Method, device, storage medium and electronic device for determining carbon loading overload

Abstract

The present application provides a method, device, storage medium, and electronic device for determining carbon loading overload. The method includes: an acquisition step: acquiring carbon loading-related parameters of a particulate trap installed on a vehicle to be tested; a determination step: determining the carbon loading at a future preset moment according to the carbon loading-related parameters and a long short-term memory network model; repeating the acquisition step and the determination step multiple times in sequence to obtain the carbon loadings at multiple future preset moments within a continuous time period, and the difference between two historical moments corresponding to the carbon loading-related parameters in two adjacent repetitions is a preset difference; in the case that at least part of the carbon loadings at multiple future preset moments are greater than a preset carbon loading threshold, it is determined that the carbon loading is overloaded. The present application inputs the carbon loading-related parameters into the long short-term memory network model to obtain the carbon loading at a future preset moment, and solves the problem of low accuracy of the early judgment scheme for carbon loading overload.

Description

Technical Field

[0001] The present application relates to the technical field of engines, and in particular, to a method for determining carbon loading overload, a device for determining carbon loading overload, a computer-readable storage medium, and an electronic device. Background Art

[0002] Overloading of the particulate filter carbon loading has an adverse impact on the fuel economy and power performance of the whole vehicle. The accumulated carbon may generate a large temperature gradient during combustion, which may cause cracking or melting of the particulate filter. After the particulate filter overload is triggered, even if the torque is limited immediately, it will bring potential hazards to safe driving.

[0003] Currently, the accuracy of the early judgment scheme for particulate filter carbon loading overload is relatively low. Summary of the Invention

[0004] The main objective of the present application is to provide a method for determining carbon loading overload, a device for determining carbon loading overload, a computer-readable storage medium, and an electronic device, so as to at least solve the problem of relatively low accuracy of the early judgment scheme for particulate filter carbon loading overload in the prior art.

[0005] To achieve the above objective, according to one aspect of the present application, a method for determining carbon loading overload is provided, including: an acquisition step: acquiring carbon loading related parameters of a particulate filter installed on a vehicle to be tested, where the carbon loading related parameters are related parameters acquired within a preset time period, and the preset time period is a time period from a historical moment to the current moment. Among them, the carbon loading related parameters include at least one of the following: the engine speed, the engine torque, the engine fuel injection volume, the engine exhaust gas volume flow rate, the upstream temperature of the particulate filter, the differential pressure of the particulate filter, and the carbon loading of the particulate filter calculated in real time using a mechanism model; a determination step: determining the carbon loading at a future preset moment according to the carbon loading related parameters and a long short-term memory network model, where the time difference between the future preset moment and the current moment is equal to the duration of the preset time period. Among them, the carbon loading related parameters are used as the input of the long short-term memory network model, and the carbon loading at the future preset moment is used as the output of the long short-term memory network model; repeating the acquisition step and the determination step multiple times in sequence to obtain the carbon loading at multiple future preset moments within a continuous time period, and the difference between the two historical moments corresponding to the carbon loading related parameters in two adjacent repetitions is a preset difference; in the case that at least part of the carbon loading at multiple future preset moments is greater than a preset carbon loading threshold, determining carbon loading overload.

[0006] Optionally, before determining the carbon loading at a future preset moment according to the carbon loading related parameter and the long short-term memory network model, the method further includes: constructing an initial long short-term memory network model; training and testing the initial long short-term memory network model with a training set and a test set to obtain a long short-term memory network model, where the training set includes a first data set and a second data set, the test set includes a third data set, the first data set includes carbon loading related parameters of multiple carbon loading overloaded vehicles in a historical period before overload, the second data set includes carbon loading related parameters of multiple carbon loading non-overloaded vehicles in the historical period, the third data set includes real-time carbon loading related parameters of a target vehicle, and the engines and the particulate traps of the multiple carbon loading overloaded vehicles, the multiple carbon loading non-overloaded vehicles, the target vehicle, and the vehicle to be measured are of the same model.

[0007] Optionally, during the process of training the initial long short-term memory network model with the training set, the method further includes: within a preset hyperparameter range, using the mean squared error as a loss function and optimizing the parameters in the initial long short-term memory network model by using a gradient descent algorithm to obtain optimized parameters, where the initial long short-term memory network model includes an input layer, a first hidden layer, a second hidden layer, and an output layer, the first hidden layer and the second hidden layer each include a plurality of hidden neurons, and the parameters include: first target weights corresponding to each input layer node, first target thresholds corresponding to each first hidden layer node, second target thresholds corresponding to each second hidden layer node, and second target weights corresponding to each output layer node; replacing the parameters before optimization with the optimized parameters.

[0008] Optionally, after training the initial long short-term memory network model with the training set to obtain the long short-term memory network model, the method further includes: using a preset test index to test the prediction accuracy of the initial long short-term memory network model; in the case where the prediction accuracy meets a preset accuracy range, determining the initial long short-term memory network model as the long short-term memory network model.

[0009] Optionally, before training and testing the initial long short-term memory network model with a training set and a test set to obtain a long short-term memory network model, the method further includes: obtaining model-related parameters of a plurality of the vehicles with carbon loading overload, a plurality of the vehicles with non-overloaded carbon loading, the target vehicle, and the vehicle to be tested, where the model-related parameters include: engine-related parameters and particulate filter-related parameters, and the engine-related parameters include at least one of the following: cylinder diameter, stroke, designed explosion pressure, rated speed, maximum torque speed, rated power, physical and chemical properties of the surface of the piston and the cylinder liner; the particulate filter-related parameters include at least one of the following: model, porosity, pore diameter, mesh number parameter; when the model-related parameters of the plurality of the vehicles with carbon loading overload, the plurality of the vehicles with non-overloaded carbon loading, the target vehicle, and the vehicle to be tested are the same, it is determined that the engines and the particulate filters of the plurality of the vehicles with carbon loading overload, the plurality of the vehicles with non-overloaded carbon loading, the target vehicle, and the vehicle to be tested are of the same model.

[0010] Optionally, when at least part of the carbon loadings at a plurality of the future preset times are greater than a preset carbon loading threshold, determining that the carbon loading is overloaded includes: obtaining a first duration of the carbon loadings at the future preset times that are greater than a first preset carbon loading threshold, and when it is determined that the ratio of the first duration to the continuous time period is greater than a preset ratio, determining that the carbon loading overload is a first-level overload; obtaining a second duration of the carbon loadings at the future preset times that are greater than a second preset carbon loading threshold, and when the second duration is greater than or equal to a preset duration, determining that the carbon loading overload is a second-level overload, where the second preset carbon loading threshold is greater than the first preset carbon loading threshold.

[0011] Optionally, the method further includes: setting an intelligent operation and maintenance platform for the carbon loading overload, where the intelligent operation and maintenance platform has a plurality of display areas, and the plurality of display areas include a first display area and a second display area; the first display area is used to display real-time state parameters of the vehicle to be tested and the carbon loadings at a plurality of the future preset times, and the real-time state parameters include at least one of the following: the carbon loading of the particulate filter calculated in real time by a mechanism model at the current moment, the upstream temperature of the particulate filter at the current moment; the second display area is used to display a first fault code and a second fault code, where the first fault code is used to represent the first-level overload, and the second fault code is used to represent the second-level overload.

[0012] According to another aspect of the present application, a device for determining carbon loading overload is provided. The device includes: an acquisition unit configured to perform an acquisition step of acquiring carbon loading related parameters of a particulate filter installed on a vehicle to be tested. The carbon loading related parameters are parameters acquired within a preset time period, which is a time period from a historical moment to the current moment. The carbon loading related parameters include at least one of the following: the engine speed, the engine torque, the engine fuel injection amount, the engine exhaust gas volume flow rate, the upstream temperature of the particulate filter, the differential pressure of the particulate filter, and the carbon loading of the particulate filter calculated in real time using a mechanism model; a determination unit configured to perform a determination step of determining the carbon loading at a future preset moment according to the carbon loading related parameters and a long short-term memory network model. The time difference between the future preset moment and the current moment is equal to the duration of the preset time period. The carbon loading related parameters serve as the input of the long short-term memory network model, and the carbon loading at the future preset moment serves as the output of the long short-term memory network model; a repeated acquisition unit configured to repeatedly perform the acquisition step and the determination step multiple times in sequence to obtain the carbon loading at multiple future preset moments within a continuous time period, and the difference between two historical moments corresponding to the carbon loading related parameters in two adjacent repetitions is a preset difference; an overload determination unit configured to determine carbon loading overload when at least part of the carbon loading at multiple future preset moments is greater than a preset carbon loading threshold.

[0013] According to still another aspect of the present application, a computer-readable storage medium is provided. The computer-readable storage medium includes a stored program. When the program runs, it controls the device where the computer-readable storage medium is located to execute any one of the methods for determining carbon loading overload.

[0014] According to yet another aspect of the present application, an electronic device is provided, including: one or more processors, a memory, and one or more programs. The one or more programs are stored in the memory and are configured to be executed by the one or more processors. The one or more programs include those for executing any one of the methods for determining carbon loading overload.

[0015] Applying the technical solution of the present application, the first acquisition step is to acquire the carbon loading related parameters of the particulate trap installed on the vehicle to be tested. The carbon loading related parameters are the related parameters acquired within a preset time period, and the preset time period is the time period from the historical moment to the current moment. Among them, the carbon loading related parameters include at least one of the following: the engine speed, the engine torque, the engine fuel injection amount, the engine exhaust gas volume flow rate, the upstream temperature of the particulate trap, the pressure difference of the particulate trap, and the carbon loading of the particulate trap calculated in real time by the mechanism model. Then, the determination step is to determine the carbon loading at a future preset moment according to the carbon loading related parameters and the long short-term memory network model. The time difference between the future preset moment and the current moment is equal to the duration of the preset time period. Among them, the carbon loading related parameters are used as the input of the long short-term memory network model, and the carbon loading at the future preset moment is used as the output of the long short-term memory network model. Then, the acquisition step and the determination step are repeated multiple times in sequence to obtain the carbon loadings at multiple future preset moments within a continuous time period, and the difference between the two historical moments corresponding to the carbon loading related parameters in two adjacent repetitions is the preset difference. Finally, when at least part of the carbon loadings at multiple future preset moments is greater than the preset carbon loading threshold, it is determined that the carbon loading is overloaded. The present application inputs the carbon loading related parameters of the particulate trap acquired within the time period from the historical moment to the current moment into the long short-term memory network model, so as to obtain the carbon loading at the future preset moment, and can determine in advance whether the carbon loading is overloaded, solving the problem of low accuracy of the prior art in the early judgment of the carbon loading overload of the particulate trap. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The specification drawings forming a part of the present application are used to provide a further understanding of the present application. The schematic embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation to the present application. In the drawings:

[0017] Figure 1 The hardware structure block diagram of a mobile terminal for executing a method for determining carbon loading overload provided in an embodiment of the present application is shown;

[0018] Figure 2 The flowchart of a method for determining carbon loading overload provided in an embodiment of the present application is shown;

[0019] Figure 3 The flowchart of a specific method for determining carbon loading overload provided in an embodiment of the present application is shown;

[0020] Figure 4 The structure block diagram of a device for determining carbon loading overload provided in an embodiment of the present application is shown. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0021] It should be noted that, without conflict, the embodiments in the present application and the features in the embodiments may be combined with each other. The following will describe the present application in detail with reference to the drawings and in combination with the embodiments.

[0022] In order to enable those skilled in the art to better understand the solutions of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in 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 of ordinary skill in the art without making creative efforts shall fall within the protection scope of the present application.

[0023] It should be noted that the terms "first", "second", etc. in the specification, claims and the above drawings of the present application are used to distinguish similar objects, and do not necessarily need to be used to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so as to describe the embodiments of the present application here. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device including a series of steps or units does not necessarily need to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0024] For the convenience of description, some nouns or terms related to the embodiments of the present application are described below:

[0025] Long Short-Term Memory Network Model (LSTM): A special recurrent neural network structure. Due to its unique design structure, this network can alleviate the long-term dependence problem and has the ability to learn the law of long-term performance degradation and use it to update the current state;

[0026] Diesel Particulate Filter (DPF) carbon deposition: A kind of solid particle in the engine exhaust intercepted by the DPF. The core is solid carbon (C), and the outside contains polymers generated due to incomplete engine combustion;

[0027] Diesel Particulate Filter (DPF) overload: As carbon particles gradually accumulate in the DPF, it will cause the carbon particle carrying capacity of the DPF to exceed the limit. According to the amount of carbon particle accumulation, it can be divided into level1 overload and level2 overload. Generally speaking, the limit of level1 overload is 5.5 g / L, and that of level2 overload is 6.5 g / L;

[0028] Diesel Particulate Filter (DPF) differential pressure: The pressure difference between the inlet end and the outlet end of the DPF, which is mainly affected by the number of particles intercepted in the DPF.

[0029] As introduced in the background art, the prior art has a low accuracy in the early judgment scheme for the carbon loading overload of the particulate filter. To solve the problem of low accuracy in the early judgment scheme for carbon loading overload, embodiments of the present application provide a method for determining carbon loading overload, a device for determining carbon loading overload, a computer-readable storage medium, and an electronic device.

[0030] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention.

[0031] The method embodiments provided in the embodiments of the present application can be executed on a mobile terminal, a computer terminal, or a similar computing device. Taking running on a mobile terminal as an example, Figure 1 is a hardware structure block diagram of a mobile terminal for a method of determining carbon loading overload according to an embodiment of the present invention. As Figure 1 shown, the mobile terminal may include one or more ( Figure 1 only one is shown in the figure) processors 102 (the processor 102 may include, but is not limited to, a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data. Among them, the above mobile terminal may further include a transmission device 106 for communication functions and an input / output device 108. Those of ordinary skill in the art can understand that Figure 1 the structure shown is only schematic and does not limit the structure of the above mobile terminal. For example, the mobile terminal may further include more or fewer components than Figure 1 shown in the figure, or have a different configuration from Figure 1 shown in the figure.

[0032] The memory 104 can be used to store computer programs, such as software programs and modules of application software, such as the computer program corresponding to the display method of device information in the embodiments of the present invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, implements the above-mentioned method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memories, or other non-volatile solid-state memories. In some instances, the memory 104 may further include a memory remotely disposed relative to the processor 102, and these remote memories may be connected to the mobile terminal through a network. Examples of the above-mentioned network include but are not limited to the Internet, an enterprise intranet, a local area network, a mobile communication network, and combinations thereof. The transmission device 106 is used to receive or send data via a network. Specific examples of the above-mentioned network may include a wireless network provided by a communication provider of the mobile terminal. In one instance, the transmission device 106 includes a network adapter (Network Interface Controller, abbreviated as NIC), which can be connected to other network devices through a base station and thus can communicate with the Internet. In one instance, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0033] In this embodiment, a method for determining carbon loading overload running on a mobile terminal, a computer terminal, or a similar computing device is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0034] Figure 2 It is a flowchart of a method for determining carbon loading overload according to an embodiment of the present application. As Figure 2 shown, the method includes the following steps:

[0035] Step S201, acquisition step: Acquire carbon loading related parameters of a particulate filter installed on a vehicle to be tested. The above carbon loading related parameters are related parameters acquired within a preset time period, and the preset time period is a time period from a historical moment to the current moment. Among them, the carbon loading related parameters include at least one of the following: the engine speed, the engine torque, the engine fuel injection volume, the engine exhaust gas volume flow, the upstream temperature of the particulate filter, the differential pressure of the particulate filter, and the carbon loading of the particulate filter calculated in real time using a mechanism model;

[0036] Specifically, data such as the operating parameters of the engine, the upstream temperature of the DPF, the differential pressure of the DPF, and the carbon loading calculated by the mechanism model are collected and transmitted back through sensors and Internet of Things technology. The mechanism model, also known as the white box model, is an accurate mathematical model established based on the internal mechanism of the object, the production process, or the transfer mechanism of the material flow. Here, the historical moment is a preset moment before the current moment. For example, if the preset time period is 50 minutes, the historical moment is the moment 50 minutes before the current moment.

[0037] Step S202, determination step: According to the above carbon loading related parameters and the long short-term memory network model, determine the carbon loading at a future preset moment. The time difference between the above future preset moment and the above current moment is equal to the duration of the above preset time period. Among them, the above carbon loading related parameters serve as the input of the above long short-term memory network model, and the carbon loading at the above future preset moment serves as the output of the above long short-term memory network model;

[0038] Specifically, when the preset time period is 50 minutes, the future preset moment is the moment 50 minutes after the current moment. The long short-term memory network model is a commonly used deep neural network model that can model sequence data and has a certain memory ability, suitable for tasks such as time series prediction. In this solution, the long short-term memory network model can make full use of the carbon loading related parameter information at the historical moment to predict the carbon loading at the future preset moment.

[0039] Step S203, repeat the above acquisition step and the above determination step multiple times in sequence to obtain the carbon loading at multiple above future preset moments within a continuous time period, and the difference between the two above historical moments corresponding to the above carbon loading related parameters in two adjacent repetitions is a preset difference;

[0040] Specifically, the carbon loading related parameters are input into the sliding window in real time. The sliding window contains 15 sample points. The sample points include the rotational speed, torque, fuel injection quantity, exhaust gas volume flow rate, DPF upstream temperature, DPF differential pressure, and carbon loading within the current moment and the past 50 minutes. The output is the increment of the carbon loading at the moment 50 minutes in the future. Adding the predicted increment of the carbon loading to the carbon loading at the current moment can obtain the carbon loading at the moment 50 minutes in the future. By establishing a non-linear mapping relationship between the input and output, the purpose of predicting the carbon loading 50 minutes later is achieved. When the sliding window advances by 1 second, the predicted carbon loading also advances by 1 second, and continuous carbon loading predictions 50 minutes later can be obtained by predicting in sequence.

[0041] Step S204, in the case where at least part of the carbon loading at multiple above future preset moments is greater than the preset carbon loading threshold, determine that the carbon loading is overloaded.

[0042] Specifically, in order to avoid false alarms as much as possible, the predicted carbon loading at a future preset time must first exceed the limit value, and there is at least one carbon loading at a future preset time greater than the preset carbon loading threshold within a continuous time period before it is determined that the carbon loading is overloaded. By predicting and analyzing the carbon loading situation of the particulate filter during vehicle driving and making timely adaptive adjustments to avoid overloading of the carbon loading and prevent the situation of torque limitation in a timely manner, which may pose a hidden danger to safe driving. In a possible implementation, when the carbon loadings at multiple future preset times are all less than or equal to the preset carbon loading threshold, it is determined that the carbon loading is not overloaded.

[0043] The embodiment of the present application provides a method for determining carbon loading overload. First, an acquisition step: acquiring carbon loading related parameters of a particulate filter installed on a vehicle to be tested; then a determination step: determining the carbon loading at a future preset time according to the above carbon loading related parameters and a long short-term memory network model; then repeating the above acquisition step and the above determination step multiple times in sequence to obtain the carbon loadings at multiple future preset times within a continuous time period, and the difference between the two historical times corresponding to the above carbon loading related parameters in two adjacent repetitions is a preset difference; finally, when at least some of the carbon loadings at multiple future preset times are greater than the preset carbon loading threshold, it is determined that the carbon loading is overloaded. The present application acquires the carbon loading related parameters at the current time and within a continuous historical time period before the current time of the particulate filter, predicts the carbon loading at a future preset time based on the long short-term memory network model (LSTM), and combines the preset carbon loading threshold and the preset difference to determine whether there is an overload of the carbon loading in the vehicle, so as to ensure that the particulate filter will not have an overload of the carbon loading, and solves the problem of low accuracy in the prior art for the early judgment scheme of the carbon loading overload of the particulate filter.

[0044] As a possible implementation, before determining the carbon loading at a future preset time according to the above carbon loading related parameters and the long short-term memory network model, the above method further includes:

[0045] Step S301, constructing an initial long short-term memory network model;

[0046] Specifically, the entire initial long short-term memory network model is divided into an input layer, a first hidden layer, a second hidden layer, and an output layer. The training of the neural network consists of multiple iterations, and both forward propagation and backward propagation steps are required in each iteration. The core gating units of the long short-term memory network model include a forgetting gate, an input gate, and an output gate. During the forward propagation process of the hidden layer, the forgetting gate reads the state s of the hidden layer at the previous moment t-1 and the input x at the current moment t, perform a non - linear mapping of the sigmoid function. Each element of the output of the sigmoid function (which is a vector) is a real number between 0 and 1, representing the weight (or proportion) for allowing the corresponding information to pass. For example, 0 means not allowing any information to pass, and 1 means allowing all information to pass, as shown in (Equation 1). The forget gate outputs a vector f1 with values in the range (0, 1). The input to the input gate is also s t-1 and x t , W fs , W fx and b f are the learnable weights between the input and output of the forget gate. As shown in (Equation 2), the first - half sigmoid - function non - linear mapping determines which information needs to be updated. The tanh layer is equivalent to sorting out the information at the previous and current time points. At the output of the input gate for the current cell state C t the vector f2 that needs to be updated. Among them, W is , W ix , b i , W Cs , W Cx and b C are the learnable weights between the input and output of the input gate. Generally speaking, the current cell state C t is to forget unimportant information under the previous cell state C t-1 and update a part of new information f2 according to the new input, as shown in (Equation 3), thus forming a long - term memory chain. The output gate has two parts of input. One part is the state s t-1 of the hidden layer at the previous time step. The other part is the input x t at the current time step. After passing through the non - linear mapping of the sigmoid function, it determines the update of the short - term memory for the current output o t , as shown in (Equation 4). Among them, W os , W ox and b o are the learnable weights between the input and output of the output gate. Then, the current cell state C t is processed through tanh and multiplied by o t to determine the output yt of the output gate, as shown in (Equation 5). That is, use the input at the current time step and the output at the previous time step for calculation to obtain the output at the current time step, to model the relationship between the input and output, and the historical correlation between the outputs. The specific formulas are as follows:

[0047] f1 = sigmoid(W fs s t-1 +W fx x t +b f ) (Equation 1)

[0048] f2 = sigmoid(W is s t-1 + W ix x t + b i ) × tanh(W Cs s t-1 + W Cx x t + b C ) (Formula 2)

[0049]

[0050] o t = sigmoid(W os s t-1 + W ox x t + b o ) (Formula 4)

[0051] y t = o t · tanh(C t ) (Formula 5)

[0052] In addition, during the backpropagation process of the hidden layer, the gradients of each parameter and the hidden state are calculated in sequence; first, the hidden state of the last time step is calculated, then the parameters of the last time step are calculated, and then the hidden state of the penultimate time step is calculated, and so on; after obtaining the gradients of each parameter, each parameter is subtracted by a set multiple of its gradient to complete the backpropagation;

[0053] Step S302, use the training set and the test set to train and test the above initial long short-term memory network model to obtain a long short-term memory network model. The above training set includes a first data set and a second data set, the above test set includes a third data set, the above first data set includes the overload carbon loading related parameters of multiple carbon loading overloaded vehicles during the historical period before overload, the above second data set includes the non-overload carbon loading related parameters of multiple carbon loading non-overloaded vehicles during the above historical period, the above third data set includes the real-time carbon loading related parameters of the target vehicle, and the engines and the particulate traps of the above multiple carbon loading overloaded vehicles, the above multiple carbon loading non-overloaded vehicles, the above target vehicle, and the vehicle to be tested are of the same model.

[0054] Specifically, before model training, it is necessary to perform Kalman filtering and resampling on the obtained training data. The filtering is to filter out abnormal signals in the engine. For the second-level sampling interval of the engine, the data volume is too large, which is very computationally resource-intensive and it is difficult to ensure the real-time transmission of data in the digital twin platform. Therefore, the training data is resampled to collect one data every 200 s. After constructing the long short-term memory network model, the training set and the test set are used for training and testing to obtain an accurate long short-term memory network model.

[0055] It can be seen that by using multiple data sets and observing and training the data sets from multiple perspectives, it can help improve the quality and accuracy of the training data set. And keeping the test data set separately can avoid problems such as data leakage and overfitting, so that the trained long short-term memory network model is more accurate and reliable. By establishing the training set and the test set, it provides an effective means for the training and testing of the long short-term memory network model, so that a more accurate and reliable long short-term memory network model can be obtained, and it provides basic support for the subsequent steps.

[0056] As a possible implementation method, during the process of training the above initial long short-term memory network model with the training set, the above method further includes:

[0057] Step S401, within a preset hyperparameter range, using the mean square error as the loss function, the gradient descent algorithm is used to optimize the parameters in the above initial long short-term memory network model to obtain optimized parameters. The above initial long short-term memory network model includes an input layer, a first hidden layer, a second hidden layer, and an output layer. The above first hidden layer and the above second hidden layer each include a plurality of hidden neurons. The above parameters include: the first target weights corresponding to each input layer node, the first target thresholds corresponding to each first hidden layer node, the second target thresholds corresponding to each second hidden layer node, and the second target weights corresponding to each output layer node;

[0058] Specifically, 2 hidden layers are set in the initial long short-term memory network model, each hidden layer contains 100 neurons, the mean square error is used as the loss function, and the Adam gradient descent algorithm is used to optimize the weights and thresholds. The learning rate lr is 0.001, and the LSTM model parameters when the loss function is minimized are found through 30 iterations.

[0059] Step S402, using the above optimized parameters to replace the parameters before optimization.

[0060] It can be seen that when training a deep learning model, the selection of hyperparameters has a very important impact on the performance and effect of the model. Traditional methods often use grid search and other methods to find the optimal hyperparameters, but this method requires a large number of experiments and repeated training, which is very time-consuming and computationally resource-intensive. Using the gradient descent algorithm within the preset hyperparameter range to find the optimal hyperparameters is more practical and feasible. In addition, since the long short-term memory network model contains multiple hidden neurons and multiple target weights and other parameters, the parameters before optimization are often randomly initialized and are usually not optimal. Therefore, when using the gradient descent algorithm to optimize the model parameters, the model parameters can be optimized more carefully and comprehensively, so as to obtain more accurate and effective model parameters. This helps to improve the prediction accuracy and generalization ability of the model, and enhance the practicality and application value of the model.

[0061] As a possible implementation manner, training the above initial long short-term memory network model with a training set to obtain the above long short-term memory network model, the above method further includes:

[0062] Step S501, using a preset test index to test the prediction accuracy of the above initial long short-term memory network model;

[0063] Specifically, using the mean squared error MSE and the goodness of fit R 2 as the indexes for testing the model, evaluating the prediction accuracy of the model through the preset test index, and further determining whether the model meets the preset accuracy range.

[0064] Step S502, when the above prediction accuracy meets the preset accuracy range, determining the above initial long short-term memory network model as the above long short-term memory network model.

[0065] It can be seen that if the prediction accuracy of the model meets the preset accuracy requirements, the initial model can be determined as the long short-term memory network model, and then this model can be used for actual applications. Compared with the traditional model training method, it has higher controllability and flexibility. In actual applications, it is often necessary to design test indexes according to specific scenarios and requirements, and determine the final model according to the test results. Therefore, introducing test indexes in the model training process can more effectively improve the prediction accuracy of the model, thereby enhancing the reliability and practicality of the model.

[0066] As a possible implementation manner, before training and testing the above initial long short-term memory network model with a training set and a test set to obtain the long short-term memory network model, the above method further includes:

[0067] Step S601: Obtain the model-related parameters of multiple aforementioned vehicles with carbon loading overload, multiple aforementioned vehicles with non-overloaded carbon loading, the aforementioned target vehicle, and the aforementioned vehicle to be tested. The model-related parameters include: engine-related parameters and particulate filter-related parameters. The engine-related parameters include at least one of the following: cylinder diameter, stroke, designed explosion pressure, rated speed, maximum torque speed, rated power, physical and chemical properties of the piston and cylinder liner surfaces. The particulate filter-related parameters include at least one of the following: model, porosity, pore diameter, mesh number parameter;

[0068] Specifically, obtain the design-related parameters of the engine and key components, including cylinder diameter, stroke, designed explosion pressure, rated speed, maximum torque speed, rated power, and physical and chemical properties of the piston and cylinder liner surfaces, and obtain the model, porosity, pore diameter, and mesh number parameters of the DPF. Ensure that the historical data of engines and DPFs of the same model are used to train the time series prediction model of carbon loading. In addition, the operating scenarios and uses of the actual vehicles should be kept as consistent as possible to ensure the characteristic of the same distribution of data features, so that the model can converge through training.

[0069] Step S602: When the model-related parameters of multiple aforementioned vehicles with carbon loading overload, multiple aforementioned vehicles with non-overloaded carbon loading, the aforementioned target vehicle, and the aforementioned vehicle to be tested are the same, determine that the engines and the particulate filters of multiple aforementioned vehicles with carbon loading overload, multiple aforementioned vehicles with non-overloaded carbon loading, the aforementioned target vehicle, and the aforementioned vehicle to be tested are of the same model.

[0070] It can be seen that engines and particulate filters of different types and models have differences in working principles, emission standards, etc. By obtaining the model-related parameters of the vehicles and ensuring that the engine and particulate filter models of each vehicle are the same, the unity and accuracy in training the model can be improved, making the trained model more in line with the actual requirements.

[0071] As a possible implementation, when at least part of the carbon loading at multiple aforementioned future preset times is greater than the preset carbon loading threshold, determining carbon loading overload includes:

[0072] Step S701: Obtain the first duration of the carbon loading at the aforementioned future preset times that is greater than the first preset carbon loading threshold, and when it is determined that the ratio of the first duration to the continuous time period is greater than the preset ratio, determine that the carbon loading overload is a first-level overload;

[0073] Specifically, the diagnostic strategy sets different rules according to the severity of the overload. The limit value when the carbon loading is overloaded, the rules when the overload is triggered, and the time ratio can all be set through the rule interaction module. For the first-level overload, which is less severe, false alarms should be avoided as much as possible. Therefore, the carbon loading at the predicted future preset moment must first exceed the first preset carbon loading threshold, and it is determined that the ratio of the above-mentioned continuous duration to the above-mentioned continuous time period is greater than the preset ratio. For example, the time ratio of overload within 5 consecutive minutes should exceed 60% to trigger the first-level overload.

[0074] Step S702, obtain the second continuous duration of the carbon loading at the above-mentioned future preset moment that is greater than the second preset carbon loading threshold, and when the above-mentioned second continuous duration is greater than or equal to the preset duration, determine that the above-mentioned carbon loading overload is a second-level overload, and the above-mentioned second preset carbon loading threshold is greater than the above-mentioned first preset carbon loading threshold.

[0075] Specifically, since the second-level overload is more severe than the first-level overload and there is a risk of DPF melting and torque limitation, the carbon loading at the predicted future preset moment must first exceed the second preset carbon loading threshold, and the continuous duration of the excess is greater than or equal to the preset duration. For example, if the carbon loading exceeds the second preset carbon loading threshold within 5 consecutive seconds, the second-level overload can be triggered.

[0076] As a possible implementation, the above method further includes:

[0077] Step S801, set up the intelligent operation and maintenance platform for the above-mentioned carbon loading overload. The intelligent operation and maintenance platform has multiple display areas, and the multiple display areas include a first display area and a second display area;

[0078] Specifically, the intelligent operation and maintenance platform has multiple display areas, which can facilitate the remote monitoring and management of the vehicle, effectively improving the maintainability and manageability of the vehicle.

[0079] Step S802, the above-mentioned first display area is used to display the real-time status parameters of the above-mentioned vehicle to be tested and the carbon loading at multiple above-mentioned future preset moments. The real-time status parameters include at least one of the following: the carbon loading of the above-mentioned particulate trap calculated in real time using a mechanism model at the current moment, the upstream temperature of the above-mentioned particulate trap at the current moment;

[0080] Specifically, through the intelligent operation and maintenance platform, the real-time status parameters of the vehicle and the carbon loading data at the future preset moment can be obtained in real time, the carbon loading overload situation can be discovered in time, and accurate positioning and diagnosis can be carried out. By using a mechanism model to calculate the carbon loading of the particulate trap and monitoring real-time status parameters such as the upstream temperature, the carbon loading overload situation can be more accurately reflected and classified and displayed, facilitating the user to quickly locate the problem.

[0081] In step S803, the second display area is used to display a first fault code and a second fault code. The first fault code is used to characterize the first-level overload, and the second fault code is used to characterize the second-level overload.

[0082] Specifically, the first-level overload and the second-level overload are displayed on the intelligent operation and maintenance platform in the form of the first fault code and the second fault code, which facilitates the monitoring personnel to timely understand the fault situation of the vehicle to be tested, and timely adjust the particulate filter of the vehicle to be tested according to the fault situation, so as to avoid the situation of carbon loading overload of the particulate filter.

[0083] In order to enable those skilled in the art to more clearly understand the technical solution of the present application, the implementation process of the method for determining carbon loading overload of the present application will be described in detail below with reference to specific embodiments.

[0084] This embodiment relates to a specific method for determining carbon loading overload. A long short-term memory network model for predicting DPF overload includes a data acquisition module, a data integration module, a data cleaning module, a time series prediction model training module, a time series prediction model prediction module, a rule interaction module, and an intelligent operation and maintenance module for DPF overload, as Figure 3 shown, and includes the following steps:

[0085] Step S1: The data acquisition module, on the premise of ensuring the historical data of engines and DPFs of the same model, acquires carbon loading-related parameters. The carbon loading-related parameters include the engine operating condition parameters such as rotational speed, torque, fuel injection volume, exhaust gas volume flow rate, DPF upstream temperature, DPF pressure difference, and the carbon loading calculated in real time through the model.

[0086] Step S2: The data integration module, through the Internet of Things technology, is connected to the engine digital twin platform in real time, and imports these carbon loading-related parameters into the data integration module for integrated analysis of the data.

[0087] Step S3: The data cleaning module. Before model training, the original data needs to be subjected to Kalman filtering and resampling. The filtering is to filter out abnormal signals in the engine. For the second-level sampling interval of the engine, the data volume is too large, which consumes a lot of computing resources and it is difficult to ensure the real-time transmission of data on the digital twin platform. Therefore, a training data is collected at a resampling interval of 200 s for the original data.

[0088] Step S4: The time series prediction model training module uses the training set and the test set to train and test the above initial long short-term memory network model, and imports the integrated data into the data-driven model. The corresponding input contains the data of a sliding window, and the sliding window contains 15 sample points, that is, it contains the rotational speed, torque, fuel injection quantity, exhaust gas volume flow rate, DPF upstream temperature, DPF differential pressure, and carbon loading of 3000s. The output is the increment of the carbon loading after 50 minutes. By establishing the non-linear mapping relationship between the input and the output, the purpose of predicting the carbon loading after 50 minutes is achieved;

[0089] Step S5: The time series prediction model prediction module. For the trained model, the historical data of the current moment and the previous 3000s is input, and the increment of the carbon loading after 50 minutes can be predicted. Adding the increment of the carbon loading after 50 minutes to the carbon loading of the current state is the predicted carbon loading after 50 minutes. When the sliding window advances 1s, the predicted carbon loading also advances 1 second. By recursively predicting in this way, the continuous predicted carbon loading after 50 minutes can be obtained.

[0090] Step S6: The rule interaction module sets different rules according to the severity of the overload for the diagnostic strategy. The limit value when the carbon loading is overloaded, the rule when the overload is triggered, and the time ratio can all be set through the rule interaction module.

[0091] Step S7: The intelligent operation and maintenance module for DPF overload. On the operation and maintenance platform, the real-time status and the predicted carbon loading of multiple vehicles can be monitored simultaneously. The real-time monitored status includes the DPF carbon loading and the DPF upstream temperature at the current moment.

[0092] Embodiments of the present application can, based on the data collected during engine operation and in combination with the engine digital twin platform, establish the relationship between the historical data of the engine and DPF operation and the carbon loading increment after 50 minutes, thereby obtaining the predicted value of the carbon loading after 50 minutes. By setting certain rules, diagnostic strategies for two types of DPF overload situations are set, realizing the prediction of DPF overload. Based on the above, an intelligent operation and maintenance platform for DPF overload is built, which meets the ability to monitor the health status of DPFs of multiple vehicles simultaneously, including the real-time monitoring of the carbon loading and upstream temperature of the DPF, as well as the prediction of DPF overload. When the system prediction result shows that the DPF is overloaded, an alarm signal is output, and the overload type, overload time, and carbon loading within 5 minutes of continuous monitoring of the overload are recorded. Compared with the traditional method: it can comprehensively consider various factors affecting the carbon loading of the DPF, integrate historical data through the twin platform, and realize the function of early prediction of the DPF; set an alarm in advance to ensure that the engine will not have problems with the vehicle's fuel economy and power performance due to DPF overload, and can also avoid DPF melting and torque limitation due to overload; the prediction of carbon loading is highly efficient, and it can simultaneously monitor the real-time health status of DPFs of multiple vehicles and give an early warning of DPF overload in a timely manner.

[0093] Embodiments of the present application also provide a device for determining carbon loading overload. It should be noted that the device for determining carbon loading overload in the embodiments of the present application can be used to execute the method for determining carbon loading overload provided in the embodiments of the present application. This device is used to implement the above embodiments and preferred implementation manners, and those that have been described will not be repeated. As used hereinafter, the term "module" can be a combination of software and / or hardware that can achieve a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation in hardware, or a combination of software and hardware is also possible and contemplated.

[0094] The following introduces the device for determining carbon loading overload provided in the embodiments of the present application.

[0095] Figure 4 is a schematic diagram of the device for determining carbon loading overload according to an embodiment of the present application. As Figure 4 shown, the device includes: an acquisition unit 41, a determination unit 42, a repeated acquisition unit 43, and an overload determination unit 44.

[0096] An acquisition unit 41 is configured to perform an acquisition step of acquiring parameters related to the carbon loading of a particulate trap installed on a vehicle to be tested. The parameters related to the carbon loading are parameters acquired within a preset time period, and the preset time period is a time period from a historical moment to the current moment. The parameters related to the carbon loading include at least one of the following: the engine speed, the engine torque, the engine fuel injection quantity, the engine exhaust gas volume flow rate, the upstream temperature of the particulate trap, the differential pressure of the particulate trap, and the carbon loading of the particulate trap calculated in real time using a mechanism model.

[0097] Specifically, data such as the operating parameters of the engine, the upstream temperature of the DPF, the differential pressure of the DPF, and the carbon loading calculated using a mechanism model are collected and transmitted back through sensors and Internet of Things technology. The mechanism model is also called a white box model, which is an accurate mathematical model established based on the internal mechanism of the object, the production process, or the transfer mechanism of the material flow. The historical moment is a preset moment before the current moment. For example, if the preset time period is 50 minutes, the historical moment is the moment 50 minutes before the current moment.

[0098] A determination unit 42 is configured to perform a determination step of determining the carbon loading at a future preset moment based on the parameters related to the carbon loading and a long short-term memory network model. The time difference between the future preset moment and the current moment is equal to the duration of the preset time period. The parameters related to the carbon loading are used as the input of the long short-term memory network model, and the carbon loading at the future preset moment is used as the output of the long short-term memory network model.

[0099] Specifically, when the preset time period is 50 minutes, the future preset moment is the moment 50 minutes after the current moment. The long short-term memory network model is a commonly used deep neural network model that can model sequential data and has a certain memory ability, suitable for tasks such as time series prediction. In this solution, the long short-term memory network model can make full use of the information of the parameters related to the carbon loading at the historical moment to predict the carbon loading at the future preset moment.

[0100] A repeated acquisition unit 43 is configured to repeatedly perform the above acquisition step and the above determination step multiple times in sequence to obtain the carbon loadings at multiple future preset moments within a continuous time period, and the difference between the two historical moments corresponding to the parameters related to the carbon loading in two adjacent repetitions is a preset difference.

[0101] Specifically, parameters related to the carbon loading are input into a sliding window in real time. The sliding window contains 15 sample points, and the sample points include the rotational speed, torque, fuel injection quantity, exhaust gas volume flow rate, temperature upstream of the DPF, DPF differential pressure, and carbon loading at the current moment and in the past 50 minutes. The output is the increment of the carbon loading at a moment 50 minutes in the future. By adding the predicted increment of the carbon loading to the carbon loading at the current moment, the carbon loading at a moment 50 minutes in the future can be obtained. By establishing a non-linear mapping relationship between the input and the output, the purpose of predicting the carbon loading 50 minutes later can be achieved. When the sliding window advances by 1 second, the predicted carbon loading also advances by 1 second, and the predicted carbon loading 50 minutes later can be obtained sequentially through continuous prediction.

[0102] The overload determination unit 44 is configured to determine that the carbon loading is overloaded when at least some of the carbon loadings at multiple future preset moments are greater than a preset carbon loading threshold.

[0103] Specifically, in order to avoid false alarms as much as possible, the carbon loading predicted at a future preset moment must first exceed the limit value, and there is at least one carbon loading at a future preset moment greater than the preset carbon loading threshold within a continuous time period before it is determined that the carbon loading is overloaded. By predicting and analyzing the carbon loading situation of the particulate trap during vehicle driving and performing adaptive adjustment in a timely manner, carbon loading overload can be avoided, and the situation of torque limitation in a timely manner, which poses a hidden danger to safe driving, can be avoided. In a possible implementation manner, it is determined that the carbon loading is not overloaded when the carbon loadings at multiple future preset moments are all less than or equal to the preset carbon loading threshold.

[0104] Embodiments of the present application provide a device for determining carbon loading overload. The device includes: an acquisition unit, a determination unit, a repeated acquisition unit, and an overload determination unit. The acquisition unit is configured to perform an acquisition step of acquiring carbon loading related parameters of a particulate filter installed on a vehicle to be measured. The carbon loading related parameters are related parameters acquired within a preset time period, and the preset time period is a time period from a historical moment to the current moment. The determination unit is configured to perform a determination step of determining the carbon loading at a future preset moment according to the carbon loading related parameters and a long short-term memory network model. The time difference between the future preset moment and the current moment is equal to the duration of the preset time period. The repeated acquisition unit repeatedly performs the acquisition step and the determination step multiple times in sequence to acquire the carbon loading at multiple future preset moments within a continuous time period, and the difference between two historical moments corresponding to the carbon loading related parameters in two adjacent repetitions is a preset difference. The overload determination unit is configured to determine carbon loading overload when at least part of the carbon loading at multiple future preset moments is greater than a preset carbon loading threshold. The present application acquires carbon loading related parameters within a continuous historical time period before and including the current moment of the particulate filter, predicts the carbon loading at a future preset moment based on a long short-term memory network model (LSTM), and combines a preset carbon loading threshold and a preset difference to determine whether there is a carbon loading overload situation in the vehicle, so as to ensure that the particulate filter does not have a carbon loading overload situation, and solves the problem of low accuracy of the prior art in the early judgment scheme for carbon loading overload of the particulate filter.

[0105] As a possible implementation, the device further includes: a model construction unit and a model training unit.

[0106] The model construction unit is configured to construct an initial long short-term memory network model.

[0107] Specifically, the entire initial long short-term memory network model is divided into an input layer, a first hidden layer, a second hidden layer, and an output layer. The training of the neural network consists of multiple iterations, and in each iteration, two steps of forward propagation and backward propagation are required. Operations are performed using the input of the current time step and the output of the previous time step to obtain the output of the current time step, so as to model the relationship between the input and the output, and the historical correlation between the outputs. In addition, during the backward propagation process of the hidden layer, the gradients of each parameter and the hidden state are calculated in sequence. First, the hidden state of the last time step is calculated, then the parameters of the last time step are calculated, and then the hidden state of the penultimate time step is calculated, and so on. After obtaining the gradient of each parameter, each parameter is subtracted by a set multiple of its gradient to complete the backward propagation.

[0108] A model training unit is used to train and test the above initial long short-term memory network model with a training set and a test set to obtain a long short-term memory network model. The above training set includes a first data set and a second data set, the above test set includes a third data set, the above first data set includes overload carbon loading related parameters of multiple carbon loading overloaded vehicles during a historical time period before overload, the above second data set includes non-overload carbon loading related parameters of multiple carbon loading non-overloaded vehicles during the above historical time period, the above third data set includes real-time carbon loading related parameters of a target vehicle, and the engines and the particulate traps of the above multiple carbon loading overloaded vehicles, the above multiple carbon loading non-overloaded vehicles, the above target vehicle and the vehicle to be tested are of the same model.

[0109] Specifically, before model training, it is necessary to perform Kalman filtering and resampling on the obtained training data. The filtering is to filter abnormal signals in the engine. For the second-level sampling interval of the engine, the data volume is too large, which is very computationally resource-intensive and it is difficult to ensure the real-time transmission of data in the digital twin platform. Therefore, the training data is resampled to collect one data every 200s. After constructing the long short-term memory network model, the training set and the test set are used for training and testing to obtain an accurate long short-term memory network model.

[0110] Thus, by using multiple data sets and observing and training the data sets from multiple perspectives, it can help improve the quality and accuracy of the training data set. And by separately retaining the test data set, problems such as data leakage and overfitting can be avoided, so that the trained long short-term memory network model is more accurate and reliable. Through the establishment of the training set and the test set, an effective means is provided for the training and testing of the long short-term memory network model, so that a more accurate and reliable long short-term memory network model can be obtained and provide basic support for subsequent steps.

[0111] As a possible implementation manner, the above device further includes: a model optimization unit and a parameter adoption unit.

[0112] The model optimization unit is used to optimize the parameters in the above initial long short-term memory network model with the mean square error as the loss function using the gradient descent algorithm within a preset hyperparameter range to obtain optimized parameters. The above initial long short-term memory network model includes an input layer, a first hidden layer, a second hidden layer, and an output layer. The above first hidden layer and the above second hidden layer respectively include multiple hidden neurons. The above parameters include: the first target weights corresponding to each input layer node, the first target thresholds corresponding to each first hidden layer node, the second target thresholds corresponding to each second hidden layer node, and the second target weights corresponding to each output layer node;

[0113] Specifically, the initial long short-term memory network model is set with 2 hidden layers, each hidden layer contains 100 neurons, the mean squared error is used as the loss function, and the Adam gradient descent algorithm is used to optimize the weights and thresholds. The learning rate lr is 0.001, and the LSTM model parameters when the loss function is minimized are found through 30 iterations.

[0114] The parameter adoption unit is used to replace the parameters before optimization with the optimized parameters above.

[0115] Thus, it can be seen that when training a deep learning model, the selection of hyperparameters has a very important impact on the performance and effect of the model. Traditional methods often use methods such as grid search to find the optimal hyperparameters, but this method requires a large number of experiments and repeated training, which is very time-consuming and computationally resource-intensive. Using the gradient descent algorithm within the preset hyperparameter range to find the optimal hyperparameters is more practical and feasible. In addition, since the long short-term memory network model contains multiple hidden neurons and multiple target weights and other parameters, the parameters before optimization are often randomly initialized and are usually not optimal. Therefore, when using the gradient descent algorithm to optimize the model parameters, the model parameters can be optimized more carefully and comprehensively, so as to obtain more accurate and effective model parameters. This helps to improve the prediction accuracy and generalization ability of the model, and enhance the practicality and application value of the model.

[0116] As a possible implementation, the above device further includes:

[0117] The model accuracy inspection unit is used to inspect the prediction accuracy of the above initial long short-term memory network model by using preset inspection indicators;

[0118] Specifically, taking the mean squared error MSE and the goodness of fit R 2 as the indicators to inspect the model, the prediction accuracy of the model is evaluated through the preset inspection indicators, and then it is determined whether the model meets the preset accuracy range.

[0119] The model determination unit is used to determine the above initial long short-term memory network model as the above long short-term memory network model when the above prediction accuracy meets the preset accuracy range.

[0120] Thus, it can be seen that if the prediction accuracy of the model meets the preset accuracy requirements, the initial model can be determined as the long short-term memory network model, and then this model can be used for actual applications. Compared with traditional model training methods, it has higher controllability and flexibility. In actual applications, it is often necessary to design inspection indicators according to specific scenarios and requirements, and determine the final model according to the inspection results. Therefore, introducing inspection indicators in the model training process can more effectively improve the prediction accuracy of the model, thereby enhancing the reliability and practicality of the model.

[0121] As a possible implementation, the above-mentioned device further includes:

[0122] A second acquisition unit, configured to acquire model-related parameters of multiple above-mentioned carbon loading overloaded vehicles, multiple above-mentioned carbon loading non-overloaded vehicles, the above-mentioned target vehicle, and the above-mentioned vehicle to be measured. The model-related parameters include: engine-related parameters and particulate filter-related parameters. The engine-related parameters include at least one of the following: cylinder diameter, stroke, design explosion pressure, rated speed, maximum torque speed, rated power, physical and chemical properties of the piston and cylinder liner surface. The particulate filter-related parameters include at least one of the following: model, porosity, pore diameter, mesh number parameter;

[0123] Specifically, acquire the design-related parameters of the engine and key components, including cylinder diameter, stroke, design explosion pressure, rated speed, maximum torque speed, rated power, and physical and chemical properties of the piston and cylinder liner surface, and obtain the model, porosity, pore diameter, and mesh number parameters of the DPF. Ensure that the historical data of the same model engine and DPF are used to train the time series prediction model of carbon loading. In addition, the operating scenarios and uses of the actual vehicles should be kept as consistent as possible to ensure the characteristic of the same distribution of data features, so that the model can converge through training.

[0124] A second determination unit, configured to determine that the engines and the particulate filters of the multiple above-mentioned carbon loading overloaded vehicles, the multiple above-mentioned carbon loading non-overloaded vehicles, the above-mentioned target vehicle, and the above-mentioned vehicle to be measured are of the same model when the above-mentioned model-related parameters of the multiple above-mentioned carbon loading overloaded vehicles, the multiple above-mentioned carbon loading non-overloaded vehicles, the above-mentioned target vehicle, and the above-mentioned vehicle to be measured are the same.

[0125] It can be seen that engines and particulate filters of different types and models have differences in working principles, emission standards, etc. By acquiring the model-related parameters of the vehicle and ensuring that the engine and particulate filter models of each vehicle are the same, the unity and accuracy in training the model can be improved, making the trained model more in line with the actual requirements.

[0126] As a possible implementation, the above-mentioned overload determination unit includes:

[0127] A first determination module, configured to acquire a first duration of the carbon loading at the above-mentioned future preset moment greater than a first preset carbon loading threshold, and determine that the above-mentioned carbon loading overload is a first-level overload when it is determined that the ratio of the above-mentioned first duration to the above-mentioned continuous time period is greater than a preset ratio;

[0128] Specifically, the diagnostic strategy sets different rules according to the severity of the overload. The limit value when the carbon loading is overloaded, the rules when the overload is triggered, and the time ratio can all be set through the rule interaction module. For the first-level overload, which is less severe, false alarms should be avoided as much as possible. Therefore, the carbon loading at a future preset moment predicted should first exceed the first preset carbon loading threshold, and it is determined that the ratio of the above-mentioned continuous duration of the excess to the above-mentioned continuous time period is greater than the preset ratio. For example, the time ratio of overload within 5 consecutive minutes should exceed 60%, triggering the first-level overload.

[0129] A second determination module, configured to obtain a second continuous duration of the carbon loading at the future preset moment that is greater than the second preset carbon loading threshold, and determine that the carbon loading overload is a second-level overload when the second continuous duration is greater than or equal to the preset duration, where the second preset carbon loading threshold is greater than the first preset carbon loading threshold.

[0130] Specifically, since the second-level overload is more severe than the first-level overload and there is a risk of DPF melting and torque limitation, the carbon loading at a future preset moment predicted should first exceed the second preset carbon loading threshold, and the continuous duration of the excess is greater than or equal to the preset duration. For example, if the carbon loading exceeds the second preset carbon loading threshold within 5 consecutive seconds, the second-level overload can be triggered.

[0131] As a possible implementation manner, the above device further includes:

[0132] A display setting unit, configured to set the intelligent operation and maintenance platform for the carbon loading overload, where the intelligent operation and maintenance platform has a plurality of display areas, and the plurality of display areas include a first display area and a second display area;

[0133] Specifically, the intelligent operation and maintenance platform has a plurality of display areas, which can facilitate remote monitoring and management of the vehicle, effectively improving the maintainability and manageability of the vehicle.

[0134] A first display unit, where the first display area is used to display the real-time status parameters of the vehicle to be tested and the carbon loading at a plurality of the future preset moments, and the real-time status parameters include at least one of the following: the carbon loading of the particulate trap calculated in real time using a mechanism model at the current moment, the upstream temperature of the particulate trap at the current moment;

[0135] Specifically, through the intelligent operation and maintenance platform, the real-time status parameters of the vehicle and the carbon loading data at future preset moments can be obtained in real time, the carbon loading overload situation can be detected in time, and accurate positioning and diagnosis can be performed. By using a mechanism model to calculate the carbon loading of the particulate trap and monitoring real-time status parameters such as the upstream temperature, the carbon loading overload situation can be more accurately reflected and classified and displayed, facilitating users to quickly locate problems.

[0136] A second display unit, where the second display area is used to display a first fault code and a second fault code. The first fault code is used to represent the first-level overload, and the second fault code is used to represent the second-level overload.

[0137] Specifically, the first-level overload and the second-level overload are displayed in the form of the first fault code and the second fault code on the intelligent operation and maintenance platform, which is convenient for the monitoring personnel to timely understand the fault situation of the vehicle to be tested, and timely adjust the particulate filter of the vehicle to be tested according to the fault situation, so as to avoid the situation of carbon loading overload in the particulate filter.

[0138] The device for determining carbon loading overload includes a processor and a memory. The obtaining unit, the determining unit, the repeated obtaining unit, the overload determining unit, etc. are all stored in the memory as program units, and the processor executes the program units stored in the memory to implement corresponding functions. All the above modules are located in the same processor; or, the above modules are respectively located in different processors in any combination form.

[0139] The processor contains a kernel, and the kernel retrieves the corresponding program unit from the memory. One or more kernels can be set, and by adjusting the kernel parameters, the problem of low accuracy of the early judgment scheme for carbon loading overload in the particulate filter in the prior art can be solved.

[0140] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of, for example, read-only memory (ROM) or flash memory (flash RAM), and the memory includes at least one storage chip.

[0141] An embodiment of the present invention provides a computer-readable storage medium, and the computer-readable storage medium includes a stored program. When the program runs, it controls the device where the computer-readable storage medium is located to execute the method for determining carbon loading overload.

[0142] Specifically, the method for determining carbon loading overload includes:

[0143] Step S201, obtaining step: obtaining carbon loading related parameters of the particulate filter installed on the vehicle to be tested. The carbon loading related parameters are related parameters obtained within a preset time period, and the preset time period is a time period from a historical moment to the current moment. Among them, the carbon loading related parameters include at least one of the following: the engine speed, the engine torque, the engine fuel injection volume, the engine exhaust gas volume flow, the upstream temperature of the particulate filter, the pressure difference of the particulate filter, and the carbon loading of the particulate filter calculated in real time by using a mechanism model;

[0144] Step S202, determination step: Determine the carbon loading at a future preset moment according to the above carbon loading related parameters and the long short-term memory network model. The time difference between the future preset moment and the current moment is equal to the duration of the preset time period. Among them, the carbon loading related parameters are used as the input of the long short-term memory network model, and the carbon loading at the future preset moment is used as the output of the long short-term memory network model;

[0145] Step S203, repeat the above acquisition step and the above determination step multiple times in sequence to obtain the carbon loadings at multiple future preset moments within a continuous time period, and the difference between the two historical moments corresponding to the carbon loading related parameters in two adjacent repetitions is a preset difference;

[0146] Step S204, when at least part of the carbon loadings at multiple future preset moments are greater than a preset carbon loading threshold, determine that the carbon loading is overloaded.

[0147] Optionally, before determining the carbon loading at a future preset moment according to the above carbon loading related parameters and the long short-term memory network model, the method further includes: constructing an initial long short-term memory network model; training and testing the initial long short-term memory network model with a training set and a test set to obtain the long short-term memory network model. The training set includes a first data set and a second data set, the test set includes a third data set. The first data set includes the overloaded carbon loading related parameters of multiple carbon loading overloaded vehicles during the historical period before overload, the second data set includes the non-overloaded carbon loading related parameters of multiple carbon loading non-overloaded vehicles during the historical period, the third data set includes the real-time carbon loading related parameters of the target vehicle. The engines and the particulate traps of the multiple carbon loading overloaded vehicles, the multiple carbon loading non-overloaded vehicles, the target vehicle and the vehicle to be tested are of the same model.

[0148] Optionally, during the process of training the initial long short-term memory network model with the training set, the method further includes: within a preset hyperparameter range, using the mean square error as the loss function and adopting the gradient descent algorithm to optimize the parameters in the initial long short-term memory network model to obtain optimized parameters. The initial long short-term memory network model includes an input layer, a first hidden layer, a second hidden layer, and an output layer. The first hidden layer and the second hidden layer respectively include multiple hidden neurons. The parameters include: the first target weights corresponding to each input layer node, the first target thresholds corresponding to each first hidden layer node, the second target thresholds corresponding to each second hidden layer node, and the second target weights corresponding to each output layer node; replacing the parameters before optimization with the above optimized parameters.

[0149] Optionally, the above initial long short-term memory network model is trained using a training set to obtain the above long short-term memory network model. The above method further includes: using a preset test index to test the prediction accuracy of the above initial long short-term memory network model; in the case where the above prediction accuracy meets the preset accuracy range, determining the above initial long short-term memory network model as the above long short-term memory network model.

[0150] Optionally, before training and testing the above initial long short-term memory network model using a training set and a test set to obtain a long short-term memory network model, the above method further includes: obtaining model-related parameters of multiple above carbon loading overloaded vehicles, multiple above carbon loading non-overloaded vehicles, the above target vehicle, and the above vehicle to be tested. The above model-related parameters include: engine-related parameters and particulate filter-related parameters. The above engine-related parameters include at least one of the following: cylinder bore, stroke, design explosion pressure, rated speed, maximum torque speed, rated power, physical and chemical properties of the piston and cylinder liner surface. The above particulate filter-related parameters include at least one of the following: model, porosity, pore diameter, mesh number parameter; in the case where the above model-related parameters of multiple above carbon loading overloaded vehicles, multiple above carbon loading non-overloaded vehicles, the above target vehicle, and the above vehicle to be tested are the same, determining that the engines and the above particulate filters of multiple above carbon loading overloaded vehicles, multiple above carbon loading non-overloaded vehicles, the above target vehicle, and the above vehicle to be tested are of the same model.

[0151] Optionally, in the case where at least part of the carbon loading at multiple above future preset times is greater than a preset carbon loading threshold, determining carbon loading overload includes: obtaining a first duration of the carbon loading at the above future preset times greater than a first preset carbon loading threshold, and in the case where it is determined that the ratio of the above first duration to the above continuous time period is greater than a preset ratio, determining that the above carbon loading overload is a first-level overload; obtaining a second duration of the carbon loading at the above future preset times greater than a second preset carbon loading threshold, and in the case where the above second duration is greater than or equal to a preset duration, determining that the above carbon loading overload is a second-level overload, and the above second preset carbon loading threshold is greater than the above first preset carbon loading threshold.

[0152] Optionally, the above method further includes: setting the intelligent operation and maintenance platform with the above carbon loading overload, the intelligent operation and maintenance platform having a plurality of display areas, the plurality of display areas including a first display area and a second display area; the first display area is used to display the real-time status parameters of the vehicle to be tested and the carbon loading at a plurality of the above future preset moments, and the real-time status parameters include at least one of the following: the carbon loading of the particulate filter obtained by real-time calculation using a mechanism model at the current moment, the upstream temperature of the particulate filter at the current moment; the second display area is used to display a first fault code and a second fault code, the first fault code is used to characterize the first-level overload, and the second fault code is used to characterize the second-level overload.

[0153] An embodiment of the present invention provides a processor, which is used to run a program, wherein when the program runs, it executes the method for determining carbon loading overload.

[0154] Specifically, the method for determining carbon loading overload includes:

[0155] Step S201, acquisition step: acquiring the carbon loading related parameters of the particulate filter installed on the vehicle to be tested, the carbon loading related parameters being the related parameters acquired within a preset time period, the preset time period being the time period from a historical moment to the current moment, wherein the carbon loading related parameters include at least one of the following: the engine speed, the engine torque, the engine fuel injection amount, the engine exhaust gas volume flow rate, the upstream temperature of the particulate filter, the differential pressure of the particulate filter, and the carbon loading of the particulate filter obtained by real-time calculation using a mechanism model;

[0156] Step S202, determination step: determining the carbon loading at a future preset moment according to the above carbon loading related parameters and a long short-term memory network model, the time difference between the future preset moment and the current moment being equal to the duration of the preset time period, wherein the carbon loading related parameters serve as the input of the long short-term memory network model, and the carbon loading at the future preset moment serves as the output of the long short-term memory network model;

[0157] Step S203, repeating the above acquisition step and the above determination step multiple times in sequence to obtain the carbon loading at a plurality of the above future preset moments within a continuous time period, and the difference between the two above historical moments corresponding to the carbon loading related parameters in two adjacent repetitions being a preset difference;

[0158] Step S204, when at least part of the carbon loading at a plurality of the above future preset moments is greater than a preset carbon loading threshold, determining carbon loading overload.

[0159] Optionally, before determining the carbon loading at a future preset time according to the above carbon loading-related parameters and the long short-term memory network model, the above method further includes: constructing an initial long short-term memory network model; training and testing the initial long short-term memory network model using a training set and a test set to obtain a long short-term memory network model. The training set includes a first data set and a second data set, and the test set includes a third data set. The first data set includes carbon loading-related parameters of multiple carbon-loaded-over vehicles during a historical period before the overload. The second data set includes carbon loading-related parameters of multiple non-carbon-loaded-over vehicles during the historical period. The third data set includes real-time carbon loading-related parameters of the target vehicle. The engines and the particulate traps of the multiple carbon-loaded-over vehicles, the multiple non-carbon-loaded-over vehicles, the target vehicle, and the vehicle to be measured are of the same model.

[0160] Optionally, during the process of training the initial long short-term memory network model using the training set, the above method further includes: within a preset hyperparameter range, using the mean square error as the loss function and adopting the gradient descent algorithm to optimize the parameters in the initial long short-term memory network model to obtain optimized parameters. The initial long short-term memory network model includes an input layer, a first hidden layer, a second hidden layer, and an output layer. The first hidden layer and the second hidden layer respectively include multiple hidden neurons. The parameters include: the first target weights corresponding to each input layer node, the first target thresholds corresponding to each first hidden layer node, the second target thresholds corresponding to each second hidden layer node, and the second target weights corresponding to each output layer node; replacing the parameters before optimization with the above optimized parameters.

[0161] Optionally, after training the initial long short-term memory network model using the training set to obtain the long short-term memory network model, the above method further includes: using a preset test index to test the prediction accuracy of the initial long short-term memory network model; when the prediction accuracy meets the preset accuracy range, determining the initial long short-term memory network model as the long short-term memory network model.

[0162] Optionally, before training and testing the above initial long short-term memory network model with a training set and a test set to obtain a long short-term memory network model, the above method further includes: obtaining model-related parameters of multiple above-mentioned vehicles with carbon loading overload, multiple above-mentioned vehicles with non-overloaded carbon loading, the above target vehicle, and the above vehicle to be tested. The above model-related parameters include: engine-related parameters and particulate filter-related parameters. The above engine-related parameters include at least one of the following: cylinder diameter, stroke, design explosion pressure, rated speed, maximum torque speed, rated power, physical and chemical properties of the piston and cylinder liner surface. The above particulate filter-related parameters include at least one of the following: model, porosity, pore diameter, mesh parameter; when the above model-related parameters of multiple above-mentioned vehicles with carbon loading overload, multiple above-mentioned vehicles with non-overloaded carbon loading, the above target vehicle, and the above vehicle to be tested are the same, it is determined that the engines and the particulate filters of the multiple above-mentioned vehicles with carbon loading overload, the multiple above-mentioned vehicles with non-overloaded carbon loading, the above target vehicle, and the above vehicle to be tested are of the same model.

[0163] Optionally, when at least part of the carbon loading at multiple above-mentioned future preset times is greater than a preset carbon loading threshold, determining that the carbon loading is overloaded includes: obtaining a first duration of the carbon loading at the above future preset times that is greater than a first preset carbon loading threshold, and when it is determined that the ratio of the above first duration to the above continuous time period is greater than a preset ratio, determining that the above carbon loading overload is a first-level overload; obtaining a second duration of the carbon loading at the above future preset times that is greater than a second preset carbon loading threshold, and when the above second duration is greater than or equal to a preset duration, determining that the above carbon loading overload is a second-level overload, and the above second preset carbon loading threshold is greater than the above first preset carbon loading threshold.

[0164] Optionally, the above method further includes: setting an intelligent operation and maintenance platform for the above carbon loading overload. The intelligent operation and maintenance platform has multiple display areas, and the multiple display areas include a first display area and a second display area; the first display area is used to display the real-time state parameters of the above vehicle to be tested and the carbon loading at multiple above-mentioned future preset times. The real-time state parameters include at least one of the following: the carbon loading of the above particulate filter calculated in real time by a mechanism model at the current moment, the upstream temperature of the above particulate filter at the current moment; the second display area is used to display a first fault code and a second fault code. The first fault code is used to represent the above first-level overload, and the second fault code is used to represent the second-level overload.

[0165] An embodiment of the present invention provides a device, which includes a processor, a memory, and a program stored on the memory and executable on the processor. The device herein can be a server, a PC, a PAD, a mobile phone, etc. When the processor executes the program, it implements the method steps for determining the carbon loading overload described above.

[0166] The present application also provides a computer program product which, when executed on a data processing device, is adapted to execute a program initialized with at least the method steps for determining the overloading of the carbon loading described above.

[0167] Obviously, those skilled in the art should understand that the above-mentioned modules or steps of the present invention can be implemented by a general-purpose computing device. They can be concentrated on a single computing device or distributed on a network composed of multiple computing devices. They can be implemented by program codes executable by the computing device. Thus, they can be stored in a storage device and executed by the computing device. And in some cases, the steps shown or described can be executed in a different order from here, or they can be separately fabricated into individual integrated circuit modules, or multiple modules or steps among them can be fabricated into a single integrated circuit module for implementation. In this way, the present invention is not limited to any specific combination of hardware and software.

[0168] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program codes.

[0169] The present application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as the combination of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processors of general-purpose computers, special-purpose computers, embedded processors, or other programmable data processing devices to generate a machine, such that the instructions executed by the processors of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one or more of the flows Figure 1 or multiple flows and / or blocks

[0170] These computer program instructions can also be stored in a computer-readable memory capable of guiding the computer or other programmable data processing devices to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means, and the instruction means implements the functions specified in Figure 1 one or more of the flows Figure 1 or multiple flows and / or blocks

[0171] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus, causing a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, so that the instructions executed on the computer or other programmable apparatus provide steps for implementing the functions specified in one process or a plurality of processes and / or blocks Figure 1 one process or a plurality of processes and / or blocks Figure 1 steps for implementing the functions specified in one block or a plurality of blocks.

[0172] In a typical configuration, a computing device includes one or more processors (CPUs), an input / output interface, a network interface, and memory.

[0173] The memory may include non-permanent memory in the form of computer-readable media, random access memory (RAM) and / or non-volatile memory such as read-only memory (ROM) or flash memory (flash RAM). The memory is an example of computer-readable media.

[0174] Computer-readable media includes both permanent and non-permanent, removable and non-removable media implemented by any method or technology for storing information. The information may be computer-readable instructions, data structures, program modules, 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, compact disc read-only memory (CD-ROM), digital versatile discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transitory media such as modulated data signals and carrier waves.

[0175] It should also be noted that the term "comprises", "comprising" or any other variation thereof is intended to cover a non-exclusive inclusion, such that a process, method, commodity or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, commodity or device. Without further limitation, an element defined by the statement "comprising an ……" does not exclude the presence of additional identical elements in the process, method, commodity or device comprising the element.

[0176] From the above description, it can be seen that the above embodiments of the present application achieve the following technical effects:

[0177] 1) An embodiment of the present application provides a method for determining carbon loading overload. First, the acquisition step: acquire the carbon loading related parameters of the particulate filter installed on the vehicle to be tested; then the determination step: determine the carbon loading at a future preset time according to the above carbon loading related parameters and the long short-term memory network model; then repeat the above acquisition step and the above determination step multiple times in sequence to obtain the carbon loading at multiple future preset times within a continuous time period, and the difference between the two historical times corresponding to the above carbon loading related parameters in two adjacent repetitions is a preset difference; finally, when at least part of the carbon loading at multiple future preset times is greater than the preset carbon loading threshold, determine that the carbon loading is overloaded. The present application acquires the carbon loading related parameters of the particulate filter at the current time and within a continuous historical time period before the current time, predicts the carbon loading at a future preset time based on the long short-term memory network model (LSTM), and combines the preset carbon loading threshold and the preset difference to determine whether there is a carbon loading overload situation in the vehicle, so as to ensure that the particulate filter does not have a carbon loading overload situation, and solves the problem of low accuracy of the prior judgment scheme for carbon loading overload of the particulate filter.

[0178] 2) An embodiment of the present application provides a device for determining carbon loading overload. The device includes: an acquisition unit, a determination unit, a repeated acquisition unit, and an overload determination unit. The acquisition unit is used to execute the acquisition step to acquire the carbon loading related parameters of the particulate filter installed on the vehicle to be tested. The above carbon loading related parameters are the related parameters acquired within a preset time period, and the preset time period is the time period from a historical time to the current time; the determination unit is used to execute the determination step to determine the carbon loading at a future preset time according to the above carbon loading related parameters and the long short-term memory network model. The time difference between the above future preset time and the above current time is equal to the duration of the preset time period; the repeated acquisition unit repeats the above acquisition step and the above determination step multiple times in sequence to obtain the carbon loading at multiple future preset times within a continuous time period, and the difference between the two historical times corresponding to the above carbon loading related parameters in two adjacent repetitions is a preset difference; the overload determination unit is used to determine that the carbon loading is overloaded when at least part of the carbon loading at multiple future preset times is greater than the preset carbon loading threshold. The present application acquires the carbon loading related parameters of the particulate filter at the current time and within a continuous historical time period before the current time, predicts the carbon loading at a future preset time based on the long short-term memory network model (LSTM), and combines the preset carbon loading threshold and the preset difference to determine whether there is a carbon loading overload situation in the vehicle, so as to ensure that the particulate filter does not have a carbon loading overload situation, and solves the problem of low accuracy of the prior judgment scheme for carbon loading overload of the particulate filter.

[0179] The above are only the preferred embodiments of the present application and are not intended to limit the present application. For those skilled in the art, various changes and modifications can be made to the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included within the protection scope of the present application.

Claims

1. A method for determining carbon loading overload, characterized in that, Including: Obtaining step: Obtain the carbon loading related parameters of the particulate filter installed on the vehicle to be tested. The carbon loading related parameters are the related parameters obtained within a preset time period, and the preset time period is the time period from a historical moment to the current moment. Among them, the carbon loading related parameters include at least one of the following: the engine speed, the engine torque, the engine fuel injection amount, the engine exhaust gas volume flow rate, the upstream temperature of the particulate filter, the differential pressure of the particulate filter, and the carbon loading of the particulate filter calculated in real time using a mechanism model; Determining step: According to the carbon loading related parameters and the long short-term memory network model, determine the carbon loading at a future preset moment. The time difference between the future preset moment and the current moment is equal to the duration of the preset time period. Among them, the carbon loading related parameters are used as the input of the long short-term memory network model, and the carbon loading at the future preset moment is used as the output of the long short-term memory network model; Repeat the obtaining step and the determining step multiple times in sequence to obtain the carbon loadings at multiple future preset moments within a continuous time period, and the difference between the two historical moments corresponding to the carbon loading related parameters in two adjacent repetitions is a preset difference; When at least part of the carbon loadings at multiple future preset moments are greater than a preset carbon loading threshold, determine that the carbon loading is overloaded.

2. The method according to claim 1, wherein Before determining the carbon loading at a future preset moment according to the carbon loading related parameters and the long short-term memory network model, the method further includes: Construct an initial long short-term memory network model; Train and test the initial long short-term memory network model using a training set and a test set to obtain a long short-term memory network model. The training set includes a first data set and a second data set, and the test set includes a third data set. The first data set includes the overload carbon loading related parameters of multiple carbon loading overloaded vehicles within the historical time period before overload. The second data set includes the non-overload carbon loading related parameters of multiple carbon loading non-overloaded vehicles within the historical time period. The third data set includes the real-time carbon loading related parameters of the target vehicle. The engines and the particulate filters of the multiple carbon loading overloaded vehicles, the multiple carbon loading non-overloaded vehicles, the target vehicle, and the vehicle to be tested are of the same model.

3. The method according to claim 2, wherein During the process of training the initial long short-term memory network model using the training set, the method further includes: Within a preset hyperparameter range, use the mean square error as the loss function and adopt the gradient descent algorithm to optimize the parameters in the initial long short-term memory network model to obtain optimized parameters. The initial long short-term memory network model includes an input layer, a first hidden layer, a second hidden layer, and an output layer. The first hidden layer and the second hidden layer respectively include multiple hidden neurons. The parameters include: the first target weights corresponding to each input layer node, the first target thresholds corresponding to each first hidden layer node, the second target thresholds corresponding to each second hidden layer node, and the second target weights corresponding to each output layer node; Replace the parameters before optimization with the optimized parameters.

4. The method according to claim 2, wherein Training the initial long short-term memory network model with a training set to obtain the long short-term memory network model, the method further comprising: Using a preset test index to test the prediction accuracy of the initial long short-term memory network model; When the prediction accuracy meets a preset accuracy range, determining the initial long short-term memory network model as the long short-term memory network model.

5. The method according to claim 2, characterized in that, Before training and testing the initial long short-term memory network model with a training set and a test set to obtain a long short-term memory network model, the method further comprises: Obtaining model-related parameters of multiple carbon loading overloaded vehicles, multiple carbon loading non-overloaded vehicles, the target vehicle, and the vehicle to be tested, the model-related parameters including: engine-related parameters and particulate filter-related parameters, the engine-related parameters including at least one of the following: cylinder diameter, stroke, design explosion pressure, rated speed, maximum torque speed, rated power, physical and chemical properties of the piston and cylinder liner surface, and the particulate filter-related parameters including at least one of the following: model, porosity, pore diameter, mesh number parameter; When the model-related parameters of multiple carbon loading overloaded vehicles, multiple carbon loading non-overloaded vehicles, the target vehicle, and the vehicle to be tested are the same, determining that the engines and the particulate filters of multiple carbon loading overloaded vehicles, multiple carbon loading non-overloaded vehicles, the target vehicle, and the vehicle to be tested are of the same model.

6. The method according to claim 1, wherein When at least part of the carbon loading at multiple future preset times is greater than a preset carbon loading threshold, determining carbon loading overload, including: Obtaining a first duration of the carbon loading at the future preset time greater than a first preset carbon loading threshold, and when determining that the ratio of the first duration to the continuous time period is greater than a preset ratio, determining that the carbon loading overload is a first-level overload; Obtaining a second duration of the carbon loading at the future preset time greater than a second preset carbon loading threshold, and when the second duration is greater than or equal to a preset duration, determining that the carbon loading overload is a second-level overload, the second preset carbon loading threshold being greater than the first preset carbon loading threshold.

7. The method according to claim 6, the method further comprising: Setting an intelligent operation and maintenance platform for carbon loading overload, the intelligent operation and maintenance platform having a plurality of display areas, the plurality of display areas including a first display area and a second display area; The first display area is used to display the real-time state parameters of the vehicle to be tested and the carbon loading at multiple future preset times, the real-time state parameters including at least one of the following: the carbon loading of the particulate filter calculated in real time by a mechanism model at the current moment, the upstream temperature of the particulate filter at the current moment; The second display area is used to display a first fault code and a second fault code, the first fault code being used to characterize the first-level overload, and the second fault code being used to characterize the second-level overload.

8. An apparatus for determining carbon loading overload, characterized in that, The device includes: An acquisition unit, configured to perform an acquisition step of acquiring parameters related to the carbon loading of a particulate trap installed on a vehicle to be tested, where the parameters related to the carbon loading are parameters acquired within a preset time period, and the preset time period is a time period from a historical moment to the current moment. The parameters related to the carbon loading include at least one of the following: the rotational speed of the engine, the torque of the engine, the fuel injection amount of the engine, the exhaust gas volume flow rate of the engine, the upstream temperature of the particulate trap, the pressure difference of the particulate trap, and the carbon loading of the particulate trap calculated in real time using a mechanism model; A determination unit, configured to perform a determination step of determining the carbon loading at a future preset moment according to the parameters related to the carbon loading and a long short-term memory network model. The time difference between the future preset moment and the current moment is equal to the duration of the preset time period. The parameters related to the carbon loading are used as the input of the long short-term memory network model, and the carbon loading at the future preset moment is used as the output of the long short-term memory network model; A repeated acquisition unit, configured to repeatedly perform the acquisition step and the determination step multiple times in sequence to acquire the carbon loading at multiple future preset moments within a continuous time period, and the difference between two historical moments corresponding to the parameters related to the carbon loading in two adjacent repetitions is a preset difference; An overload determination unit, configured to determine that the carbon loading is overloaded when at least part of the carbon loading at multiple future preset moments is greater than a preset carbon loading threshold.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, where when the program runs, it controls the device where the computer-readable storage medium is located to execute the method for determining carbon loading overload according to any one of claims 1 to 7.

10. An electronic device, characterized in that, Comprising: One or more processors, a memory, and one or more programs, where the one or more programs are stored in the memory and are configured to be executed by the one or more processors. The one or more programs include a method for executing the determination of carbon loading overload according to any one of claims 1 to 7.

Images