Vehicle-mounted intelligent controller and vehicle

Abstract

The application relates to the technical field of controllers, in particular to a vehicle-mounted intelligent controller and a vehicle, which comprise a controller hardware and a control model, wherein the controller hardware comprises a plurality of input ports, a control unit and a machine learning processing unit; the control model comprises a plurality of control submodels, the plurality of control submodels are distributedly stored in the control unit and the machine learning processing unit, the control unit calculates input signals of the plurality of input ports to obtain target signals, the machine learning processing unit determines control signals according to the target signals, and the control unit controls the vehicle according to the control signals. Therefore, the problems that the control method is simple, it is difficult to adapt to complex environments, the precision and efficiency are limited, and the method is easily affected by the state of a prototype in the prior art are solved.

Description

Technical Field

[0001] This application relates to the field of controller technology, and in particular to an in-vehicle intelligent controller and vehicle. Background Technology

[0002] With the development of technology, automobiles are becoming increasingly intelligent, and in-vehicle intelligent systems are an important component of this intelligence. For vehicles equipped with internal combustion engines as their power system, in-vehicle intelligent systems mainly rely on ECUs (Electronic Control Units) to collect vehicle status information through various sensors and perform calculations and controls according to preset programs to achieve precise control of the vehicle.

[0003] In existing technologies, in-vehicle intelligent systems monitor and control various vehicle states through preset rules and thresholds. However, this control method is too simplistic, unable to cope with complex and changing environments, lacks comprehensive operational condition coverage, cannot guarantee the accuracy of global model calculations, and struggles to achieve precise vehicle control. Furthermore, the calibration process is extremely time-consuming. Moreover, the control model has high requirements for prototype consistency; if the prototype exhibits significant variance or is aged, the model's fidelity deteriorates. Summary of the Invention

[0004] This application provides an in-vehicle intelligent controller and a vehicle to solve the problems of existing control methods being simple, difficult to adapt to complex environments, limited in accuracy and efficiency, and easily affected by the prototype status.

[0005] The first aspect of this application provides an in-vehicle intelligent controller, including: controller hardware and a control model, wherein the controller hardware includes multiple input ports, a control unit, and a machine learning processing unit; and a control module, wherein the control model includes multiple control sub-models, the multiple control sub-models are distributed and stored in the control unit and the machine learning processing unit, the control unit calculates the target signal from the input signals of the multiple input ports, the machine learning processing unit determines the control signal based on the target signal, and the control unit controls the vehicle based on the control signal.

[0006] Optionally, the multiple control sub-models include a main control sub-model, an air circuit model, an oil circuit model, and a torque model, wherein the main control sub-model is stored in the control unit, and the air circuit model, oil circuit model, and torque model are stored in the machine learning processing unit.

[0007] Optionally, the gas path model, oil path model, and torque model have the same model structure, wherein the model structure includes a local model composed of a polynomial model and corresponding weights.

[0008] Optionally, the model structure is as follows:

[0009]

[0010]

[0011] Among them, w ij θ represents the weights from input variable i to intermediate output variable j. i The bias from input variable i to intermediate output variable j. Here, a, b, and c are the weights corresponding to the local model, p and q are the coefficients of the local model, e is the exponential function, and y is the weights of the local model. k For local model output, This is the output of the first layer of the network, and also the local model y. k Input, For local y k .

[0012] Optionally, the air circuit model calculates and outputs the relative intake volume based on at least one of the following signals calculated by the control unit: engine speed, intake manifold temperature signal, intake manifold pressure, intake VVT ​​angle, exhaust VVT angle, boost pressure, coolant temperature, ambient temperature, and ambient pressure, and transmits the relative intake volume to the control unit; the fuel circuit model calculates and transmits the injection time to the control unit based on at least one of the following signals calculated by the control unit: engine speed, intake manifold temperature, intake manifold pressure, intake VVT ​​angle, exhaust VVT angle, injection angle, fuel rail pressure, and air-fuel mixture concentration, as well as the relative intake volume calculated by the air circuit model; the torque model calculates and outputs the torque based on at least one of the following signals calculated by the control unit: engine speed, ignition angle, intake VVT ​​angle, exhaust VVT angle, coolant temperature, and air-fuel mixture concentration, as well as the relative intake volume calculated by the air circuit model, and transmits the output torque to the control unit.

[0013] Optionally, the multiple input ports include multiple switch signal input ports, trigger signal input ports, pulse signal input ports, analog signal input ports, and CAN signal input ports.

[0014] Optionally, the controller hardware also includes a bus, a random access memory (RAM) unit, and a flash memory unit, wherein multiple input ports communicate with the bus, and the control unit, machine learning processing unit, RAM unit, and flash memory unit interact with the bus via signals.

[0015] Optionally, the controller hardware also includes multiple output ports, wherein the multiple output ports output control signals to the actuators of the vehicle.

[0016] Optionally, the multiple output ports include multiple switch control signal output ports, pulse signal output ports, duty cycle signal output ports, communication ports, and drive ports.

[0017] A second aspect of this application provides a vehicle including the on-board intelligent controller described above.

[0018] Therefore, this application has the following beneficial effects:

[0019] This application's embodiments collect real-time data through multiple input ports. Combined with a control unit and a machine learning processing unit with a built-in deep reinforcement learning model, it can adjust the control strategy in real time according to different operating conditions, achieving more refined and accurate vehicle control. The built-in machine learning processing unit effectively reduces computational load and time latency, easily coping with complex and ever-changing environmental challenges. Thus, it solves the technical problems of existing technologies, such as simple control methods that are difficult to adapt to complex environments, limited accuracy and efficiency, and susceptibility to prototype status.

[0020] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0021] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0022] Figure 1 This is an example diagram of an in-vehicle intelligent controller provided according to an embodiment of this application;

[0023] Figure 2 This is a schematic diagram of the controller hardware structure provided according to an embodiment of this application;

[0024] Figure 3 This is a schematic diagram illustrating the interaction and storage relationships of model information according to embodiments of this application;

[0025] Figure 4 This is a schematic diagram of the model principle provided according to the embodiments of this application;

[0026] Figure 5 A flowchart illustrating model adaptation based on embodiments of this application;

[0027] Figure 6 This is a model accuracy error distribution diagram provided according to the embodiments of this application;

[0028] Figure 7 This is a schematic diagram illustrating the model accuracy according to an embodiment of this application. Detailed Implementation

[0029] The embodiments of this application are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0030] The following description, with reference to the accompanying drawings, describes an embodiment of the vehicle-mounted intelligent controller and vehicle of this application. Addressing the problem of inaccurate vehicle control mentioned in the background section, this application provides an vehicle-mounted intelligent controller that collects real-time data through multiple input ports. Combined with a control unit and a machine learning processing unit with a built-in deep reinforcement learning model, it can adjust the control strategy in real time according to different operating conditions, achieving more refined and accurate vehicle control. The built-in machine learning processing unit effectively reduces computational load and time latency, easily coping with complex and ever-changing environmental challenges. Thus, it solves the problems of existing technologies where control methods are simple, difficult to adapt to complex environments, limited in accuracy and efficiency, and easily affected by prototype status.

[0031] Specifically, Figure 1 This is a block diagram of an embodiment of the vehicle-mounted intelligent controller of this application.

[0032] like Figure 1 As shown, the vehicle-mounted intelligent controller 10 includes: controller hardware 100 and control model 200.

[0033] In the embodiments of this application, such as Figure 2 As shown, the controller hardware 100 includes multiple input ports, a control unit 1032, and a machine learning processing unit 1033; the control model 200 includes multiple control sub-models 201, which are distributed and stored in the control unit 1032 and the machine learning processing unit 1033. The control unit 1032 calculates the target signal from the input signals of the multiple input ports, the machine learning processing unit 1033 determines the control signal 1032 based on the target signal, and the control unit 1032 controls the vehicle based on the control signal.

[0034] The multiple input ports may include multiple switch signal input ports 1011, trigger signal input ports 1012, pulse signal input ports 1013, analog signal input ports 1014, and CAN signal input ports 1015.

[0035] It is understood that, in this embodiment, the controller hardware 100 integrates multiple input ports, a control unit 1032, and a machine learning processing unit 1033 to achieve comprehensive acquisition and intelligent processing of real-time vehicle data. The control unit 1032 calculates and analyzes real-time data from vehicle sensors to generate target signals; while the machine learning processing unit 1033 uses deep reinforcement learning algorithms to optimize the target signals online, quickly determining the optimal control signal, which not only improves the accuracy of the control strategy but also significantly enhances its adaptability to complex environmental changes. Furthermore, the distributed storage design of the control model 200 deploys multiple sub-models in the control unit 1032 and the machine learning processing unit 1033 respectively, fully utilizing the performance advantages of each hardware component and improving flexibility and processing efficiency.

[0036] In the embodiments of this application, such as Figure 2 As shown, the controller hardware 100 includes a bus 1031, a random access memory unit 1034, and a flash memory unit 1035. Multiple input ports communicate with the bus 1031, and the control unit 1032, the machine learning processing unit 1033, the random access memory unit 1034, and the flash memory unit 1035 interact with the bus 1031 via signals.

[0037] It is understood that the embodiments of this application support complex data analysis and machine learning applications through efficient data transmission and processing mechanisms, ensuring stable operation and providing reliable performance in various scenarios.

[0038] In the embodiments of this application, such as Figure 2 As shown, the controller hardware 100 also includes multiple output ports, wherein the multiple output ports output control signals to the actuators of the vehicle.

[0039] In the embodiments of this application, such as Figure 2 As shown, the multiple output ports include multiple switch control signal output ports 1051, pulse signal output ports 1052, duty cycle signal output ports 1053, communication ports 1054, and drive ports 1055.

[0040] It is understood that the multiple output ports in this application embodiment include a switch control signal output port 1051, a pulse signal output port 1052, a duty cycle signal output port 1053, a communication port 1054, and a drive port 1055, which are used to realize the switch control, precise time control, state adjustment, data exchange, and high-power drive of external devices, and together constitute an important part of the controller hardware 100, supporting diverse control tasks and application scenarios.

[0041] In the embodiments of this application, such as Figure 3As shown, multiple control sub-models include a main control sub-model 201, an air circuit model 202, an oil circuit model 203, and a torque model 204. The main control sub-model 201 is stored in the control unit 1032, and the air circuit model 202, the oil circuit model 203, and the torque model 204 are stored in the machine learning processing unit 1033.

[0042] Understandably, in this embodiment, the main control sub-model 201 is stored in the control unit 1032, responsible for overall coordination and control strategy formulation; while the air path model 202, fuel path model 203, and torque model 204 are stored in the machine learning processing unit 1033, focusing on simulating complex processes such as engine intake, fuel injection, and predicting output torque, respectively. This distributed storage method fully utilizes the performance advantages of the control unit 1032 and the machine learning processing unit 1033, achieving efficient and precise management of engine control, thereby improving performance and optimizing combustion efficiency.

[0043] In the embodiments of this application, such as Figure 4 As shown, the air circuit model 202, oil circuit model 203 and torque model 204 have the same model structure.

[0044] The model structure may include local models composed of polynomial models and corresponding weights.

[0045] It is understood that the air path model 202, oil path model 203 and torque model 204 in the embodiments of this application all adopt a combination of polynomial models and their local models and corresponding weights, which facilitates the subsequent calculation of relative intake volume, injection time and output torque.

[0046] In this embodiment of the application, the model structure is as follows:

[0047]

[0048]

[0049] Among them, w ij θ represents the weights from input variable i to intermediate output variable j. i The bias from input variable i to intermediate output variable j. Here, a, b, and c are the weights corresponding to the local model, p and q are the coefficients of the local model, e is the exponential function, and y is the weights of the local model. k For local model output, This is the output of the first layer of the network, and also the local model y. k Input, For local y k .

[0050] In this embodiment, the air path model 202 calculates and outputs the relative intake volume based on at least one signal calculated by the control unit 1032, including engine speed, intake manifold temperature signal, intake manifold pressure, intake VVT ​​angle, exhaust VVT angle, boost pressure, water temperature, ambient temperature, and ambient pressure, and transmits the relative intake volume to the control unit 1032; the fuel path model 203 calculates the injection time based on at least one signal calculated by the control unit 1032, including engine speed, intake manifold temperature, intake manifold pressure, intake VVT ​​angle, exhaust VVT angle, injection angle, fuel rail pressure, and air-fuel mixture concentration, and the relative intake volume calculated by the air path model 202, and transmits the injection time to the control unit 1032; the torque model 204 calculates and outputs the torque based on at least one signal calculated by the control unit 1032, including engine speed, ignition angle, intake VVT ​​angle, exhaust VVT angle, water temperature, and air-fuel mixture concentration, and the relative intake volume calculated by the air path model 202, and transmits the output torque to the control unit 1032.

[0051] It is understood that in this embodiment, the air path model 202 is responsible for evaluating the intake state and calculating the relative intake volume, providing key inputs for other models; the fuel path model 203 optimizes the fuel injection strategy based on these inputs and other signals to ensure that the engine can obtain the best fuel economy and emission performance under different operating conditions; and the torque model 204 predicts and evaluates the engine's output torque in real time, providing an important reference for the formulation of control strategies.

[0052] Specifically, the airflow model 202 obtains signals such as engine speed n, intake manifold temperature tim, intake manifold pressure pim, intake VVT ​​angle vin, exhaust VVT angle vex, boost pressure pbt, coolant temperature two, ambient temperature tair, and ambient pressure pair from the control unit 1032, calculates and outputs the relative intake volume rl, and transmits it to the control unit 1032. The airflow model 202 is a black box model.

[0053] rl=f(pim,tim,n,vin,vex,pbt,two,tair,pair)

[0054] The fuel system model 203 obtains signals such as engine speed n, intake manifold temperature tim, intake manifold pressure pim, intake VVT ​​angle vin, exhaust VVT angle vex, injection angle iao, fuel rail pressure fup, and air-fuel mixture concentration λ from the control unit 1032. It also obtains the relative intake air volume rl from the air system model 202, calculates and outputs the injection time ti, and transmits it to the control unit 1032. The fuel system model 203 is a black box model.

[0055] ti=f(rl,λ,pim,tim,n,vin,vex,iao,fup)

[0056] Torque model 204 obtains signals such as engine speed n, ignition angle iga, intake VVT ​​angle vin, exhaust VVT angle vex, water temperature two, and air-fuel mixture concentration λ from control unit 1032, obtains relative intake volume rl from air path model 202, calculates the output torque TORQ, and transmits it to control unit 1032. Torque model 204 is a black box model.

[0057] TORQ=f(n,rl,λ,iga,vin,vex,two).

[0058] It should be noted that when the prototype exhibits significant variance or ages, leading to a deterioration in model fidelity, the prototype robustness adaptive function can correct these deviations. The self-learning mode includes forced self-learning and range adaptation. Significant prototype variance or aging is primarily determined by the mixture correction coefficient in control unit 1032. When the correction coefficient exceeds a certain set limit for a specified period, range adaptation is triggered. For example... Figure 5 As shown, adaptive testing begins, triggering the test program. This program executes the built-in tests in module 1033 and calculates the accuracy based on the test data. If the model accuracy meets the requirements, the adaptive function ends. If the model accuracy does not meet the requirements, the model parameters are adjusted, and the test is repeated to assess model accuracy again until the accuracy meets the requirements. Alternatively, the adaptive function can be forced according to user needs.

[0059] The following will provide a detailed explanation of the air circuit model 202, the oil circuit model 203, and the torque model 204.

[0060] When gas path model 202 is offline:

[0061] S1. Conduct engine tests on an engine test bench, measuring steady-state data under different speeds, loads, intake VVT, and exhaust VVT. The operating conditions must cover the entire working space of the engine. Measured variables include engine speed n, intake manifold temperature tim, intake manifold pressure pim, intake VVT ​​angle vin, exhaust VVT angle vex, boost pressure pbt, coolant temperature two, ambient temperature pair, ambient pressure pair, relative intake volume rl, etc. The data is stored in local test equipment.

[0062] It should be noted that typical operating conditions can be speeds of 1000, 1500, 2000, 2500, 3000, 3500, 4000, 4500, 5000, 5500, and 6000 rpm, and loads can be defined using relative intake volume: 20, 25, 30, 40, 50, 60, 80, 100, 120, 150, and 180 degrees. The absolute angles of the intake VVT ​​are 0, 10, 20, 30, 40, and 50 degrees; the absolute angles of the exhaust VVT are 0, 10, 20, 30, 40, and 50 degrees. The specific values ​​of speed, load, and VVT are defined according to the actual range of the engine and must cover the entire operating range.

[0063] S2. Perform offline processing on the data collected by S1 on the local device to remove abnormal data, such as null values, non-numerical values, and boundary condition control anomalies, such as abnormal temperature and abnormal combustion stability values, to ensure data quality.

[0064] S3. Scale the data processed in S2 on the local device to map all data to the same scale. Scale to 0-1, ensuring that each feature data has the same impact on the output. The scaling method is as follows:

[0065]

[0066] in, The input is scaled down, x is the original input, a0 is a value 10% smaller than the minimum value of the input data, and a1 is a value 10% larger than the maximum value of the input data.

[0067] S4. Perform a second transformation on the data after the changes in S3 on the local device to make the data closer to a normal distribution. Transformation method:

[0068]

[0069] in, This is the result of a quadratic change, where λ is the coefficient of change, for example, λ = 0.5.

[0070] S5. Optimize the core black-box model parameters on the local device, using a stochastic gradient descent algorithm to iteratively optimize the parameters until the model output rl and the experimental test rl accuracy meet the requirement that the deviation of 95% of the test points is within ±5%. Figure 6 He Ru Figure 7 As shown, the model structure is as follows:

[0071]

[0072]

[0073] Among them, w ijθ represents the weights from input variable i to intermediate output variable j. i The bias from input variable i to intermediate output variable j. Here, a, b, and c are the weights corresponding to the local model, p and q are the coefficients of the local model, e is the exponential function, and y is the weights of the local model. k For local model output, This is the output of the first layer of the network, and also the local model y. k Input, For local y k .

[0074] S6. Integrate the model trained by S5 into a dedicated machine learning computing unit to interact with the main torque structure model and calculate the relative intake volume in real time.

[0075] It should be noted that another algorithm is as follows:

[0076]

[0077] Where, σ m σ represents the length scale of each variable in the multidimensional space. f It is the signal variance of the kernel function, used to control the degree of local correlation.

[0078] When gas path model 202 is online:

[0079] S1. Conduct engine tests on an engine test bench, measuring steady-state data under different speeds, loads, intake VVTs, and exhaust VVTs. The operating conditions must cover the entire working space of the engine. Measured variables include engine speed n, intake manifold temperature tim, intake manifold pressure pim, intake VVT ​​angle vin, exhaust VVT angle vex, boost pressure pbt, coolant temperature two, ambient temperature tair, ambient pressure pair, relative intake volume rl, etc. The data is stored in real time in the flash memory unit 1035.

[0080] It should be noted that typical operating conditions can be speeds of 1000, 1500, 2000, 2500, 3000, 3500, 4000, 4500, 5000, 5500, and 6000 rpm, and loads can be defined using relative intake volume: 20, 25, 30, 40, 50, 60, 80, 100, 120, 150, and 180 degrees. The absolute angles of the intake VVT ​​are 0, 10, 20, 30, 40, and 50 degrees; the absolute angles of the exhaust VVT are 0, 10, 20, 30, 40, and 50 degrees. The specific values ​​of speed, load, and VVT are defined according to the actual range of the engine and must cover the entire operating range.

[0081] S2, the machine learning processing unit 1033 obtains the data collected by S1 in real time from the flash memory unit 1035, performs cleaning processing, removes abnormal data, such as null values, non-numerical values, and boundary condition control abnormal values, such as temperature abnormalities and combustion stability abnormal values, to ensure data quality.

[0082] S3, the machine learning processing unit 1033 scales the data processed by S2, mapping all data to the same scale. The scaling is between 0 and 1, ensuring that each feature data has the same impact on the output. The scaling method is as follows:

[0083]

[0084] in, The input is scaled down, x is the original input, a0 is a value 10% smaller than the minimum value of the input data, and a1 is a value 10% larger than the maximum value of the input data.

[0085] S4, the machine learning processing unit 1033 performs a secondary transformation on the data after the transformation in S3 to make the data closer to a normal distribution. The transformation method is as follows:

[0086]

[0087] in, This is the result of a quadratic change, where λ is the coefficient of change, for example, λ = 0.5.

[0088] S5, the machine learning processing unit 1033, optimizes the core black-box model parameters using a stochastic gradient algorithm, iteratively optimizing the parameters until the model output rl and the experimental test rl accuracy meet the requirement that the deviation of 95% of the test points is within ±5%. Figure 6 He Ru Figure 7 As shown, the model structure is as follows:

[0089]

[0090] Among them, w ij θ represents the weights from input variable i to intermediate output variable j. i The bias from input variable i to intermediate output variable j. Here, a, b, and c are the weights corresponding to the local model, p and q are the coefficients of the local model, e is the exponential function, and y is the weights of the local model. k For local model output, This is the output of the first layer of the network, and also the local model y. k Input, For local y k .

[0091] S6. Solidify the optimized model parameters of S5 and store them in the machine learning processing unit 1033. Interact with the main torque structure model to calculate the relative intake volume in real time.

[0092] When the oil circuit model 203 is offline:

[0093] S1. Conduct engine tests on an engine test bench, measuring steady-state data under different speeds, loads, intake VVTs, and exhaust VVTs. The operating conditions must cover the entire working space of the engine. Measured variables include speed n, relative intake volume rl, intake manifold temperature tim, intake manifold pressure pim, intake VVT ​​angle vin, exhaust VVT angle vex, injection angle iao, fuel rail pressure fup, and mixture concentration λ, etc. The data is stored in local test equipment.

[0094] It should be noted that typical operating conditions can be speeds of 1000, 1500, 2000, 2500, 3000, 3500, 4000, 4500, 5000, 5500, and 6000 rpm. Loads can be defined using relative intake air volume: 20, 25, 30, 40, 50, 60, 80, 100, 120, 150, and 180. Intake and exhaust VVT angles are set for both intake and exhaust. Injection angles are 250 to 350 degrees, with intervals of 10 degrees. Rail pressures are 15 to 35 degrees, with intervals of 5 degrees. Specific values ​​are defined based on the actual conditions of the engine and must cover the entire operating range.

[0095] S2. Perform offline processing on the data collected by S1 on the local device to remove abnormal data, such as null values, non-numerical values, and boundary condition control anomalies, such as abnormal temperature and abnormal combustion stability values, to ensure data quality.

[0096] S3. Scale the data processed in S2 on the local device to map all data to the same scale. Scale to 0-1, ensuring that each feature data has the same impact on the output. The scaling method is as follows:

[0097]

[0098] in, The input is scaled down, x is the original input, a0 is a value 10% smaller than the minimum value of the input data, and a1 is a value 10% larger than the maximum value of the input data.

[0099] S4. Perform a second transformation on the data after the changes in S3 on the local device to make the data closer to a normal distribution. The transformation method is as follows:

[0100]

[0101] in, This is the result of a quadratic change, where λ is the coefficient of change, for example, λ = 0.5.

[0102] S5. Optimize the core black-box model parameters on the local device, using a stochastic gradient descent algorithm to iteratively optimize the parameters until the model output rl and the experimental test rl accuracy meet the requirement that the deviation of 95% of the test points is within ±5%. Figure 6 He Ru Figure 7 As shown, the model structure is as follows:

[0103]

[0104] Among them, w ij θ represents the weights from input variable i to intermediate output variable j. i The bias from input variable i to intermediate output variable j. Here, a, b, and c are the weights corresponding to the local model, p and q are the coefficients of the local model, e is the exponential function, and y is the weights of the local model. k For local model output, This is the output of the first layer of the network, and also the local model y. k Input, For local y k .

[0105] S6. Integrate the model trained by S5 into a dedicated machine learning computing unit to interact with the main torque structure model and calculate the relative intake volume in real time.

[0106] It should be noted that another algorithm is as follows:

[0107]

[0108] Where, σ m σ represents the length scale of each variable in the multidimensional space. f It is the signal variance of the kernel function, used to control the degree of local correlation.

[0109] When the oil circuit model 203 is online:

[0110] S1. Conduct engine tests on an engine test bench, measuring steady-state data under different speeds, loads, intake VVTs, and exhaust VVTs. The operating conditions must cover the entire working space of the engine. Measured variables include speed n, relative intake volume rl, intake manifold temperature tim, intake manifold pressure pim, intake VVT ​​angle vin, exhaust VVT angle vex, injection angle iao, fuel rail pressure fup, and mixture concentration λ, etc. The data is stored in real time in the flash memory unit 1035.

[0111] It should be noted that typical operating conditions can be speeds of 1000, 1500, 2000, 2500, 3000, 3500, 4000, 4500, 5000, 5500, and 6000 rpm. Loads can be defined using relative intake air volume: 20, 25, 30, 40, 50, 60, 80, 100, 120, 150, and 180. Intake and exhaust VVT angles are set for both intake and exhaust. Injection angles are 250 to 350 degrees, with intervals of 10 degrees. Rail pressures are 15 to 35 degrees, with intervals of 5 degrees. Specific values ​​are defined based on the actual conditions of the engine and must cover the entire operating range.

[0112] S2, the machine learning processing unit 1033 obtains the data collected by S1 in real time from the flash memory unit 1035, performs cleaning processing, removes abnormal data, such as null values, non-numerical values, and boundary condition control abnormal values, such as temperature abnormalities and combustion stability abnormal values, to ensure data quality.

[0113] S3, the machine learning processing unit 1033 scales the data processed by S2, mapping all data to the same scale. The scaling is between 0 and 1, ensuring that each feature data has the same impact on the output. The scaling method is as follows:

[0114]

[0115] in, The input is scaled down, x is the original input, a0 is a value 10% smaller than the minimum value of the input data, and a1 is a value 10% larger than the maximum value of the input data.

[0116] S4, the machine learning processing unit 1033 performs a secondary transformation on the data after the transformation in S3, making the data closer to a normal distribution. The transformation method is as follows:

[0117]

[0118] in, This is the result of a quadratic change, where λ is the coefficient of change, for example, λ = 0.5.

[0119] S5, the machine learning processing unit 1033, optimizes the core black-box model parameters using a stochastic gradient algorithm, iteratively optimizing the parameters until the model output rl and the experimental test rl accuracy meet the requirement that the deviation of 95% of the test points is within ±5%. Figure 6 He Ru Figure 7 As shown, the model structure is as follows:

[0120]

[0121] Among them, w ij θ represents the weights from input variable i to intermediate output variable j.i The bias from input variable i to intermediate output variable j. Here, a, b, and c are the weights corresponding to the local model, p and q are the coefficients of the local model, e is the exponential function, and y is the weights of the local model. k For local model output, This is the output of the first layer of the network, and also the local model y. k Input, For local y k .

[0122] S6. Solidify the optimized model parameters of S5 and store them in the machine learning processing unit 1033. Interact with the main torque structure model to calculate the relative intake volume in real time.

[0123] When torque model 204 is offline:

[0124] S1. Conduct engine tests on an engine test bench, measuring steady-state data under different speeds, loads, intake VVT, and exhaust VVT. The operating conditions must cover the entire working space of the engine. Measured variables include speed n, ignition angle iga, intake VVT ​​angle vin, exhaust VVT angle vex, coolant temperature two, and air-fuel mixture concentration λ. The data is stored in local test equipment.

[0125] It should be noted that typical operating conditions can be speeds of 1000, 1500, 2000, 2500, 3000, 3500, 4000, 4500, 5000, 5500, and 6000 rpm. Loads can be defined using relative intake air volume: 20, 25, 30, 40, 50, 60, 80, 100, 120, 150, and 180 degrees. Intake and exhaust VVT angle settings include using only intake, using only exhaust, not using either intake or exhaust, and using both. Ignition angle changes are 0, -3, -6, -9, -12, and -15 degrees. Specific values ​​are defined according to the actual conditions of the engine and must cover the entire operating range.

[0126] S2. Perform offline processing on the data collected by S1 on the local device to remove abnormal data, such as null values, non-numerical values, and boundary condition control anomalies, such as abnormal temperature and abnormal combustion stability values, to ensure data quality.

[0127] S3. Scale the data processed in S2 on the local device to map all data to the same scale. Scale to 0-1, ensuring that each feature data has the same impact on the output. The scaling method is as follows:

[0128]

[0129] in, The input is scaled down, x is the original input, a0 is a value 10% smaller than the minimum value of the input data, and a1 is a value 10% larger than the maximum value of the input data.

[0130] S4. Perform a second transformation on the data after the changes in S3 on the local device to make the data closer to a normal distribution. Transformation method:

[0131]

[0132] in, This is the result of a quadratic change, where λ is the coefficient of change, for example, λ = 0.5.

[0133] S5. Optimize the core black-box model parameters on the local device, using a stochastic gradient descent algorithm to iteratively optimize the parameters until the model output rl and the experimental test rl accuracy meet the requirement that the deviation of 95% of the test points is within ±5%. Figure 6 He Ru Figure 7 As shown, the model structure is as follows:

[0134]

[0135]

[0136] Among them, w ij θ represents the weights from input variable i to intermediate output variable j. i The bias from input variable i to intermediate output variable j. Here, a, b, and c are the weights corresponding to the local model, p and q are the coefficients of the local model, e is the exponential function, and y is the weights of the local model. k For local model output, This is the output of the first layer of the network, and also the local model y. k Input, For local y k .

[0137] S6. Integrate the model trained by S5 into a dedicated machine learning computing unit to interact with the main torque structure model and calculate the relative intake volume in real time.

[0138] It should be noted that another algorithm is as follows:

[0139]

[0140] Where, σ m σ represents the length scale of each variable in the multidimensional space. f It is the signal variance of the kernel function, used to control the degree of local correlation.

[0141] When torque model 204 is online:

[0142] S1. Conduct engine tests on the engine test bench and measure steady-state data under different speeds, loads, intake VVTs, and exhaust VVTs. The operating conditions need to cover the entire working space of the engine. The measured variables include speed n, ignition angle iga, intake VVT ​​angle vin, exhaust VVT angle vex, water temperature two, and mixture concentration λ. The data is stored in real time in the flash memory unit 1035.

[0143] It should be noted that typical operating conditions can be speeds of 1000, 1500, 2000, 2500, 3000, 3500, 4000, 4500, 5000, 5500, and 6000 rpm. Loads can be defined using relative intake air volume: 20, 25, 30, 40, 50, 60, 80, 100, 120, 150, and 180 degrees. Intake and exhaust VVT angle settings include using only intake, using only exhaust, not using either intake or exhaust, and using both. Ignition angle changes are 0, -3, -6, -9, -12, and -15 degrees. Specific values ​​are defined according to the actual conditions of the engine and must cover the entire operating range.

[0144] S2, the machine learning processing unit 1033 obtains the data collected by S1 in real time from the flash memory unit 1035, performs cleaning processing, removes abnormal data, such as null values, non-numerical values, and boundary condition control abnormal values, such as temperature abnormalities and combustion stability abnormal values, to ensure data quality.

[0145] S3, the machine learning processing unit 1033, scales the data processed by S2, mapping all data to the same scale. Scale is taken from 0 to 1, ensuring that each feature data has the same impact on the output. The scaling method is as follows:

[0146]

[0147] in, The input is scaled down, x is the original input, a0 is a value 10% smaller than the minimum value of the input data, and a1 is a value 10% larger than the maximum value of the input data.

[0148] S4, the machine learning processing unit 1033 performs a secondary transformation on the data after the transformation in S3, making the data closer to a normal distribution. The transformation method is as follows:

[0149]

[0150] in, This is the result of a quadratic change, where λ is the coefficient of change, for example, λ = 0.5.

[0151] S5, the machine learning processing unit 1033, optimizes the core black-box model parameters using a stochastic gradient algorithm, iteratively optimizing the parameters until the model output rl and the experimental test rl accuracy meet the requirement that the deviation of 95% of the test points is within ±5%. Figure 6 He Ru Figure 7 As shown, the model structure is as follows:

[0152]

[0153] Among them, w ij θ represents the weights from input variable i to intermediate output variable j. i The bias from input variable i to intermediate output variable j. Here, a, b, and c are the weights corresponding to the local model, p and q are the coefficients of the local model, e is the exponential function, and y is the weights of the local model. k For local model output, This is the output of the first layer of the network, and also the local model y. k Input, For local y k .

[0154] S6. Solidify the optimized model parameters of S5 and store them in the machine learning processing unit 1033. Interact with the main torque structure model to calculate the relative intake volume in real time.

[0155] The vehicle-mounted intelligent controller proposed in this application collects real-time data through multiple input ports. Combined with a control unit and a machine learning processing unit with a built-in deep reinforcement learning model, it can adjust the control strategy in real time according to different operating conditions, achieving more refined and accurate vehicle control. The built-in machine learning processing unit effectively reduces computational load and time latency, easily coping with complex and ever-changing environmental challenges. This solves the problems of existing technologies, such as simple control methods that are difficult to adapt to complex environments, limited accuracy and efficiency, and susceptibility to prototype status.

[0156] This application also provides a vehicle, including the above-described on-board intelligent controller.

[0157] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0158] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0159] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or N executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0160] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any of the following techniques known in the art, or a combination thereof: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (FPGAs), field-programmable gate arrays (FPGAs), etc.

[0161] Those skilled in the art will understand that all or part of the steps of the methods implementing the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0162] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A vehicle-mounted intelligent controller, characterized in that, include: Controller hardware and control model, among which, The controller hardware includes multiple input ports, a control unit, and a machine learning processing unit; The control model includes multiple control sub-models, which are distributed and stored in the control unit and the machine learning processing unit. The control unit calculates the target signal from the input signals of the multiple input ports, the machine learning processing unit determines the control signal based on the target signal, and the control unit controls the vehicle based on the control signal. The multiple control sub-models include a main control sub-model, an air circuit model, an oil circuit model, and a torque model, wherein the main control sub-model is stored in the control unit, and the air circuit model, the oil circuit model, and the torque model are stored in the machine learning processing unit; The gas path model, the oil path model, and the torque model have the same model structure, wherein the model structure includes a local model composed of a polynomial model and corresponding weights; The model structure is as follows: in, For input variable i to the kth output variable The weight, For input variable i to the kth output variable The bias, For the weights corresponding to the local model, All are local model coefficients. All are coefficients of the weighting function. It is an exponential function with the natural constant e as its base. Output for local model.

2. The vehicle-mounted intelligent controller according to claim 1, characterized in that, The air path model calculates and outputs the relative intake volume based on at least one signal calculated by the control unit, including the engine speed, intake manifold temperature signal, intake manifold pressure, intake VVT ​​angle, exhaust VVT angle, boost pressure, water temperature, ambient temperature, and ambient pressure, and transmits the relative intake volume to the control unit. The fuel circuit model calculates the injection time based on at least one signal calculated by the control unit, including the engine speed, intake manifold temperature, intake manifold pressure, intake VVT ​​angle, exhaust VVT angle, injection angle, fuel rail pressure, and air-fuel mixture concentration, as well as the relative intake volume calculated by the fuel circuit model, and transmits the injection time to the control unit. The torque model calculates the output torque based on at least one signal from the control unit, including the engine speed, ignition angle, intake VVT ​​angle, exhaust VVT angle, water temperature, and air-fuel mixture concentration, as well as the relative intake volume calculated by the air path model, and transmits the output torque to the control unit.

3. The vehicle-mounted intelligent controller according to claim 1, characterized in that, The multiple input ports include multiple switch signal input ports, trigger signal input ports, pulse signal input ports, analog signal input ports, and CAN signal input ports.

4. The vehicle-mounted intelligent controller according to claim 1, characterized in that, The controller hardware also includes a bus, a random access memory (RAM) unit, and a flash memory unit. The plurality of input ports communicate with the bus, and the control unit, the machine learning processing unit, the RAM unit, and the flash memory unit interact with the bus via signals.

5. The vehicle-mounted intelligent controller according to claim 1, characterized in that, The controller hardware also includes multiple output ports, wherein the multiple output ports output the control signals to the actuators of the vehicle.

6. The vehicle-mounted intelligent controller according to claim 5, characterized in that, The multiple output ports include multiple switch control signal output ports, pulse signal output ports, duty cycle signal output ports, communication ports, and drive ports.

7. A vehicle, characterized in that, Includes the vehicle-mounted intelligent controller as described in any one of claims 1-6.

Images