Carbon loading determination method, apparatus, computer-readable storage medium, and electronic device

Abstract

This invention provides a method, apparatus, computer-readable storage medium, and electronic device for determining carbon loading. Different carbon loading determination models can be used under different conditions to determine the carbon loading, and then the final carbon loading is determined. Since the pre-trained passive regeneration carbon loading determination model is specifically designed to determine the carbon loading of the particulate trap in a passive regeneration state, its accuracy is higher.

Description

Technical Field

[0001] This invention relates to the field of vehicle emission technology, and in particular to a method, apparatus, computer-readable storage medium, and electronic device for determining carbon load. Background Technology

[0002] A particulate filter (DPF) is a filter element that filters particulate matter. It collects all impurities such as black smoke produced by the engine, thereby reducing particulate emissions.

[0003] The carbon loading model based on differential pressure sensors calculates the accumulated carbon loading in the DPF based on the differential pressure sensor readings across the DPF. When the accumulated carbon loading reaches a certain value, the system triggers regenerative combustion to remove the accumulated carbon particles from the DPF.

[0004] However, the carbon loading calculations based on current differential pressure sensor-based carbon loading models have a large error. Summary of the Invention

[0005] The purpose of this invention is to provide a method, apparatus, computer-readable storage medium, and electronic device for determining carbon loading, in order to solve the problem of large errors in carbon loading. The specific technical solution is as follows:

[0006] A method for determining carbon loading includes:

[0007] Determine the current state of the particle trap;

[0008] If the particle trap is in a passive regeneration state and the duration of the passive regeneration state exceeds a preset duration, then a pre-trained passive regeneration carbon loading determination model is used to determine at least one first carbon loading.

[0009] If the particulate trap is in a carbonized state, at least one second carbon load is determined using a pre-trained carbon loading determination model.

[0010] The third carbon loading is determined based on the at least one first carbon loading and the at least one second carbon loading.

[0011] Optionally, determining the current state of the particle trap includes:

[0012] The exhaust gas volumetric flow rate and the upstream exhaust temperature of the particulate filter are obtained.

[0013] If the exhaust gas volume flow rate is within a preset flow range and the upstream exhaust temperature is within a preset temperature range, then the current state of the particulate filter is determined to be a passive regeneration state.

[0014] Optionally, determining the third carbon loading based on the at least one first carbon loading and the at least one second carbon loading includes:

[0015] Wavelet denoising is performed on the carbon load sequence consisting of at least one first carbon load and at least one second carbon load, and sliding flat filtering is performed on the carbon load sequence that has undergone wavelet denoising.

[0016] The third carbon load is determined from the carbon load sequence that has undergone the sliding flat filtering process.

[0017] Optionally, the inputs to the pre-trained passive regenerated carbon loading determination model are the exhaust gas volumetric flow rate and the upstream and downstream pressure difference of the particulate filter, and the inputs to the pre-trained carbon deposition carbon loading determination model are the exhaust gas volumetric flow rate and the upstream and downstream pressure difference of the particulate filter.

[0018] Optional,

[0019] The training process of the passive regeneration carbon loading determination model includes: obtaining a first initial model and first training data, wherein the first training data includes the exhaust gas volumetric flow rate under the passive regeneration state and the upstream and downstream pressure difference of the particulate filter under the passive regeneration state, and the first initial model is a nonlinear autoregressive neural network (NARX) model with external input; training the first initial model using the first training data, wherein during the training process, the momentum-adaptive gradient descent method is used to adjust the parameters of the network structure of the first initial model, wherein the parameters include: the threshold of the hidden layer, the weight of the hidden layer, the threshold of the output layer, and the weight of the output layer;

[0020] And / or,

[0021] The training process of the carbon loading determination model includes: obtaining a second initial model and second training data, wherein the second training data includes the exhaust gas volume flow rate under carbon deposition conditions and the upstream and downstream pressure difference of the particulate filter under carbon deposition conditions; the second initial model is a nonlinear autoregressive neural network (NARX) model with external input; training the second initial model using the second training data; during the training process, adjusting the network structure parameters of the second initial model using the momentum-adaptive gradient descent method, wherein the parameters include: the threshold of the hidden layer, the weight of the hidden layer, the threshold of the output layer, and the weight of the output layer.

[0022] A carbon loading determination device, comprising:

[0023] The state determination unit is used to determine the current state of the particle collector;

[0024] The first determining unit is used to determine at least one first carbon load using a pre-trained passive regeneration carbon load determination model if the particle trap is in a passive regeneration state and the duration of the passive regeneration state exceeds a preset duration.

[0025] The second determining unit is used to determine at least one second carbon load using a pre-trained carbon load determination model if the particulate trap is in a carbon deposit state.

[0026] The third determining unit is used to determine the third carbon loading based on the at least one first carbon loading and the at least one second carbon loading.

[0027] Optionally, the state determination unit is specifically used for:

[0028] Obtain the exhaust gas volumetric flow rate and the upstream exhaust temperature of the particulate filter; if the exhaust gas volumetric flow rate is within a preset flow range and the upstream exhaust temperature is within a preset temperature range, then determine that the current state of the particulate filter is a passive regeneration state.

[0029] Optionally, the third determining unit is specifically used to: perform wavelet denoising on the carbon load sequence composed of the at least one first carbon load and the at least one second carbon load; perform sliding flat filtering on the carbon load sequence that has undergone wavelet denoising; and determine a third carbon load from the carbon load sequence that has undergone sliding flat filtering.

[0030] A computer-readable storage medium storing a program that, when executed by a processor, implements any of the above-described methods for determining carbon loading.

[0031] An electronic device includes at least one processor, at least one memory and a bus connected to the processor; wherein the processor and the memory communicate with each other via the bus; the processor is used to call program instructions in the memory to execute any of the carbon loading determination methods described above.

[0032] This invention provides a carbon loading determination method, apparatus, computer-readable storage medium, and electronic device. Different carbon loading determination models can be used in different states to determine the carbon loading, and then the final carbon loading is determined. Since the pre-trained passive regeneration carbon loading determination model is specifically designed to determine the carbon loading of a particulate trap in a passive regeneration state, its accuracy is higher. Furthermore, by limiting the duration of continuous passive regeneration to a preset time, this application avoids frequent switching of the carbon loading determination model. Frequent switching can lead to frequent fluctuations in the determined carbon loading, resulting in poor stability and affecting its accuracy.

[0033] Of course, any product or method implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0034] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0035] Figure 1 A flowchart of a method for determining carbon loading provided in an embodiment of the present invention;

[0036] Figure 2 This is a schematic diagram of a carbon loading determination device provided in an embodiment of the present invention;

[0037] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0038] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0039] The inventors of this application have discovered that the main reason for the large error in carbon loading calculations based on differential pressure sensors is that when the DPF temperature exceeds 320℃, the pyrolysis rate of soluble organic matter (SOF) attached to the carbon particles inside the DPF accelerates, and passive regeneration begins. Microscopically, NO2 preferentially reacts with the surface of the carbon particles, resulting in a rapid decrease in flow resistance and a smaller differential pressure. However, the carbon loading does not decrease significantly, thus reducing the accuracy of carbon loading estimations calculated by the differential pressure sensor-based carbon loading model.

[0040] like Figure 1 As shown, this embodiment of the invention provides a method for determining carbon loading, which may include:

[0041] S100, Determine the current state of the particle collector;

[0042] Optionally, the particulate filter can be in several states, such as passive regeneration, carbon buildup, and active regeneration.

[0043] Among them, passive regeneration state refers to the state in which the particle trap is undergoing passive regeneration.

[0044] When the engine operates in the high-speed, high-load range, the exhaust temperature will be relatively high. The NO in the exhaust gas will be oxidized into NO2 after passing through soluble organic carbon (DOC). The NO2 reacts with the captured carbon particles inside the DPF to generate gaseous pollutants that are discharged – this is passive regeneration.

[0045] When the vehicle's ECU calculates or detects that the DPF is full of carbon, it will automatically trigger the on-road regeneration function. At this time, the particulate filter is in active regeneration mode. In active regeneration mode, the vehicle injects diesel fuel through the engine's rear injection or the seventh fuel injector, causing the soot to react with O2 at high temperature (above 500°C), which generally occurs cyclically.

[0046] Optionally, step S100 of this application may specifically include:

[0047] Obtain the exhaust gas volumetric flow rate and the upstream exhaust temperature of the particulate filter;

[0048] If the exhaust gas volume flow rate is within the preset flow range and the upstream exhaust temperature is within the preset temperature range, then the current state of the particulate filter is determined to be passive regeneration state.

[0049] Optionally, the preset flow rate range can be 300kg / h to 1000kg / h, or other ranges such as 350kg / h to 950kg / h.

[0050] Optionally, the preset temperature range can be 300 to 400 degrees, or other ranges such as 230 to 350 degrees.

[0051] This application can identify whether the particulate filter is in active regeneration mode by using active regeneration commands issued by the ECU.

[0052] Optionally, if the particulate filter is not in passive or active regeneration mode, it can be determined that the particulate filter is in a carbon buildup state.

[0053] S200. If the particle trap is in a passive regeneration state and the duration of the passive regeneration state exceeds a preset duration, then at least one first carbon load is determined using a pre-trained passive regeneration carbon load determination model.

[0054] By limiting the duration of the continuous passive regeneration state to a preset duration, this application can avoid frequent switching of the carbon loading determination model. Frequent switching may cause frequent fluctuations in the determined carbon loading, resulting in poor stability and affecting its accuracy.

[0055] Optionally, the preset duration can be 1 minute or other durations.

[0056] The pre-trained passive regeneration carbon loading determination model is specifically designed to determine the carbon loading of a particulate filter in passive regeneration mode. Specifically, the inputs to the pre-trained passive regeneration carbon loading determination model are the exhaust gas volumetric flow rate and the pressure difference between the upstream and downstream sides of the particulate filter.

[0057] Optionally, the pre-trained passive regeneration carbon loading determination model can be trained using the exhaust gas volumetric flow rate and the upstream and downstream pressure difference of the particulate filter under passive regeneration conditions.

[0058] The model for determining passive regenerated carbon loading can be a supervised model, a semi-supervised model, or an unsupervised model.

[0059] If the passive regeneration carbon loading determination model is a supervised or semi-supervised model, the training data also includes the carbon loading of the particulate trap under passive regeneration conditions.

[0060] Understandably, over time, passive regenerated carbon loading determination models can identify multiple primary carbon loadings.

[0061] S300. If the particulate trap is in a carbonized state, at least one second carbon load is determined using a pre-trained carbon loading determination model.

[0062] The pre-trained carbon loading determination model is used to determine the carbon loading of the particulate filter in a carbon-deposited state. Specifically, the inputs to the pre-trained carbon loading determination model are the exhaust gas volumetric flow rate and the pressure difference between the upstream and downstream sides of the particulate filter.

[0063] Optionally, the pre-trained carbon loading determination model can be trained using the exhaust gas volumetric flow rate under carbon deposition conditions and the upstream and downstream pressure difference of the particulate filter under carbon deposition conditions.

[0064] The model for determining carbon loading can be a supervised model, a semi-supervised model, or an unsupervised model.

[0065] If the carbon loading determination model is a supervised or semi-supervised model, the training data also includes the carbon loading of the particulate trap under carbon deposition conditions.

[0066] Understandably, over time, carbon loading determination models can identify multiple secondary carbon loadings.

[0067] S400, determine the third carbon loading based on at least one first carbon loading and at least one second carbon loading.

[0068] Specifically, step S400 may include:

[0069] Wavelet denoising is performed on the carbon load sequence consisting of at least one first carbon load and at least one second carbon load, and sliding flat filtering is performed on the carbon load sequence that has undergone wavelet denoising.

[0070] The third carbon load is determined from the carbon load sequence that has undergone sliding flat filtering.

[0071] Optionally, the passive regenerated carbon loading determination model and the carbon deposition carbon loading determination model can be neural network models.

[0072] Optionally, in another embodiment, the training process of the passive regeneration carbon loading determination model includes: obtaining a first initial model and first training data, wherein the first training data includes the exhaust gas volume flow rate under passive regeneration and the upstream and downstream pressure difference of the particulate filter under passive regeneration, and the first initial model is a nonlinear autoregressive neural network (NARX) model with external input; training the first initial model using the first training data, wherein during the training process, the parameters of the network structure of the first initial model are adjusted using the momentum-adaptive gradient descent method, wherein the parameters include: the threshold of the hidden layer, the weight of the hidden layer, the threshold of the output layer, and the weight of the output layer.

[0073] When training the passive regenerated carbon loading determination model, this application allows for data initialization with a finite delay step size. For example, setting the finite delay step size to 10 steps means the model will sample data in units of 10. If there are fewer than 10 data points, this application can fill in the missing data using existing data through repetition or by using previously sampled historical data.

[0074] By using the aforementioned finite delay step, this application can achieve a filtering effect, reduce interference from abnormal data, and improve data accuracy.

[0075] Optionally, in another embodiment of the present invention, Figure 1 The method shown may also include:

[0076] If the particulate trap is in an active regeneration state, at least one fourth carbon load is determined using a pre-trained active regeneration carbon loading determination model.

[0077] Step S400 may specifically include:

[0078] The third carbon loading is determined based on at least one first carbon loading, at least one second carbon loading, and at least one fourth carbon loading.

[0079] The active regenerative carbon loading determination model can be any existing mature active regenerative carbon loading determination model, and this application does not limit the active regenerative carbon loading determination model.

[0080] Optionally, in another embodiment, the training process of the carbon loading determination model includes: obtaining a second initial model and second training data, wherein the second training data includes the exhaust gas volume flow rate under carbon deposition and the upstream and downstream pressure difference of the particulate filter under carbon deposition, and the second initial model is a nonlinear autoregressive neural network (NARX) model with external input; training the second initial model using the second training data, wherein during the training process, the parameters of the network structure of the second initial model are adjusted using the momentum-adaptive gradient descent method, wherein the parameters include: the threshold of the hidden layer, the weight of the hidden layer, the threshold of the output layer, and the weight of the output layer.

[0081] Optionally, this application can be based on the formula to Calculate the hidden layer respectively Input of each node Hidden layer Output of each node Output layer Input of each node Output layer Output of each node ,formula to middle The weights of the hidden layer, This is the threshold for the hidden layer. The weights of the output layer, For the output layer threshold, For hidden layer transfer functions, The weights and thresholds are learned using gradient descent (traingd algorithm) for the output layer transfer function. It equals the product of the number of inputs and the delay step.

[0082] Optionally, this application can be based on the formula to Calculate the output layer's first... The adjustment amount of the weight of each node Output layer Correction amount of node threshold Hidden layer The adjustment amount of the weight of each node Hidden layer Correction amount of node threshold In the formula The weights of the hidden layer, This is the threshold for the hidden layer. The weights of the output layer, For the output layer threshold, It is the inverse function of the hidden layer transfer function. It is the inverse function of the output layer transfer function. The learning rate is set to a fixed value of 0.04.

[0083]

[0084]

[0085]

[0086]

[0087]

[0088]

[0089]

[0090]

[0091] The momentum-adaptive gradient descent (TraingDX algorithm) method is used to learn weights and thresholds. Compared to gradient descent (Traingd algorithm), it introduces a momentum factor and a variable learning rate, allowing it to escape local optima and achieve higher training accuracy. The formula is as follows: ~ For learning algorithms, where Mean square error, For input, For the number of training sessions, Momentum factor For learning rate, For the first The sum of squared errors of each step. According to the formula... Calculate the momentum factor using the formula Calculate the learning rate, where For the first The sum of squared errors of each step.

[0092] (9)

[0093] (10)

[0094] (11)

[0095] Corresponding to the above method embodiments, this application also provides a carbon loading determination device, such as... Figure 2 As shown, the device may include:

[0096] The state determination unit 100 is used to determine the current state of the particle collector;

[0097] The first determining unit 200 is used to determine at least one first carbon load using a pre-trained passive regeneration carbon load determination model if the particulate trap is in a passive regeneration state and the duration of the passive regeneration state exceeds a preset duration.

[0098] The second determining unit 300 is used to determine at least one second carbon load using a pre-trained carbon load determination model if the particulate trap is in a carbon deposit state.

[0099] The third determining unit 400 is used to determine a third carbon loading based on the at least one first carbon loading and the at least one second carbon loading.

[0100] Optionally, the state determination unit 100 is specifically used to obtain the exhaust gas volume flow rate and the upstream exhaust temperature of the particulate filter; if the exhaust gas volume flow rate is within a preset flow range and the upstream exhaust temperature is within a preset temperature range, then the current state of the particulate filter is determined to be a passive regeneration state.

[0101] Optionally, the third determining unit 400 is specifically used to: perform wavelet denoising processing on the carbon load sequence composed of the at least one first carbon load and the at least one second carbon load; perform sliding flat filtering processing on the carbon load sequence that has undergone wavelet denoising processing; and determine a third carbon load from the carbon load sequence that has undergone sliding flat filtering processing.

[0102] Optionally, the inputs to the pre-trained passive regenerated carbon loading determination model are the exhaust gas volumetric flow rate and the upstream and downstream pressure difference of the particulate filter, and the inputs to the pre-trained carbon deposition carbon loading determination model are the exhaust gas volumetric flow rate and the upstream and downstream pressure difference of the particulate filter.

[0103] Optionally, it may also include a first training unit and / or a second training unit.

[0104] The first training unit is used to obtain a first initial model and first training data. The first training data includes the exhaust gas volumetric flow rate under the passive regeneration state and the upstream and downstream pressure difference of the particulate filter under the passive regeneration state. The first initial model is a nonlinear autoregressive neural network (NARX) model with external input. The first initial model is trained using the first training data. During the training process, the momentum-adaptive gradient descent method is used to adjust the parameters of the network structure of the first initial model. The parameters include: the threshold of the hidden layer, the weight of the hidden layer, the threshold of the output layer, and the weight of the output layer.

[0105] The second training unit is used to obtain a second initial model and second training data. The second training data includes the volumetric flow rate of exhaust gas under carbon deposit conditions and the upstream and downstream pressure difference of the particulate filter under carbon deposit conditions. The second initial model is a nonlinear autoregressive neural network (NARX) model with external input. The second initial model is trained using the second training data. During the training process, the momentum-adaptive gradient descent method is used to adjust the parameters of the network structure of the second initial model. The parameters include: the threshold of the hidden layer, the weight of the hidden layer, the threshold of the output layer, and the weight of the output layer.

[0106] The carbon load determination device includes a processor and a memory. The aforementioned state determination unit, first determination unit, second determination unit, and third determination unit are all stored in the memory as program units. The processor executes the aforementioned program units stored in the memory to achieve the corresponding functions.

[0107] The processor contains a core, which retrieves the corresponding program unit from memory. One or more cores can be configured, and the carbon load is determined by adjusting the core parameters.

[0108] This invention provides a storage medium storing a program that, when executed by a processor, implements the carbon loading determination method.

[0109] This invention provides a processor for running a program, wherein the program executes the carbon loading determination method during runtime.

[0110] like Figure 3 As shown, this embodiment of the invention provides an electronic device 70, which includes at least one processor 701, at least one memory 702 connected to the processor 701, and a bus 703. The processor 701 and the memory 702 communicate with each other via the bus 703. The processor 701 is used to call program instructions in the memory 702 to execute the aforementioned carbon loading determination method. The electronic device described herein can be a server, PC, PAD, mobile phone, ECU (Electronic Control Unit), VCU (Vehicle Control Unit), MCU (Micro Controller Unit), HCU (Hybrid Control Unit), etc.

[0111] This application also provides a computer program product that, when executed on a data processing device, is adapted to perform a program that initializes the steps included in the above-described carbon loading determination method.

[0112] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0113] In a typical configuration, the device includes one or more processors (CPUs), memory, and a bus. The device may also include input / output interfaces, network interfaces, etc.

[0114] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, like read-only memory (ROM) or flash RAM, and memory includes at least one memory chip. Memory is an example of computer-readable media.

[0115] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0116] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0117] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0118] The various embodiments in this specification are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0119] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for determining carbon loading, characterized in that, include: Determine the current state of the particle trap; If the particle trap is in a passive regeneration state and the duration of the passive regeneration state exceeds a preset duration, then a pre-trained passive regeneration carbon loading determination model is used to determine at least one first carbon loading. If the particulate trap is in a carbonized state, at least one second carbon load is determined using a pre-trained carbon loading determination model. The third carbon loading is determined based on the at least one first carbon loading and the at least one second carbon loading; Determining the third carbon loading based on the at least one first carbon loading and the at least one second carbon loading includes: Wavelet denoising is performed on the carbon load sequence consisting of at least one first carbon load and at least one second carbon load, and moving average filtering is performed on the carbon load sequence that has undergone wavelet denoising. The third carbon load is determined from the carbon load sequence that has undergone the moving average filtering process described above; The inputs to the pre-trained passive regenerated carbon load determination model are the exhaust gas volumetric flow rate and the upstream and downstream pressure difference of the particulate filter; the inputs to the pre-trained carbon deposition carbon load determination model are the exhaust gas volumetric flow rate and the upstream and downstream pressure difference of the particulate filter. The training process of the passive regeneration carbon loading determination model includes: obtaining a first initial model and first training data, wherein the first training data includes the exhaust gas volumetric flow rate under the passive regeneration state, the upstream and downstream pressure difference of the particulate filter under the passive regeneration state, and the carbon loading of the particulate filter under the passive regeneration state; the first initial model is a nonlinear autoregressive neural network (NARX) model with external input; training the first initial model using the first training data, wherein the parameters of the network structure of the first initial model are adjusted using the momentum-adaptive gradient descent method during the training process, wherein the parameters include: the threshold of the hidden layer, the weight of the hidden layer, the threshold of the output layer, and the weight of the output layer; And / or, The training process of the carbon loading determination model includes: obtaining a second initial model and second training data, wherein the second training data includes the exhaust gas volumetric flow rate under carbon deposition conditions, the upstream and downstream pressure difference of the particulate filter under carbon deposition conditions, and the carbon loading of the particulate filter under carbon deposition conditions; the second initial model is a nonlinear autoregressive neural network (NARX) model with external input; and training the second initial model using the second training data. During the training process, the parameters of the network structure of the second initial model are adjusted using the momentum-adaptive gradient descent method, wherein the parameters include: the threshold of the hidden layer, the weight of the hidden layer, the threshold of the output layer, and the weight of the output layer.

2. The method for determining carbon loading according to claim 1, characterized in that, Determining the current state of the particle trap includes: The exhaust gas volumetric flow rate and the upstream exhaust temperature of the particulate filter are obtained. If the exhaust gas volume flow rate is within a preset flow range and the upstream exhaust temperature is within a preset temperature range, then the current state of the particulate filter is determined to be a passive regeneration state.

3. A carbon loading determination device, characterized in that, The apparatus for implementing the method of claim 1 or 2, comprising: The state determination unit is used to determine the current state of the particle collector; The first determining unit is used to determine at least one first carbon load using a pre-trained passive regeneration carbon load determination model if the particle trap is in a passive regeneration state and the duration of the passive regeneration state exceeds a preset duration. The second determining unit is used to determine at least one second carbon load using a pre-trained carbon load determination model if the particulate trap is in a carbon deposit state. The third determining unit is used to determine the third carbon loading based on the at least one first carbon loading and the at least one second carbon loading.

4. The carbon loading determination device according to claim 3, characterized in that, The state determination unit is specifically used for: Obtain the exhaust gas volumetric flow rate and the upstream exhaust temperature of the particulate filter; if the exhaust gas volumetric flow rate is within a preset flow range and the upstream exhaust temperature is within a preset temperature range, then determine that the current state of the particulate filter is a passive regeneration state.

5. The carbon loading determination device according to claim 3, characterized in that, The third determining unit is specifically used to: perform wavelet denoising on the carbon load sequence composed of the at least one first carbon load and the at least one second carbon load, and perform moving average filtering on the carbon load sequence that has undergone the wavelet denoising. The third carbon load is determined from the carbon load sequence that has undergone the moving average filtering process.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program that, when executed by a processor, implements the carbon loading determination method according to claim 1 or 2.

7. An electronic device, characterized in that, The electronic device includes at least one processor, and at least one memory and a bus connected to the processor; wherein the processor and the memory communicate with each other through the bus; the processor is used to call program instructions in the memory to execute the carbon loading determination method according to claim 1 or 2.

Images