value of the state variable is obtained. However, this state observation algorithm is greatly affected by the reference model that describes the vehicle motion state. Due to changes in working conditions such as the number of occupants or the vehicle being empty or fully loaded, the sprung mass of the vehicle will change, making the model error between the reference model and the actual model larger, and then resulting in inaccurate estimated values of the observed state variables. Therefore, it has great engineering significance to identify the real-time parameters of the sprung mass and accurately estimate the complete vehicle operating state.

[0004] In the "Method for Observing and Controlling Vehicle Suspension Dynamics" disclosed in the Chinese invention patent CN115431696A by Guan Jifu et al., only the suspension dynamic stroke signal is used to estimate the state quantity of the suspension system and the road surface elevation. However, in this method, the process noise that characterizes the model error only considers the influence of road surface unevenness excitation, and does not involve the dynamic model error between the reference model used for estimation and the actual motion process due to subjective or objective factors, especially the error of the vehicle state observer prediction result caused by the change in vehicle load, i.e., the sprung mass change, and the estimation result is not accurate. Summary of the Invention

[0005] To solve the problems existing in the prior art, the present invention provides a vehicle state estimation method based on parameter identification. The method first performs parameter identification on the sprung mass of a semi-active suspension vehicle, thereby updating the state equation of the vehicle state observer. Then, based on the updated vehicle state observer and according to the signal of the suspension height sensor, the running state of the complete vehicle can be estimated.

[0006] To achieve the object of the present invention, a vehicle state estimation method based on parameter identification provided by the present invention is implemented according to the following steps:

[0007] (1) Establish an observer for sprung mass parameters:

[0008] During the driving process of the vehicle, the sprung mass is almost constant. Therefore, it can be considered that the sprung mass at the previous moment is equal to the sprung mass at the next moment, that is:

[0009] m k+1 =m k (14)

[0010] where the subscripts k and k + 1 represent moments; m is the sprung mass.

[0011] Select state variables Measurement variables System input u = [u] T , and the following observer for sprung mass parameters can be established:

[0012]

[0013] where the subscripts k and k + 1 represent moments; m s is the sprung mass, is the vertical acceleration of the sprung mass, u is the adjustable power of the shock absorber, H is the suspension resultant force including the shock absorber damping force and the spring force; w is the process noise, v is the measurement noise, and both satisfy the Gaussian distributions of w ∼ N(0, Q w ), v ∼ N(0, R v ), where Q w and R v represent the covariance matrices satisfied by the process noise and the measurement noise respectively. In addition, the noise w and v are independent of each other and independent of the state variables.

[0014] (2) Establish a vehicle state observer:

[0015] Select state variables Measurement variable y = [x s -x u T , system input u = [u] T , and the following vehicle state observer can be established: ​

[0016]

[0017] Among them, A is the system state matrix, B is the input matrix, and C is the output state matrix; ω is the process noise, and υ is the measurement noise. The two respectively satisfy ω ~ N(0, Q ω ), υ ~ N(0, R υ ), where Q ω and R υ respectively represent the covariance matrices satisfied by the process noise and the measurement noise. In addition, the noises ω and υ are independent of each other and independent of the state variables.

[0018] Considering that components such as sensors and ECUs in engineering practical applications operate at the sampling time T, therefore, the vehicle state observer is discretized:

[0019]

[0020] Thus, the discretized form of the vehicle state observer is obtained:

[0021]

[0022] Among them, the subscripts k and k + 1 represent moments, and the subscript d represents the discretized form; A d , B d , C d are matrices in the discretized form.

[0023] (3) Initialize the parameter observer and the state observer:

[0024] Set the model parameters of the sprung mass parameter observer, including setting the initial sprung mass parameter value θ0 and the initial parameter error covariance matrix According to the parameter model error and the measurement error of the sprung mass acceleration sensor, set the covariance matrices Q w and R v satisfied by the process noise w and the measurement noise v; set the sampling time T θ of the sprung mass parameter observer.

[0025] Set the model parameters of the vehicle state observer, including setting the initial vehicle state x0 and the initial state error covariance matrix According to the state model error and the measurement error of the suspension height sensor, set the covariance matrices Q ω and R υ satisfied by the process noise ω and the measurement noise υ; set the sampling time T x of the vehicle state observer.

[0026] Among them, the covariance matrix Q satisfied by the process noise w and ω representing the model errorw and Q ω is determined by the method of offline adjusting the value to optimize the estimation result, and the covariance matrix R satisfied by the measurement noises v and υ characterizing the sensor measurement error v and R υ is determined by the method of noise analysis of the sensor measurement signal.

[0027] (4) Calculate the suspension resultant force H at time k k :

[0028] The dynamic deflection x of the suspension at time k is measured by the suspension height sensor s -x u , and its signal is discretely differentiated to obtain the dynamic velocity of the suspension at time k In addition, the adjustable actuating force u of the shock absorber at time k is obtained from the damping continuously adjustable shock absorber control unit k . Thus, the suspension resultant force H at time k is calculated k :

[0029]

[0030] (5) Calculate the parameter predicted value at time k and the parameter prediction error covariance

[0031] According to the prediction equation of the sprung mass parameter observer, the parameter predicted value at time k is calculated from the parameter estimated value at time k - 1

[0032]

[0033] where the superscript hat represents the predicted value and the superscript tilde represents the estimated value.

[0034] Calculate the parameter prediction error covariance at time k The solution calculation formula is as follows:

[0035]

[0036] where the superscript hat represents the prediction error covariance and the superscript tilde represents the estimation error covariance.

[0037] (6) Calculate the parameter estimated value at time k and the parameter estimation error covariance

[0038] According to the deviation between the measurement value of the sprung mass acceleration sensor and the output predicted value of the parameter observer, the parameter predicted value at time k is feedback corrected, and the solution calculation formula is as follows:​

[0039]

[0040] Calculate the parameter estimation error covariance at time k The solution formula is as follows:

[0041]

[0042] In equations (6) and (7):

[0043]

[0044] Where, is the parameter deviation gain at time k, is the measured value of the sprung mass acceleration sensor at time k, and I represents the identity matrix.

[0045] The parameter estimation error covariance obtained from equation (7) is used for the calculation of the parameter prediction error covariance at the next time .

[0046] (7) Update the vehicle state observer at time k:

[0047] Take the reciprocal of the parameter estimated value to obtain the estimated value of the sprung mass at time k Substitute the estimated value of the sprung mass at time k into equation (2), update the vehicle state observer, and discretize the vehicle state observer according to the sampling time T x of the state observer.

[0048] (8) Calculate the state predicted value at time k and the state prediction error covariance

[0049] The adjustable actuator force u at time k - 1 is obtained from the continuously variable damping shock absorber control unit k-1 , and according to the prediction equation of the vehicle state observer, the state predicted value at time k is calculated from the state estimated value at time k - 1 The solution formula is as follows:

[0050]

[0051] Where, the hat on the superscript represents the predicted value, and the tilde on the superscript represents the estimated value.

[0052]

[0053] Among them, the superscript broken line represents the prediction error covariance, and the superscript wavy line represents the estimation error covariance.

[0054] (9) Calculate the state estimate at time k and the state estimation error covariance

[0055] According to the deviation between the measured value of the suspension height sensor and the output prediction value of the state observer, the state prediction value at time k is Feedback correction is performed and the calculation formula is solved as follows:

[0056]

[0057] Calculate the state estimation error covariance at time k The solution calculation formula is as follows:

[0058]

[0059] In formula (11) and formula (12):

[0060]

[0061] in, is the state deviation gain at time k, is the measured value of the suspension height sensor at time k, and I represents the unit matrix.

[0062] According to formula (12), the state estimation error covariance is The state prediction error covariance for the next moment Calculation.

[0063] (10) Output the sprung mass estimate and vehicle state estimate at time k;

[0064] (11) Let k = k + 1, and repeat the above steps (4) to (10) to obtain the sprung mass estimation value and vehicle state estimation result at the next moment.

[0065] Compared with the prior art, the present invention has at least the following advantages:

[0066] 1) Parameter identification of the sprung mass through the sprung mass acceleration sensor and the suspension height sensor can reduce the parameter error caused by changes in operating conditions such as the number of occupants or whether the vehicle is empty or fully loaded, and provide reliable parameter results for subsequent state estimation and suspension dynamic control.

[0067] 2) The sprung mass obtained based on parameter identification can be used to update the vehicle state observer in real time, which can reduce the model error of the state observer caused by parameter error and make the estimated vehicle operating state more accurate.

[0068] 3) In vehicle state estimation, only the measurement signals of the suspension height sensors are used, which can reduce the number of sensors used while ensuring the state estimation accuracy, lower the cost of the semi-active suspension control system, give full play to the control effect of the semi-active suspension, and further achieve better control performance.

[0069] 4) In the estimation method provided by the present invention, the process noise w and ω that characterize the model error satisfy the covariance matrix Q w and Q ω , and the measurement noise v and υ that characterize the sensor measurement error satisfy the covariance matrix R v and R υ , which can improve the accuracy of the estimation result. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1 is a schematic flowchart of a vehicle state estimation method based on parameter identification provided by an embodiment of the present invention.

[0071] Figure 2 is a schematic diagram of a quarter semi-active suspension vehicle model in an embodiment of the present invention.

[0072] Figure 3 is a schematic diagram of the system framework in an embodiment of the present invention.

[0073] Figure 4 is a comparison chart of the estimation result, actual result, and original estimation result of the sprung mass in an embodiment of the present invention.

[0074] Figure 5 is a comparison chart of the estimation result, actual result, and original estimation result of the suspension dynamic deflection in an embodiment of the present invention.

[0075] Figure 6 is a comparison chart of the estimation result, actual result, and original estimation result of the sprung mass velocity in an embodiment of the present invention.

[0076] Figure 7 is a comparison chart of the estimation result, actual result, and original estimation result of the tire dynamic deformation in an embodiment of the present invention.

[0077] Figure 8 is a comparison chart of the estimation result, actual result, and original estimation result of the unsprung mass velocity in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0078] The present invention will be further described in detail below with reference to the accompanying drawings and by way of examples.

[0079] Please refer to Figure 1, a vehicle state estimation method based on parameter identification provided by the present invention is implemented as follows:

[0080] (1) Establish an observer for the sprung mass parameter:

[0081] In some embodiments of the present invention, during the driving process of the vehicle, the sprung mass is almost constant. Therefore, it can be considered that the sprung mass at the previous moment is equal to the sprung mass at the next moment, that is:

[0082] m k+1 = m k (14)

[0083] where the subscripts k and k + 1 represent moments.

[0084] For a quarter-vehicle model with a damping continuously adjustable semi-active suspension, the following dynamic equation for the sprung mass is:

[0085]

[0086] where m s is the sprung mass, x s is the vertical displacement of the sprung mass, c s is the fixed damping coefficient of the shock absorber, k s is the suspension spring stiffness, u is the adjustable actuating force of the shock absorber, are the first derivatives of x s and x u respectively, is the second derivative of x s , and x u is the vertical displacement of the unsprung mass.

[0087] Then equation (14) can be rewritten as:

[0088]

[0089] where H is the suspension resultant force including the shock absorber damping force and the spring force, and its expression is:

[0090]

[0091] Select the state variable measurement variable system input u = [u] T , and the following observer for the sprung mass parameter can be established:

[0092]

[0093] where the subscripts k and k + 1 represent moments; θ and θ k+1 represent the reciprocals of the sprung mass at moment k and moment k + 1 respectively, and yk+1 denotes the measured value of the unsprung mass acceleration sensor at time \(k + 1\), \(H\) k denotes the suspension resultant force at time \(k\); \(w\) k is the process noise at time \(k\), \(v\) k is the measurement noise at time \(k\), and the two respectively satisfy \(w\sim N(0,Q\) w ), \(v\sim N(0,R\) v ) Gaussian distributions, where \(Q\) w and \(R\) v respectively represent the covariance matrices satisfied by the process noise and the measurement noise. In addition, the noises \(w\) and \(v\) are independent of each other and independent of the state variables.

[0094] The selected state variable \(\theta\) is the reciprocal of the unsprung mass, so that the output equation of the parameter observer is a linear equation.

[0095] (2) Establish a vehicle state observer:

[0096] For a quarter-vehicle model with a continuously adjustable semi-active suspension, there is the following dynamic equation:

[0097]

[0098] where \(m\) s is the unsprung mass, \(m\) u is the sprung mass, \(x\) s is the vertical displacement of the sprung mass, \(x\) u is the vertical displacement of the unsprung mass, \(x\) r is the road surface excitation, \(c\) s is the fixed damping coefficient of the shock absorber, \(k\) s is the suspension spring stiffness, \(k\) t is the vertical stiffness of the tire, and \(u\) is the adjustable actuator force of the shock absorber.

[0099] Select the state variable The measurement variable \(y = [x s -x u T , the system input \(u = [u] T , and the following vehicle state observer can be established:

[0100]

[0101] In Equation (16):

[0102] C =

[1000]

[0103] where, ​is the first derivative of the state variable x, A is the system state matrix, B is the input matrix, and C is the output state matrix; ω is the process noise and υ is the measurement noise, and both satisfy ω ~ N(0, Q ω ), υ ~ N(0, R υ ), where Q ω and R υ represent the covariance matrices satisfied by the process noise and the measurement noise respectively. In addition, the noises ω and υ are independent of each other and independent of the state variable.

[0104] Considering that components such as sensors and ECUs in practical engineering applications operate at a sampling time T, therefore, the vehicle state observer is discretized:

[0105]

[0106] Thus, the discretized form of the vehicle state observer is obtained:

[0107]

[0108] where the subscripts k and k + 1 represent moments, and the subscript d represents the discretized form; A d , B d , C d are matrices in the discretized form.

[0109] (3) Initialize the parameter observer and the state observer:

[0110] Set the model parameters of the sprung mass parameter observer, including setting the initial sprung mass parameter value θ0 and the initial parameter error covariance matrix According to the parameter model error and the measurement error of the sprung mass acceleration sensor, set the covariance matrices Q w and R v satisfied by the process noise w and the measurement noise v; set the sampling time T θ of the sprung mass parameter observer.

[0111] Set the model parameters of the vehicle state observer, including setting the initial vehicle state x0 and the initial state error covariance matrix According to the state model error and the measurement error of the suspension height sensor, set the covariance matrices Q ω and R υ satisfied by the process noise ω and the measurement noise υ; set the sampling time T x of the vehicle state observer.

[0112] In some embodiments of the present invention, the covariance matrices Q w and Q ωIt is determined by the method of optimizing the estimation result by adjusting the value offline, and the covariance matrix R satisfied by the measurement noises v and υ representing the sensor measurement error v and R υ is determined by the method of noise analysis of the sensor measurement signal.

[0113] (4) Calculate the suspension resultant force H at time k k :

[0114] The dynamic deflection x of the suspension at time k is measured by the suspension height sensor s -x u , and its signal is discretely differentiated to obtain the dynamic velocity of the suspension at time k In addition, the adjustable driving force u of the shock absorber at time k is obtained from the damping continuously adjustable shock absorber control unit k . According to Equation (19), calculate the suspension resultant force H at time k k :

[0115]

[0116] (5) Calculate the parameter predicted value at time k and the parameter prediction error covariance

[0117] According to the prediction equation of the sprung mass parameter observer, the parameter predicted value at time k is calculated from the parameter estimated value at time k-1 Calculate the parameter predicted value at time k

[0118]

[0119] where the superscript hat represents the predicted value and the superscript tilde represents the estimated value.

[0120] Calculate the parameter prediction error covariance at time k The solution calculation formula is as follows:

[0121]

[0122] where the superscript hat represents the prediction error covariance and the superscript tilde represents the estimation error covariance.

[0123] (6) Calculate the parameter estimated value at time k and the parameter estimation error covariance

[0124] According to the deviation between the measurement value of the sprung mass acceleration sensor and the output predicted value of the parameter observer, the parameter predicted value at time k is feedback corrected, and the solution calculation formula is as follows:

[0125]

[0126] In Equation (6):

[0127]

[0128] Wherein, is the parameter deviation gain at time k, is the measured value of the sprung mass acceleration sensor at time k.

[0129] Calculate the parameter estimation error covariance at time k The solution calculation formula is as follows:

[0130]

[0131] Wherein, I represents the identity matrix.

[0132] The parameter estimation error covariance obtained according to Equation (7) is used for the calculation of the parameter prediction error covariance at the next moment .

[0133] (7) Update the vehicle state observer at time k:

[0134] Take the reciprocal of the parameter estimation value to obtain the estimated value of the sprung mass at time k Substitute the estimated value of the sprung mass at time k into Equation (2) to update the vehicle state observer, and discretize the vehicle state observer according to the sampling time T x of the state observer.

[0135] (8) Calculate the state prediction value at time k and the state prediction error covariance

[0136] The adjustable actuator force u at time k-1 is obtained by the continuously adjustable damper control unit k-1 , and according to the prediction equation of the vehicle state observer, the state prediction value at time k is calculated from the state estimation value at time k-1 The solution calculation formula is as follows:

[0137]

[0138] Wherein, the hatched superscript represents the predicted value, and the tilde superscript represents the estimated value.

[0139] Calculate the state prediction error covariance at time k The solution calculation formula is as follows:

[0140]

[0141] Among them, the superscript broken line represents the prediction error covariance, and the superscript wavy line represents the estimation error covariance.

[0142] (9) Calculate the state estimate value at time k and the state estimation error covariance

[0143] According to the deviation between the measurement value of the suspension height sensor and the predicted value of the output of the state observer, the state prediction value at time k is feedback corrected, and the calculation formula is as follows:

[0144]

[0145] In formula (11):

[0146]

[0147] Among them, is the state deviation gain at time k, is the measurement value of the suspension height sensor at time k.

[0148] Calculate the state estimation error covariance at time k The calculation formula is as follows:

[0149]

[0150] Among them, I represents the identity matrix.

[0151] The state estimation error covariance obtained according to formula (12) is used for the calculation of the state prediction error covariance at the next moment.

[0152] (10) Output the estimated value of the unsprung mass and the vehicle state estimation result at time k;

[0153] (11) Let k = k + 1, and repeat the above steps (4) to (10) to obtain the estimated value of the unsprung mass and the vehicle state estimation result at the next moment.

[0154] In some embodiments of the present invention, specific examples are used to verify the effectiveness of the method of the present invention:

[0155] A quarter semi-active suspension vehicle model is established in the simulation software. Among them, the vehicle parameters are shown in Table 1. At the same time, a C-class road surface excitation is established to act on the vehicle model, and the vehicle speed is set at 72 km / h. First, the sprung mass acceleration signal and the suspension dynamic deflection signal are led out from the vehicle model, and the signals are respectively connected to the sprung mass parameter observer and the vehicle state observer. Secondly, the estimated value of the sprung mass is output from the sprung mass parameter observer and connected to the vehicle state observer. Thirdly, the vehicle state estimation results are output from the vehicle state observer, namely the suspension dynamic deflection, the sprung mass speed, the tire dynamic deformation and the unsprung mass speed. Finally, the estimated value of the sprung mass and the vehicle state estimation results are compared with the actual results.

[0156] Among them, Figure 4 is a comparison chart of the estimated result, actual result and original estimated result of the sprung mass of the present invention; Figure 5 is a comparison chart of the estimated result, actual result and original estimated result of the suspension dynamic deflection of the present invention; Figure 6 is a comparison chart of the estimated result, actual result and original estimated result of the sprung mass speed of the present invention; Figure 7 is a comparison chart of the estimated result, actual result and original estimated result of the tire dynamic deformation of the present invention; Figure 8 is a comparison chart of the estimated result, actual result and original estimated result of the unsprung mass speed of the present invention. These results show that the vehicle state estimation method based on parameter identification proposed by the present invention can estimate and dynamically correct the parameter values of the sprung mass in real time, can effectively reduce the model error of the state observer caused by parameter errors, and further improve the estimation accuracy of the vehicle running state. Therefore, the method of the present invention can better meet the requirements of the semi-active suspension vehicle running state estimation, so as to give full play to the control effect of the semi-active suspension and further achieve better control performance.

[0157] Table 1 Quarter vehicle model parameters

[0158]

[0159]

[0160] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A vehicle state estimation method based on parameter identification, characterized in that, It includes the following steps: (1) Establish a sprung mass parameter observer; Select state variables , measurement variables , system input , and establish the following observer for the sprung mass parameters: (1) where the subscripts and represent time instants; represents the measured value of the sprung mass acceleration sensor at the time instant, is the sprung mass, is the vertical acceleration of the sprung mass, is the adjustable power of the shock absorber, is the suspension resultant force including the shock absorber damping force and the spring force; is the process noise, is the measurement noise, and the two respectively satisfy and the Gaussian distributions, where and respectively represent the covariance matrices satisfied by the process noise and the measurement noise; (2) Establish a vehicle state observer; Select state variables , measurement variables , system inputs , and establish the following vehicle state observer: (2) where the subscript and represent time instants; is the vertical displacement of the sprung mass, is the vertical displacement of the unsprung mass, is the adjustable power of the shock absorber; the subscript represents the discretized form; is the discretized system state matrix, is the discretized input matrix, is the discretized output state matrix; is the process noise, is the measurement noise, and the two respectively satisfy , the Gaussian distribution, where and respectively represent the covariance matrices satisfied by the process noise and the measurement noise, is the road surface excitation; (3) Initialize the sprung mass parameter observer and the vehicle state observer; Set the model parameters of the sprung mass parameter observer, including setting the initial value of the sprung mass parameter , the initial parameter error covariance matrix ; Set the process noise and the measurement noise to satisfy the covariance matrix and ; Set the sampling time of the sprung mass parameter observer; Set the model parameters of the vehicle state observer, including setting the initial vehicle state , the initial state error covariance matrix ; Set the process noise and the measurement noise to satisfy the covariance matrix and ; Set the sampling time of the vehicle state observer; (4) Calculate the suspension resultant force at the moment : (3) Among them, is the dynamic speed of the suspension, is the dynamic deflection of the suspension, is the adjustable power of the shock absorber, is the fixed damping coefficient of the shock absorber, is the stiffness of the suspension spring; (5) Calculate The predicted value of the parameter at the moment And the predicted error covariance of the parameter : (4) (5) Wherein, the superscript broken line represents the prediction situation, and the superscript wavy line represents the estimation situation; (6) Calculate The parameter estimation value at the moment And the parameter estimation error covariance : (6) (7) In equations (6) and (7): (8) Among them, is the parameter deviation gain at the moment, is the measured value of the unsprung mass acceleration sensor at the moment, represents the identity matrix; (7) Update Vehicle state observer at a moment: For the parameter estimated value Take the reciprocal to obtain The estimated value of the unsprung mass at time , and substitute The estimated value of the unsprung mass at time into Equation (2) to update the vehicle state observer, and discretize the vehicle state observer according to the sampling time of the state observer; (8) Calculate The predicted value of the state at a certain moment And the state prediction error covariance : (9) (10) Wherein, the superscript broken line represents the prediction situation, and the superscript wavy line represents the estimation situation; (9) Calculate The state estimation value at the moment And the state estimation error covariance : (11) (12) In equations (11) and (12): (13) Among them, is the state deviation gain at a moment, is the measured value of the suspension height sensor at a moment, represents the identity matrix; Output (10) The estimated value of the unsprung mass at a moment and the estimated result of the vehicle state (11) Let , repeat the above steps (4) to (10) to obtain the estimated results of the unsprung mass and vehicle state at the next moment.

2. The vehicle state estimation method based on parameter identification according to claim 1, characterized in that, Step (1) is established based on the fact that the sprung mass is almost unchanged during vehicle driving, assuming that the sprung mass at the previous moment is equal to that at the next moment, i.e.: (14) Among them, the subscript and represent moments, is the sprung mass.

3. The vehicle state estimation method based on parameter identification according to claim 1, characterized in that, Step (2) is obtained by discretizing the continuous form of the state equation, and the discretization method takes into account the sampling time. The specific method is: (15) where the subscripts and represent moments, is a state variable, is a system input, is a process noise, is a system state matrix, is an input matrix, is a sampling time.

4. A vehicle state estimation method based on parameter identification according to claim 1, characterized in that The covariance matrix satisfied by the process noise is determined by the method of offline adjusting the value to make the estimation result optimal, and the covariance matrix satisfied by the measurement noise is determined by the method of noise analysis of the sensor measurement signal.

5. A vehicle state estimation method based on parameter identification according to claim 1, characterized in that, The suspension height sensor measures the dynamic deflection of the suspension, and by performing discrete differentiation on its signal, the dynamic speed of the suspension is obtained.

6. The vehicle state estimation method based on parameter identification according to claim 1, characterized in that In steps (5) and (8), the predicted value at the next moment is calculated from the estimated value at the previous moment according to the prediction equation of the observer.

7. A vehicle state estimation method based on parameter identification according to claim 1, characterized in that According to the deviation between the measured value of the sprung mass acceleration sensor and the predicted value of the output of the vehicle parameter observer, the parameter predicted value is feedback corrected.

8. A vehicle state estimation method based on parameter identification according to claim 1, characterized in that The estimated error covariance obtained in steps (6) and (9) is used for the calculation of the prediction error covariance at the next moment.

9. A vehicle state estimation method based on parameter identification according to claim 1, characterized in that, According to the deviation between the measured value of the suspension height sensor and the predicted value of the output of the state observer, the state predicted value is feedback corrected.

10. A vehicle state estimation method based on parameter identification according to any one of claims 1-9, characterized in that, The process of obtaining the vehicle state observer in step (2) includes: Select state variables , measurement variables , system inputs , the following vehicle state observer can be established: (16) In equation (16): ， ， Among them, is the first derivative of the state variable . is the system state matrix, is the input matrix, is the output state matrix; is the process noise, is the measurement noise, and both satisfy , Gaussian distribution, where and represent the covariance matrices satisfied by the process noise and the measurement noise respectively, is the unsprung mass; Discretize the vehicle state observer: (17) The discretized form of the vehicle state observer can be obtained: (18) 。

Images