Automobile heat network parameter adaptive identification method, system and device and storage medium

Abstract

The invention relates to the technical field of new energy automobile thermal management systems, and particularly provides an automobile thermal network parameter adaptive identification method, system and device and a storage medium, and the method comprises the steps: building a whole automobile thermal network mechanism model containing to-be-identified thermal network parameters; collecting system operation data to form a training data set; calculating a data quantity and a quality index; selecting an offline identification strategy according to the data quantity index, and obtaining an initial thermal network parameter; and deploying the initial parameters to a vehicle end, selecting an online identification strategy according to the data quality index, and carrying out online updating and correction on the parameters. According to the method, an off-line and on-line fused adaptive identification framework is constructed, the optimal identification strategy is dynamically selected according to the data quantity and quality, the problem of adaptability of a single method in the whole life cycle is solved, and the precision and robustness of parameter identification are remarkably improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of new energy vehicle thermal management system, and particularly relates to a vehicle thermal network parameter adaptive identification method, system, device and storage medium. BACKGROUND

[0002] The new energy vehicle thermal management system is the core of ensuring the safety and reliability of the whole vehicle, and its precise control depends on the accuracy of the equivalent thermal network model. However, the key parameters such as thermal capacity and thermal resistance in the model are difficult to accurately identify in actual application, which constitutes a core technical challenge. The existing technical solutions have obvious limitations: the pure offline identification method depends on specific test data, and is prone to parameter mismatch and poor generalization ability when the working condition changes; and the pure online identification strategy can be dynamically updated, but is prone to estimation deviation or even divergence under the conditions of initial value selection, data quality fluctuation and vehicle-mounted computing power limitation, and has insufficient robustness. Neither relying on offline fitting nor relying on online recursion, a single parameter identification path is difficult to meet the whole life cycle requirements of the vehicle from the bench test to the full working condition running. Therefore, an integrated parameter identification method that can integrate the advantages of offline and online and adaptively adjust the strategy according to the actual data conditions is urgently needed. SUMMARY

[0003] In view of the above deficiencies of the prior art, the application provides a vehicle thermal network parameter adaptive identification method, system, device and storage medium to solve the above technical problems.

[0004] In a first aspect, the application provides a vehicle thermal network parameter adaptive identification method, comprising: establishing a mechanism model of the whole vehicle thermal network, wherein the mechanism model comprises a plurality of thermal network parameters to be identified; collecting state quantities and input quantities of the whole vehicle thermal management system during operation to form a training data set; calculating a data quantity index and a data quality index based on the training data set; selecting and executing one of a plurality of preset offline identification strategies according to the numerical range of the data quantity index to identify the thermal network parameters and obtain initial thermal network parameters; updating the initial thermal network parameters to the mechanism model at the vehicle end, and selecting and executing one of a plurality of preset online identification strategies according to the data quality index to perform online update and correction on the current thermal network parameters of the mechanism model.

[0005] In an optional embodiment, the mechanism model of the whole vehicle thermal network comprises: based on the equivalent RC network, a state space equation in the following form is established as the mechanism model:

[0006]

[0007] in, This is a state vector, representing the temperature of each key component; The input vector represents the heating power and ambient temperature; For observation vectors; The rate of change of the state vector; the thermal network parameters to be identified are parameterized in the system matrix. Input matrix Specifically, it includes the heat capacity and thermal resistance of each component; matrix sum matrix It is the parameter matrix corresponding to the observation vector.

[0008] In one optional implementation, the state variables and input variables of the vehicle thermal management system during operation are collected to form a training dataset, including: The vehicle temperature sensor collects the motor temperature, motor controller temperature, multi-function controller temperature and coolant temperature as status variables. The heat output of the motor, motor controller, and all-in-one controller is obtained as input through data from the controller local area network bus or based on current and voltage sensor data, while the coolant mass flow rate and ambient temperature are collected. All collected data are preprocessed, including handling missing values ​​and removing physically unreasonable outliers, to form the training dataset.

[0009] In an optional implementation, based on the training dataset, data quantity metrics and data quality metrics are calculated, including: The training dataset is divided into preset time windows, and data quantity indicators and data quality indicators are calculated based on the data in each time window. The formula for calculating the data quantity index is:

[0010] in, This indicates the amount of valid data within the current time window. The minimum number of samples required to complete one effective identification; The formula for calculating data quality indicators is:

[0011] in, This indicates the number of data points marked as missing or invalid within the current time window.

[0012] In an optional embodiment, according to the value range of the data quantity indicator, one of a plurality of preset offline identification strategies is selected and executed to identify the thermal network parameters, to obtain initial thermal network parameters, including: The data quantity indicator is compared with a preset first threshold value and a second threshold value . ; If , a first strategy is executed, which identifies the parameters through a physical information neural network, specifically including: taking time and input quantity as network input to predict temperature state quantity; embedding the state space equation of the mechanism model into the loss function of the neural network as a physical constraint, the loss function including a data fitting term and a physical constraint term; minimizing the loss function through an optimization algorithm, so as to identify the thermal network parameters as network parameters of the physical information neural network; If , a second strategy is executed, which identifies the parameters through a nonlinear least square method, specifically including: constructing a weighted square sum of differences between predicted temperature and measured temperature as an objective function, and minimizing the objective function through a numerical optimization algorithm to obtain the thermal network parameters; If , a third strategy is executed, which identifies the parameters by performing global parameter search within a preset physically feasible domain and selecting a set of optimal thermal network parameters according to the maximum a posteriori estimation principle.

[0013] In an optional embodiment, the set of optimal thermal network parameters is selected by performing global parameter search within a preset physically feasible domain and according to the maximum a posteriori estimation principle, including: Based on engineering prior knowledge, a physically feasible parameter value interval is set for each thermal network parameter to be identified, to form the physically feasible domain; Within the physically feasible domain, a large number of candidate thermal network parameter combinations are generated by using grid search or random sampling; For each candidate thermal network parameter combination , the following calculation is performed: The parameter combination is substituted into the mechanism model to calculate the model output under the current training data set ; The maximum a posteriori probability of the parameter combination is calculated, and the objective function thereof is composed of a likelihood term and a prior term, and is expressed as:

[0014] wherein, is the measured value of the jth sample in the training data set, ​a corresponding model prediction value, a parameter prior value based on historical data or experience, a weight coefficient of the prior term; a likelihood term, a prior term; comparing the numerical values of all candidate parameter combinations, the smallest parameter combination is selected as the optimal thermal network parameter.

[0015] In an optional embodiment, the initial thermal network parameter is updated to the mechanism model at the vehicle end, and according to the data quality index, one of a plurality of preset online identification strategies is selected and executed to perform online updating and correction of the current thermal network parameter of the mechanism model, including: comparing the data quality index Q q with a preset third threshold value a q ; if Q q ≥ a q , a fourth strategy is executed, which uses a recursive least square method with a forgetting factor to update the parameter, specifically including: obtaining the thermal network parameter estimation value θ k-1 and the covariance matrix P k-1 of the last time; calculating the gain vector K and the parameter update amount according to the observation vector y of the current time and the forgetting factor; obtaining the thermal network parameter estimation value θ k of the current time by recursion according to the error between the current observation value y k and the model prediction value ; if Q q < a q , a fifth strategy is executed, which uses a frozen parameter method, specifically: keeping the current thermal network parameter of the mechanism model unchanged, and skipping the online updating process of the current time window.

[0016] In a second aspect, the present application provides a vehicle thermal network parameter adaptive identification system, including: a model establishment module for establishing a mechanism model of the vehicle thermal network, the mechanism model including a plurality of thermal network parameters to be identified; a data acquisition module for acquiring state quantities and input quantities of the vehicle thermal management system during operation to form a training data set; an index calculation module for calculating a data quantity index and a data quality index based on the training data set; An offline identification module is configured to select and execute one of a plurality of preset offline identification strategies according to a numerical range of the data quantity indicator, so as to identify the thermal network parameter and obtain an initial thermal network parameter. An online identification module is configured to update the initial thermal network parameter to a mechanism model at a vehicle end, and select and execute one of a plurality of preset online identification strategies according to the data quality indicator, so as to perform online update and correction on a current thermal network parameter of the mechanism model.

[0017] In a third aspect, an apparatus is provided, comprising: a memory configured to store an automotive thermal network parameter adaptive identification program; a processor configured to implement steps of the automotive thermal network parameter adaptive identification method provided in the first aspect when the automotive thermal network parameter adaptive identification program is executed.

[0018] In a fourth aspect, a computer readable storage medium is provided, and the automotive thermal network parameter adaptive identification program is stored on the computer readable storage medium, and steps of the automotive thermal network parameter adaptive identification method provided in the first aspect are implemented when the automotive thermal network parameter adaptive identification program is executed by a processor.

[0019] The automotive thermal network parameter adaptive identification method, system, apparatus and storage medium provided by the present application have the beneficial effects that the adaptive identification framework combining offline and online is constructed, the optimal identification strategy is dynamically selected according to data quantity and quality, and the adaptability problem of a single method in the whole life cycle is effectively solved. The beneficial effects are that the identification accuracy and model generalization ability of the thermal network parameter under different data conditions are significantly improved; the parameter distortion and waste of computing resources caused by low-quality data are avoided through the intelligent strategy switching and update decision mechanism; the real-time control robustness and reliability of the thermal management system are enhanced, and a precise model basis is provided for the comprehensive improvement of the energy efficiency and safety of new energy vehicles. BRIEF DESCRIPTION OF DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, brief descriptions will be given below to the drawings needed to be used in the embodiments or prior art descriptions, and obviously, other drawings can also be obtained by those skilled in the art without creative effort on the premise of not paying creative effort.

[0021] Figure 1 is a schematic flowchart of a method according to an embodiment of the present application.

[0022] Figure 2 is a schematic block diagram of a system according to an embodiment of the present application.

[0023] Figure 3A structural schematic diagram of a device provided by an embodiment of the present application is shown. DETAILED DESCRIPTION

[0024] In order for those skilled in the art to better understand the technical solutions in the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative efforts should fall within the scope of protection of the present application.

[0025] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs. The terminology used in the description of the present application herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the present application.

[0026] The automobile thermal network parameter adaptive identification method provided by the embodiment of the present application is executed by a computer device, and accordingly, the automobile thermal network parameter adaptive identification system runs in the computer device.

[0027] Figure 1 is a schematic flowchart of the method of an embodiment of the present application. In which, Figure 1 The execution subject can be an automobile thermal network parameter adaptive identification system. The order of the steps in the flowchart can be changed according to different needs, and some can be omitted.

[0028] As Figure 1 shown, the method comprises: S1. establishing a mechanism model of a whole vehicle thermal network, the mechanism model comprising a plurality of thermal network parameters to be identified; S2. collecting state quantities and input quantities of the whole vehicle thermal management system in the running process to form a training data set; S3. calculating a data quantity index and a data quality index based on the training data set; S4. according to the numerical range of the data quantity index, selecting and executing one of a plurality of preset offline identification strategies to identify the thermal network parameters, and obtaining initial thermal network parameters; S5. updating the initial thermal network parameters to the mechanism model at the vehicle end, and according to the data quality index, selecting and executing one of a plurality of preset online identification strategies to perform online update and correction on the current thermal network parameters of the mechanism model.

[0029] In an embodiment of the present application, based on step S1, a possible embodiment will be given below to non-restrictively describe the specific implementation scheme thereof.

[0030] Based on the equivalent RC network, the state space equation in the following form is established as the mechanism model:

[0031]

[0032] wherein, is the state vector, representing the temperature of each key component; is the input vector, representing the heat generation power and the ambient temperature; is the observation vector; is the rate of change of the state vector; the to-be-identified thermal network parameters are parameterized in the system matrix , the input matrix , the specific heat and the thermal resistance corresponding to each component; the matrix and the matrix are the parameter matrices corresponding to the observation vector.

[0033]

[0034]

[0035] A

[0036]

[0037]

[0038]

[0039] wherein, is the coolant mass flow rate, is the specific heat capacity of the coolant, is the temperature of the all-in-one controller, is the temperature of the motor controller, is the temperature of the motor, is the heat generation power of the all-in-one controller, is the heat generation power of the motor controller, is the heat generation power of the motor, is the thermal capacity of the all-in-one controller, is the thermal capacity of the motor controller, is the thermal capacity of the motor, is the equivalent thermal capacity of the radiator, is the equivalent thermal resistance of the all-in-one controller, is the equivalent thermal resistance of the motor controller, is the equivalent thermal resistance of the motor and the coolant, is the equivalent thermal resistance of the radiator, Equivalent heat transfer capacity, Ambient temperature.

[0040] This state space is only for example, the purpose is to facilitate the reader to understand. Obviously all the heat capacity, thermal resistance, equivalent heat transfer capacity are the parameters that need to be identified. Set the physical feasible region and priori boundary.

[0041] In one embodiment of the present application, based on step S2, a possible embodiment will be given below to specifically illustrate the non-limiting embodiment.

[0042] Data acquisition: the acquisition of state variables is mainly realized by high-precision vehicle-mounted temperature sensors (such as PT1000 or NTC thermistor) deployed on key components. Specifically, these sensors directly measure the motor stator winding temperature, motor controller (IGBT module) shell temperature, multi-in-one controller power unit temperature, and cooling liquid temperature at the inlet and outlet of the cooling system, and take these temperature values as the state variables of the system.

[0043] The acquisition of input variables is obtained through multi-source data fusion: Heat generation power: for the heat generation power of the motor, motor controller and multi-in-one controller, the real-time power value calculated and issued by the upper controller is directly read through the vehicle's controller area network (CAN bus) as a priority. If there is no such signal, a calculation method based on sensor data is used, that is, the working current and voltage of each component are synchronously collected (through Hall current sensor and voltage sensor), and the formula Q=I×U×(1-η) (where η is the efficiency coefficient, which is usually obtained by table lookup method) is used for estimation.

[0044] Other input variables: the cooling liquid mass flow is measured by a flow sensor installed in the cooling circuit; the ambient temperature is provided by an external environment temperature sensor.

[0045] Data preprocessing: the raw data stream collected needs to be preprocessed to form a high-quality training data set. The preprocessing link mainly includes: Processing data missing values: for single or continuous data missing caused by signal instantaneous interruption, linear interpolation method based on time series is used for filling; for large area data loss invalid segment, it is deleted in whole.

[0046] Eliminate physically unreasonable outliers: based on the first law of thermodynamics and the working limit of components, set reasonable physical boundaries for each parameter (such as the motor temperature cannot be lower than the ambient temperature, the heat generation power cannot be negative, etc.). The interquartile range (IQR) method or sliding window Z-score method are used for joint judgment, and the abnormal points that exceed the hard boundary or statistical rule are identified and removed.

[0047] Finally, the time-synchronously preprocessed state quantity sequence and the input quantity sequence are collectively used as the training data set D for the application, providing a reliable data basis for subsequent parameter identification.

[0048] In an embodiment of the application, based on step S3, a possible embodiment will be given below to illustrate the non-limiting specific implementation thereof.

[0049] In order to adapt to the time-varying characteristics in vehicle operation and realize online identification, the continuous training data set is first divided according to a pre-set fixed time window. The window length T can be set according to actual application requirements, for example, set to 300 seconds to cover a relatively complete driving cycle. A sliding window mechanism can be used in processing to balance the calculation load and the timeliness of parameter updating.

[0050] The calculation formula of the data quantity index is:

[0051] wherein, represents the effective data quantity in the current time window. The minimum sample quantity required for completing an effective identification is a pre-set constant, and its size depends on the complexity of the thermal network mechanism model (i.e. the number of parameters to be identified) and the dynamic response characteristics of the system. For example, for a model with 10 parameters to be identified, 1000 can be set. This formula ensures that is a normalized index between 0 and 1, and when the actual data quantity N reaches or exceeds the reference value , the index saturates to 1, indicating that the data quantity is sufficient.

[0052] The calculation formula of the data quality index is:

[0053] wherein, represents the number of data points marked as missing or invalid in the current time window. These invalid points come from the physically unreasonable abnormal values identified in the aforementioned preprocessing link and the missing points that fail to be repaired by interpolation.

[0054] In an embodiment of the application, based on step S4, a possible embodiment will be given below to illustrate the non-limiting specific implementation thereof.

[0055] The data quantity index is compared with a pre-set first threshold and a second threshold , wherein ; based on the specific application scenario, , The calculated data quantity indicators The comparison is made against these two thresholds, and one of the following three strategies is adaptively selected and executed based on the comparison result.

[0056] like If the first strategy is executed, this strategy uses a physical information neural network for parameter identification. Specifically, it includes: using time and input quantities as network inputs to predict temperature state quantities; embedding the state-space equations of the mechanism model as physical constraints into the loss function of the neural network, which includes data fitting terms and physical constraint terms; and minimizing the loss function using an optimization algorithm to identify the thermal network parameters as network parameters of the physical information neural network. The specific implementation steps are as follows: (1) Organize the training dataset into training data according to the time series sampling interval.

[0057] (2) Construct a feature embedding module. This module can be a multilayer feedforward neural network, a recurrent neural network, or other neural network structures. The inputs of the feature embedding module are t and u, and the output is the temperature prediction. A training network is constructed using a physical information network structure, which parameterizes the physical quantity to be identified into trainable network parameters. ,in, This represents network parameters.

[0058] (3) The state-space equations of the aforementioned mechanism model As a physical constraint, the predicted state variables are calculated using automatic differentiation techniques. The derivative with respect to time t, and the physical constraint loss term are constructed. The specific expressions are as follows:

[0059]

[0060] in, This represents the estimated value of the measurement. This represents the total number of sampling points. This represents the estimated value of the state variables. Furthermore, the overall loss function can be expressed as follows:

[0061] in, These represent the weighting coefficients, used for active parameter tuning.

[0062] (4) Minimize using the Adam optimizer Finally, the hot network parameters (elements in matrices A and B) are identified as the network parameters after training convergence. Use reparameterization techniques to ensure that the identification parameters are always positive.

[0063] like Then, the second strategy is executed. This strategy uses nonlinear least squares method for parameter identification, specifically: constructing a weighted sum of squares of the difference between the predicted temperature and the measured temperature as the objective function, and minimizing this objective function through a numerical optimization algorithm to obtain the thermal network parameters; the specific implementation steps include: (1) Based on the mechanism model, according to the target identification and input vector The calculation model outputs the temperature. .

[0064] (2) Define the weighted squared error.

[0065] in, This represents the weight value, which can be dynamically adjusted based on data reliability.

[0066] (3) Employ numerical optimization methods such as the Levenberg-Marquardt algorithm or the trust region reflection algorithm to minimize the objective function. The optimal thermal network parameter estimates are obtained. During the optimization process, it is necessary to verify whether the parameters meet the preset physical boundary constraints.

[0067] like Then, the third strategy is executed. This strategy involves a global parameter search within a predefined physical feasible region and selects a set of optimal thermal network parameters based on the maximum a posteriori estimation principle. Specifically, this includes: Based on prior engineering knowledge, physically feasible value ranges are defined for each thermal network parameter to be identified. These ranges are determined based on: component design parameters, material properties, historical experimental data, and simulation analysis results. For example, the motor's heat capacity C... motor The value range can be set to [500, 5000] J / ℃; the equivalent thermal resistance R between the motor and the coolant motor The value range can be set to [0.05, 0.5] ℃ / W; the controller heat capacity C inv The value range can be set to [100, 2000] J / ℃, and the feasible intervals of all parameters together constitute the physical feasible domain. .

[0068] In the defined physical feasible region Within this framework, a systematic sampling strategy is employed to generate candidate parameter combinations: when the number of parameters to be identified is small (e.g., ≤5), a grid search method is used to evenly divide the interval of each parameter into several equal parts, generating all possible parameter combinations. When the parameter dimensionality is high, a Latin hypercube sampling method is used to ensure that the samples can fully cover the entire parameter space while maintaining the independence of each parameter dimension. Through these methods, thousands to tens of thousands of candidate parameter combinations are generated. A set of.

[0069] For each candidate combination of thermal network parameters Perform the following calculations: Will Substituting the values ​​into the state-space equations of the aforementioned mechanism model, the model output sequence at all time steps in the current training dataset is calculated using numerical integration methods (such as the fourth-order Runge-Kutta method). ; The objective function for calculating the maximum posterior probability of this parameter combination, consisting of a likelihood term and a prior term, is expressed as:

[0070] in, To obtain the measured value of the j-th sample in the training dataset, The corresponding model prediction value, These are prior values ​​of parameters based on historical data or experience. These are the weighting coefficients for the prior terms; For likelihood terms, It is a priori; Comparing all candidate parameter combinations The value will make The minimum parameter combination is selected as the optimal thermal network parameters.

[0071] In one embodiment of the present invention, based on step S5, a possible embodiment will be given below, and its specific implementation will be described in a non-limiting manner.

[0072] Set the third threshold α q As a threshold for data quality, it is typically set to α based on specific application requirements. q =0.7. The data quality index Q, calculated in real time, is... q The data is compared with this threshold, and an online update strategy is dynamically selected based on the comparison result.

[0073] Fourth strategy (good data quality, Q) q ≥α q When the data quality is good, the recursive least squares method with a forgetting factor is used for real-time parameter updates. The specific implementation process is as follows: Obtain the parameter estimates from the previous time step. Covariance matrix Current period observations Regression vector Forgetting factor The forgetting factor is usually a fixed value.

[0074] according to and Calculate the normalization factor The specific formula is as follows:

[0075] according to The recursive least squares gain is calculated using the following formula:

[0076] The observation error is calculated, and the identification parameters are updated. The specific formula is as follows:

[0077]

[0078] The covariance is updated using the least squares method, with the specific formula as follows:

[0079] The calculation in this step Is it a replacement? When applied to actual model calculations, the quality of the identified parameters depends on subsequent steps in determining their quality.

[0080] If Q q <α q If the current thermal network parameters of the mechanism model remain unchanged, the online update process for this time window will be skipped.

[0081] Furthermore, in other embodiments of the present invention, to ensure the stability of the parameters, the following steps are also included: 1. Calculate the amount of information in the window to determine whether parameter updates are allowed within that time window, avoiding blind parameter changes. Specifically, this includes: The formula for calculating whether the amount of information within the current time window meets the requirements for online parameter updates is as follows:

[0082]

[0083] in, This represents the measurement noise estimation matrix used for weighting. Representation mechanism model, This indicates the size of the time window, and k represents the k-th sliding window.

[0084] The flag bit is updated based on the parameter calculation of the information content signal. The specific formula is as follows:

[0085] in, This represents the operation of calculating the minimum eigenvalue of a matrix. Indicates the information content threshold. When At that time, perform an online update.

[0086] 2. When determining whether an update is permissible, an uncertainty fusion strategy is employed, aging constraints are applied, and the online update parameter results are calculated. Specifically, this includes: Let the parameters of the previous time window update be... The covariance matrix is ​​expressed as The parameters updated in this round of time window are: The covariance matrix is ​​expressed as .

[0087] The specific formula for calculating the fusion weights of the current time window parameters is as follows:

[0088]

[0089] in, This represents the lower confidence limit and is a constant greater than zero. This represents the i-th parameter.

[0090] Based on the weights calculated in step 1, the fusion parameters are calculated using the following formula:

[0091] Calculate the posterior variance using the weights calculated in step 1. The specific formula is as follows:

[0092] Update the flag bit based on the parameters from step 1, and calculate the closed-form solution for one-dimensional optimization. The specific formula is as follows:

[0093] in, Let represent the upper and lower limits of the constraint values, respectively. Considering the issue of real-time calculation, the optimization objective can be simplified to:

[0094] The specific formula for calculating the identification parameters under aging constraints is as follows:

[0095] in, This represents the aging constraint parameter.

[0096] 3. Design a unified criterion to update the results of online parameter updates. If the criterion is met, the result is locked as unique, and no further online updates are performed. Specifically, this includes: Using the Exponential Weighted Moving Average (EWMA) as a reference for determining whether an evaluation parameter can be locked into a unique result. Calculate the EWMA for parameter increment, prediction residual, and observability respectively, using the following formulas:

[0097]

[0098]

[0099] in, Indicates parameter weights, This indicates that the Frobenius norm is used for calculation.

[0100] The three indicators are treated using a hysteresis threshold approach, which involves setting different entry / exit boundary conditions and calculating the lock flag. The specific formula is as follows:

[0101] Set a saturation counter; the type can be int16.

[0102] in, Indicates the decay parameter, Indicates the upper limit of the counter. Sets the two-stage counting determination method and sets the stage threshold. The specific strategy is as follows: when When, the frozen parameters are updated; when At that time, the lock parameters are updated. The difference between freezing and locking is that when in the parameter freezing phase, if... Less than due to the influence of the decay parameter When the parameters are locked, they immediately begin updating; when the parameters are locked, they will not be updated unless an unlock event is triggered. The unlock event can be... Excessive size, vehicle repair / parts replacement, major software version switch, etc.

[0103] In some embodiments, the adaptive identification system for automotive thermal network parameters may include multiple functional modules composed of computer program segments. The computer programs for each program segment in the adaptive identification system for automotive thermal network parameters may be stored in the memory of a computer device and executed by at least one processor to perform (see details). Figure 1 (Description) The function of adaptive identification of automotive thermal network parameters.

[0104] In this embodiment, the automotive thermal network parameter adaptive identification system can be divided into multiple functional modules based on its functions, such as... Figure 2 As shown. The module referred to in this invention is a series of computer program segments that can be executed by at least one processor and perform a fixed function, and is stored in memory. In this embodiment, the functions of each module will be described in detail in subsequent embodiments.

[0105] The model building module is used to build a mechanistic model of the whole vehicle thermal network, which includes multiple thermal network parameters to be identified. The data acquisition module is used to collect the state variables and input variables of the vehicle thermal management system during operation to form a training dataset; The indicator calculation module is used to calculate data quantity indicators and data quality indicators based on the training dataset; The offline identification module is used to select and execute one of multiple preset offline identification strategies based on the numerical range of the data quantity index, so as to identify the thermal network parameters and obtain the initial thermal network parameters. The online identification module is used to update the initial thermal network parameters to the mechanism model at the vehicle end, and select and execute one of multiple preset online identification strategies according to the data quality index to update and correct the current thermal network parameters of the mechanism model online.

[0106] Figure 3 The adaptive identification method for automotive thermal network parameters provided in this application embodiment can be applied to devices. Those skilled in the art will understand that the device structure involved in the embodiments of this invention does not constitute a limitation on the device. A device may include more or fewer components than illustrated, or combine certain components, or have different component arrangements. In the embodiments of this invention, the device includes, but is not limited to, laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the embodiments of this application described and / or claimed herein.

[0107] The device 300 may include a processor 310, a memory 320, and a communication unit 330. These components communicate via one or more buses. Those skilled in the art will understand that the server structure shown in the figure does not constitute a limitation of the present invention. It may be a bus topology or a star topology, and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0108] The memory 320 can be used to store execution instructions of the processor 310. The memory 320 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. When the execution instructions in the memory 320 are executed by the processor 310, the device 300 is able to perform some or all of the steps in the above method embodiments.

[0109] The processor 310 serves as the control center of the storage device, connecting various parts of the electronic device via various interfaces and lines. It executes software programs and / or modules stored in the memory 320, and calls data stored in the memory to perform various functions of the electronic device and / or process data. The processor can be composed of integrated circuits (ICs), such as a single packaged IC or multiple packaged ICs with the same or different functions connected together. For example, the processor 310 may consist only of a central processing unit (CPU). In this embodiment of the invention, the CPU may have a single processing core or include multiple processing cores.

[0110] The communication unit 330 is used to establish a communication channel, enabling the storage device to communicate with other devices. It can receive user data sent by other devices or send user data to other devices.

[0111] The present invention also provides a computer storage medium, wherein the computer storage medium may store a program, which, when executed, may include some or all of the steps provided in the embodiments of the present invention. The storage medium may be a magnetic disk, an optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0112] Those skilled in the art will clearly understand that the techniques in the embodiments of the present invention can be implemented using software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solutions in the embodiments of the present invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, or any other medium capable of storing program code. It includes several instructions to cause a computer device (which may be a personal computer, a server, or a second device, network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.

[0113] The same or similar parts between the various embodiments in this specification can be referred to mutually. In particular, the device embodiments are basically similar to the method embodiments, so the description is relatively simple, and the relevant parts can be referred to the description in the method embodiments.

[0114] In the embodiments provided by this invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between systems or modules may be electrical, mechanical, or other forms.

[0115] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0116] In addition, the functional modules in the various embodiments of the present invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0117] Although the present invention has been described in detail with reference to the accompanying drawings and preferred embodiments, the present invention is not limited thereto. Various equivalent modifications or substitutions can be made to the embodiments of the present invention by those skilled in the art without departing from the spirit and essence of the invention, and such modifications or substitutions should all be within the scope of the present invention. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should also be covered within the protection scope of the present invention.

Claims

1. A method for parameter adaptive identification of an automotive thermal network, characterized in that, The method comprises the following steps: a mechanism model of a whole vehicle thermal network is established, wherein the mechanism model comprises a plurality of thermal network parameters to be identified; state quantities and input quantities of the whole vehicle thermal management system during operation are collected to form a training data set; data quantity indexes and data quality indexes are calculated based on the training data set; one of a plurality of preset offline identification strategies is selected and executed according to a numerical range of the data quantity indexes, so as to identify the thermal network parameters and obtain initial thermal network parameters; the initial thermal network parameters are updated to the mechanism model at the vehicle end, and one of a plurality of preset online identification strategies is selected and executed according to the data quality indexes, so as to perform online updating and correction on current thermal network parameters of the mechanism model.

2. The method of claim 1, wherein, The method for establishing a mechanism model of a whole vehicle thermal network comprises the following steps: Based on the equivalent RC network, a state space equation in the following form is established as the mechanism model: wherein, is a state vector representing the temperature of each key component; is an input vector representing the heat generation power and the ambient temperature; is an observation vector; is a rate of change of the state vector; the thermal network parameters to be identified are parameterized in the system matrix , the input matrix , the matrix and the matrix are the parameter matrices corresponding to the observation vector.

3. The method of claim 1, wherein, The state quantities and input quantities of the whole vehicle thermal management system during operation are collected to form a training data set, which comprises the following steps: The motor temperature, motor controller temperature, multi-in-one controller temperature and cooling liquid temperature are collected by the vehicle-mounted temperature sensor as the state quantities; The heating power of the motor, motor controller and multi-in-one controller are obtained as the input quantities by the controller area network bus data or based on the current and voltage sensor data, and the cooling liquid mass flow and environment temperature are collected; All the collected data are preprocessed, which comprises processing data missing values and eliminating physically unreasonable abnormal values, so as to form the training data set.

4. The method of claim 1, wherein, The data quantity indexes and data quality indexes are calculated based on the training data set, which comprises the following steps: The training data set is divided according to a preset time window, and the data quantity indexes and data quality indexes are calculated according to the data in each time window; The calculation formula of the data quantity indexes is as follows: wherein, represents the amount of valid data within the current time window, is the minimum sample size required to complete one valid recognition; The calculation formula of the data quality indexes is as follows: wherein, represents the number of data points within the current time window that are marked as missing or invalid.

5. The method of claim 1, wherein, One of a plurality of preset offline identification strategies is selected and executed according to a numerical range of the data quantity indexes, so as to identify the thermal network parameters and obtain initial thermal network parameters, which comprises the following steps: comparing the data quantity indicator with a preset first threshold value and a second threshold value and a second threshold value wherein ; If a first strategy is executed, which performs parameter identification through a physical information neural network, specifically including: taking time and input quantity as network input, predicting temperature state quantity; embedding the state space equation of the mechanism model into the loss function of the neural network as a physical constraint, which contains a data fitting term and a physical constraint term; minimizing the loss function through an optimization algorithm, thereby identifying the thermal network parameters as the network parameters of the physical information neural network; If then a second strategy is executed, which performs parameter identification by a nonlinear least squares method, specifically: a weighted sum of squares of differences between predicted and measured temperatures is constructed as an objective function, and a numerical optimization algorithm is used to minimize the objective function to obtain the thermal network parameters; If then a third strategy is executed, which searches the global parameters within a preset physically feasible region and selects a set of optimal thermal network parameters according to the maximum a posteriori estimation principle.

6. The method of claim 5, wherein, A global parameter search is performed in a preset physically feasible domain, and a set of optimal thermal network parameters is selected according to the maximum a posteriori estimation principle, which comprises the following steps: Based on engineering prior knowledge, a physically feasible parameter value interval is set for each thermal network parameter to be identified to form the physically feasible domain; A large number of candidate thermal network parameter combinations are generated in the physically feasible domain by using a grid search or random sampling method; For each candidate combination of thermal network parameters the following calculations are performed: Substitute it into the mechanism model, calculate the model output under the current training data set ; The maximum a posteriori probability of the parameter combination is calculated, and the objective function thereof is composed of a likelihood term and a prior term, which is expressed as: wherein, is the measured value of the jth sample in the training data set, is the corresponding model predicted value, is the parameter prior value based on historical data or experience, is the weight coefficient of the prior term; is the likelihood term, is the prior term; comparing all candidate parameter combinations values that will result in the smallest parameter combination is selected as the optimal thermal network parameters.

7. The method of claim 1, wherein, The initial thermal network parameters are updated to the mechanism model at the vehicle end, and one of a plurality of preset online identification strategies is selected and executed according to the data quality indexes, so as to perform online updating and correction on current thermal network parameters of the mechanism model, which comprises the following steps: comparing the data quality indicator Q q with a preset third threshold value a q ; If Q q ≥ α q , a fourth strategy is executed, which adopts a recursive least square method with a forgetting factor for parameter updating, specifically comprising: Obtain the thermal network parameter estimation value θ of the last time k-1 and the covariance matrix P k-1 ; an observation vector at the current time instant a forgetting factor, a gain vector a parameter update Based on the current observation value y k and the error between the model prediction value , the heat network parameter estimation value θ k at the current time is recursively obtained. If Q q <α q If so, the fifth strategy is executed, which adopts a frozen parameter mode, specifically: the current thermal network parameters of the mechanism model are kept unchanged, and the online updating process of the current time window is skipped.

8. An automotive thermal network parameter adaptive identification system, characterized in that, The method comprises the following steps: A mechanism model of a whole vehicle thermal network is established, wherein the mechanism model comprises a plurality of thermal network parameters to be identified; A data collection module is configured to collect state quantities and input quantities of the whole vehicle thermal management system during operation to form a training data set; An index calculation module is configured to calculate a data quantity index and a data quality index based on the training data set; An offline identification module is configured to select and execute one of a plurality of preset offline identification strategies according to a numerical range of the data quantity index to identify the thermal network parameters and obtain initial thermal network parameters; An online identification module is configured to update the initial thermal network parameters to a mechanism model at a vehicle end and select and execute one of a plurality of preset online identification strategies according to the data quality index to perform online update and correction on current thermal network parameters of the mechanism model.

9. An automotive hot network parameter adaptive identification device, characterized by, The method comprises the following steps: a memory configured to store an automobile thermal network parameter adaptive identification program; a processor configured to implement the steps of the automobile thermal network parameter adaptive identification method according to any one of claims 1-7 when the automobile thermal network parameter adaptive identification program is executed.

10. A computer readable storage medium storing a computer program, characterized in that, The readable storage medium stores an automobile thermal network parameter adaptive identification program, and the automobile thermal network parameter adaptive identification program is executed by the processor to implement the steps of the automobile thermal network parameter adaptive identification method according to any one of claims 1-7.

Images