Automobile exhaust analysis method through monitoring of automobile exhaust components

Abstract

The present invention discloses an automobile exhaust analysis method through monitoring of automobile exhaust gas components, which relates to the technical field of data processing. The method includes: constructing a monitoring data set of exhaust gas components based on the operating state of the engine, performing category uniformity analysis and data set expansion on it to obtain a target data set; performing sample annotation on the target data set to construct a training data set; building a model framework, and performing model training convergence through the training data set to obtain an exhaust analysis model; configuring exhaust gas detection equipment based on a preset processing target to perform target detection; inputting the engine monitoring parameters and the target detection results into the exhaust analysis model to determine the exhaust identification result. The present invention solves the technical problem in the prior art that due to the variability of the automobile operating state and the complexity of the emission data, the exhaust gas emission control effect is poor, and achieves the technical effect of improving the engine performance and reducing the exhaust gas emission through dynamic monitoring of exhaust gas emissions and optimization of engine control.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly to an automobile exhaust analysis method through monitoring of automobile exhaust components. Background Art

[0002] Automobiles are the most common means of transportation in modern life. At the same time, the pollution of automobile exhaust to the environment has gradually become a negative factor affecting the quality of human life. Automobile exhaust analysis is an important means to evaluate engine performance and reduce exhaust emissions. Through exhaust analysis, it is possible to judge the working state of the engine and whether the emissions of the automobile meet the standards, thereby reducing the pollution of automobile exhaust to the atmosphere at the source. However, due to the variability of the operating state of the automobile and the large variety and complexity of emission data, the accuracy of emission analysis is low, and the effect of controlling exhaust emissions is poor. Summary of the Invention

[0003] This application provides an automobile exhaust analysis method through monitoring of automobile exhaust components, which is used to solve the technical problem in the prior art that due to the variability of the operating state of the automobile and the complexity of emission data, the effect of controlling exhaust emissions is poor.

[0004] In the first aspect of this application, an automobile exhaust analysis method through monitoring of automobile exhaust components is provided. The method includes: constructing a tail gas component monitoring data set based on the engine operating state, where the tail gas component monitoring data set includes tail gas components and engine state parameters; performing category uniformity analysis on the tail gas component monitoring data set, and expanding the data set based on the category uniformity analysis result to obtain a target data set; performing sample annotation on the target data set according to a preset processing target to construct a training data set; building a model framework, and performing model training convergence through the training data set to obtain an exhaust analysis model; configuring a tail gas detection device according to the preset processing target, and performing target detection on the automobile tail gas through the tail gas detection device to obtain a target detection result; inputting the engine monitoring parameters and the target detection result into the exhaust analysis model to analyze the exhaust state of the engine and determine an exhaust recognition result.

[0005] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0006] The vehicle exhaust analysis method through vehicle exhaust component monitoring provided by this application relates to the technical field of data processing. By the engine operating state, a vehicle exhaust component monitoring data set is constructed. After performing category uniformity analysis and data set expansion on it, it is used as training data to train an exhaust analysis model. The engine monitoring parameters and vehicle exhaust detection results are input into the exhaust analysis model to determine the exhaust identification result, solving the technical problem in the prior art that due to the variability of vehicle operating states and the complexity of emission data, the vehicle exhaust emission control effect is poor, and achieving the technical effect of realizing dynamic control optimization of vehicle exhaust emissions through real-time monitoring of vehicle exhaust to improve engine performance and reduce vehicle exhaust emissions. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for description in the embodiments. Obviously, the following-described drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0008] Figure 1 Schematic flow chart of the vehicle exhaust analysis method through vehicle exhaust component monitoring provided by the embodiments of this application;

[0009] Figure 2 Schematic flow chart of obtaining the engine operating state in the vehicle exhaust analysis method through vehicle exhaust component monitoring provided by the embodiments of this application;

[0010] Figure 3 Schematic flow chart of configuring vehicle exhaust detection equipment in the vehicle exhaust analysis method through vehicle exhaust component monitoring provided by the embodiments of this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0011] This application provides a vehicle exhaust analysis method through vehicle exhaust component monitoring, which is used to solve the technical problem in the prior art that due to the variability of vehicle operating states and the complexity of emission data, the vehicle exhaust emission control effect is poor.

[0012] The following will clearly and completely describe the technical solutions in the embodiments of this application with reference to the drawings in the embodiments of this application. Obviously, the described embodiments are only some embodiments of this application, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in this application without creative efforts fall within the scope of protection of this application.

[0013] It should be noted that the terms "first", "second", etc. in the description of this application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present application described here can be implemented in an order other than those illustrated or described here. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or server that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products or devices.

[0014] Embodiment 1

[0015] As Figure 1 shown, the present application provides an automobile exhaust analysis method through monitoring the components of automobile exhaust. The method includes:

[0016] P10: Based on the engine operating state, construct a tail gas component monitoring data set, where the tail gas component monitoring data set includes tail gas components and engine state parameters;

[0017] Specifically, according to the engine operating state, construct a tail gas component monitoring data set, collect and collate detailed information on tail gas components and engine state parameters, and provide a data basis for subsequent exhaust analysis. Among them, the tail gas component monitoring data set includes tail gas components and engine state parameters. On the one hand, the tail gas components can be measured in real time through professional tail gas detection equipment. The tail gas components refer to the chemical substances contained in the gases emitted by the automobile engine, such as carbon monoxide, carbon dioxide, nitrogen oxides, hydrocarbons, etc., and are important indicators for evaluating the engine performance and whether the tail gas emission standards are met. On the other hand, the engine state parameters can be collected in real time through sensors on the engine, including various parameters describing the engine operating state, such as speed, load, temperature, fuel consumption rate, etc., which can reflect the operating conditions of the engine under different working conditions and the changes in the corresponding tail gas components.

[0018] Exemplarily, a dedicated exhaust gas monitoring device is used to collect the exhaust gas of the engine in real time under different operating conditions. The exhaust gas monitoring device includes a gas analyzer, a sensor module, etc., and can be used to monitor the main components such as CO (carbon oxides), NOx (nitrogen oxides), and HC (hydrocarbons) in the exhaust gas. Each sensor is connected to the vehicle exhaust pipe through a pipeline and can detect the emission components in real time when the engine is running. At the same time, a sensor module installed on the engine is used to collect the operating parameters of the engine. These parameters include engine speed, fuel consumption rate, load, exhaust temperature, etc., and are used to describe the performance of the engine under different operating states. These parameters are obtained in real time through a data collector to ensure the continuity and accuracy of the data. Finally, based on the engine operating state, the collected exhaust gas components are correlated with the engine operating parameters to construct an exhaust gas component monitoring data set. The constructed data set covers the exhaust characteristics and engine state parameters under different operating conditions such as idling, accelerating, decelerating, and constant speed.

[0019] Further, as Figure 2 shown, before constructing the exhaust gas component monitoring data set, the embodiment of the present application further includes step P10a, and step P10a further includes:

[0020] P11a: Configure the first engine operating state according to the instantaneous, intermittent, and continuous emission states of the exhaust gas;

[0021] P12a: Analyze the relationship between exhaust and power according to the engine speed and torsional strength, and configure the second engine operating state;

[0022] P13a: Extract the operating state characteristics from the first engine operating state and the second engine operating state, and perform random combination of the characteristics to construct a reconstructed state. Perform an engine operation evaluation on the reconstructed state, use the reconstructed state that meets the operation evaluation as a new operating state, and use the reconstructed state that does not meet the operation evaluation as a termination operating state;

[0023] P14a: Based on the new operating state, search and derive according to a set step size, and continue to perform an engine operation evaluation, and so on until there is no new operating state;

[0024] P15a: Summarize all the new operating states, the first engine operating state, and the second engine operating state to obtain the engine operating state.

[0025] It should be understood that before constructing the exhaust gas component monitoring data set, the engine operating state needs to be obtained. Specifically, according to the instantaneous, intermittent, and continuous emission states of the exhaust gas, the normal operating state of the engine is configured, and this is used as the first engine operating state. Among them, instantaneous emissions usually occur when the engine starts or accelerates, intermittent emissions may occur under specific working conditions, and continuous emissions are the states when the engine is running stably. Different emission states will directly affect the exhaust gas components and concentrations, so it is necessary to configure the corresponding engine operating state for each state.

[0026] Furthermore, by deeply analyzing the engine speed and torsional strength, where the torsional strength is the load capacity of the engine, the relationship between exhaust gas and power is obtained, that is, the mapping relationship between the engine speed and load and the exhaust gas emissions is quantitatively analyzed, and the exhaust gas components and concentrations under different engine speeds and loads are obtained, resulting in the second engine operating state.

[0027] Furthermore, operating state features are extracted from the first engine operating state and the second engine operating state. The operating state features can include the engine speed, load, temperature, fuel consumption rate, etc., which can reflect the performance of the engine under different operating states. Further, the extracted features are randomly grouped to generate a series of new reconstructed states with different characteristics, which can reflect various different working conditions and conditions that the engine may encounter during actual operation.

[0028] Furthermore, an engine operation evaluation is carried out for each reconstructed state, including a comprehensive evaluation in multiple aspects such as engine performance, emission level, and fuel economy. The reconstructed states that meet the operation evaluation are used as new added operating states, and the reconstructed states that do not meet the operation evaluation are used as terminated operating states. Further, for the reconstructed states that meet the operation evaluation, a search and derivation are carried out according to a set step size to further expand the range of the engine operating state, and the newly derived states are continuously subjected to engine operation evaluation, and so on and iterate continuously until there are no new added operating states that meet the evaluation criteria.

[0029] Finally, all the newly added operating states, the first engine operating state, and the second engine operating state are summarized to obtain a comprehensive engine operating state set, which is used as the basis for constructing the subsequent exhaust gas component monitoring data set to ensure that the exhaust gas component monitoring data set can cover the exhaust gas emissions of the engine under various operating states.

[0030] P20: Perform a category uniformity analysis on the exhaust gas component monitoring data set, and based on the category uniformity analysis result, expand the data set to obtain the target data set;

[0031] Furthermore, step P20 of the embodiment of the present application further includes:

[0032] P21: Based on the engine operating state, classify the exhaust gas component monitoring data set to obtain clusters of each state category;

[0033] P22: Decentralize the number of each state category cluster to determine the sample deviation;

[0034] P23: For those that do not meet the preset uniformity requirements, the state category clusters are supplemented to obtain the target data set. The state category cluster supplementation includes: performing genetic fission based on the original sample data and obtaining a similar sample set of the same family for correction and adjustment.

[0035] Optionally, the exhaust gas component monitoring data set is subjected to a process of class uniformity analysis and data set expansion to ensure that the number of samples of each category in the data set is relatively balanced, thereby improving the accuracy and generalization ability of subsequent model training. Specifically, based on the engine operating state, the exhaust gas component monitoring data set is classified according to different engine operating states to obtain each state category cluster, wherein each state category cluster contains multiple exhaust gas component data with similar engine operating states.

[0036] Furthermore, the number of each state category cluster is decentralized to evaluate the distribution of the number of samples in each cluster and determine the sample deviation. The decentralized processing is a data preprocessing technology used to eliminate the offset or trend in the data set so that the data is more in line with the normal distribution. The sample deviation can reflect the imbalance of the number of samples in each state category cluster. If the number of samples in some state categories is too small, it is supplemented by the data set expansion technology to ensure that the data set has enough samples to support model training.

[0037] Further, the uniformity of each state category cluster is determined with reference to the sample deviation degree. If the number of samples in any state category cluster does not meet the preset uniformity requirement, for example, the number of samples in some categories is too small, then state category cluster supplementation is required. The supplementation method may include genetic fission based on the original sample data, that is, generating new sample data through operations such as replication and mutation. For example, for the category cluster that does not meet the uniformity requirement, a genetic fission algorithm is applied for expansion. Some samples in the cluster are selected. New samples are generated using crossover and mutation operations. Crossover is to take partial features from two samples and combine them into a new sample; mutation is to apply random perturbations to the sample features to generate mutant samples. Samples similar to the original samples are found through the K-Nearest Neighbor (KNN) method for deviation correction adjustment to ensure that the generated samples are similar in features to the original samples. The category uniformity analysis is performed again on the expanded dataset to verify whether the number of samples meets the balance requirement and ensure that there are sufficient samples in the dataset under various operating states. Further, a set of similar samples of the same family is obtained for deviation correction adjustment, that is, more samples are obtained from the similar dataset to supplement the insufficient categories. By analyzing the similarity between each state category cluster, samples are extracted from the similar clusters for data correction and expansion. For example, for the insufficient exhaust gas data under high load conditions, samples can be extracted from the similar medium load state cluster and interpolation and adjustment are performed using machine learning methods to meet the data expansion requirements. Thus, the number of samples in the insufficient categories is effectively increased, making the number of samples in each category in the dataset more balanced to obtain the target dataset.

[0038] P30: Sample annotation is performed on the target dataset according to a preset processing target to construct a training dataset;

[0039] Further, step P30 of the embodiment of the present application further includes:

[0040] P31: The preset processing target includes the target components of the exhaust gas and the target state parameters of the engine. Sample screening is performed in the target dataset according to the preset processing target, and the preset processing target is used as the identification target for sample annotation;

[0041] P32: Positive and negative sample categories are set according to the preset processing target, and positive and negative samples of the target dataset are constructed according to the category setting requirements. The positive sample has a direct recognition relationship with the preset processing target, and the negative sample has an opposite recognition relationship with the positive sample.

[0042] Specifically, the target dataset is sample-annotated according to a preset processing target, which is usually set according to task requirements and includes target exhaust components such as carbon monoxide and nitrogen oxides, as well as target engine state parameters such as speed and load. Sample screening is performed on the target dataset according to the preset processing target. By determining whether the content of the exhaust components in the sample reaches a threshold or whether the engine state parameters meet specific conditions, qualified samples are screened out, and the corresponding preset processing target is used as the identification target to annotate the screened samples, obtaining a screened target dataset. The annotation can be binary, for example, the presence or absence of the target component, or continuous, for example, the specific content of the target component or the specific value of the engine state parameter.

[0043] Further, positive and negative sample categories are set according to the preset processing target. Among them, the positive sample usually refers to a sample that meets certain conditions or has certain characteristics, and the positive sample has a direct recognition relationship with the preset processing target, while the negative sample is a sample opposite to the positive sample. Exemplarily, if the preset processing target is to detect whether the carbon monoxide content in the exhaust exceeds the standard, then the sample with the carbon monoxide content exceeding the threshold can be set as the positive sample, and the sample with the content lower than the threshold can be set as the negative sample. Further, the samples in the screened target dataset are classified according to the category setting requirements to construct positive and negative samples, obtaining a training dataset for training the model to distinguish different sample categories.

[0044] P40: Build a model framework and perform model training convergence through the training dataset to obtain an exhaust analysis model;

[0045] Further, step P40 of the embodiment of the present application further includes:

[0046] P41: Building the model framework includes an identification and screening layer, an input layer, a hidden layer, an output layer, and a fully connected layer;

[0047] P42: Set an identification and screening target based on the target dataset, and obtain samples with annotations matching the target dataset through the input layer;

[0048] P43: Use the training data entering the input layer to train the hidden layer, obtain the analysis result and output it through the output layer. The fully connected layer connects all neurons to obtain an exhaust identification result, which is used to describe the exhaust result caused by the engine operating parameters. Among them, a convergence target is set based on the training dataset, and a target test is performed according to the output result. When the convergence target is met, the training is completed.

[0049] It should be understood that, based on the principles of machine learning, a model framework is built. The model framework includes an identification and screening layer, an input layer, a hidden layer, an output layer, and a fully connected layer. Among them, the identification and screening layer is responsible for screening out samples that match the preset processing target from the target dataset, ensuring that only relevant samples enter the model for training. The input layer is used to receive the screened samples as input data. The hidden layer is the core part of the model and is responsible for learning and extracting the features of the data. The output layer outputs the learning results of the hidden layer in a specific form. The fully connected layer is responsible for connecting all neurons to obtain the final exhaust identification result.

[0050] Furthermore, according to the characteristics of the target dataset and the preset processing target, an identification and screening target is set, such as specific exhaust gas components, engine state parameters, etc. Then, samples that match the identification and screening target are screened out from the target dataset and enter the model through the input layer for training.

[0051] Furthermore, the hidden layer is trained using the training data that enters the input layer, and a convergence target is set based on the characteristics of the training dataset and the requirements of the model performance. The convergence target can include the accuracy rate of the model output data and the number of iterations. Specifically, the training data is analyzed by the hidden layer, the analysis results are output through the output layer, and by comparing the difference between the output data and the expected data, the network parameters of the model are adjusted. This process is repeated for multiple iterations of training and network parameter adjustment until the output result reaches the convergence target, completing the training of the hidden layer. All neurons are connected through the fully connected layer to obtain the exhaust analysis model. The exhaust identification result can be obtained through the exhaust analysis model, and the exhaust identification result is used to describe the exhaust result caused by the engine operating parameters.

[0052] Exemplarily, the specific process of model training can be as follows. First, from the augmented target dataset, specific exhaust gas components (such as carbon monoxide, carbon dioxide) and engine state parameters (such as speed, load) are selected as preset processing targets, and positive example samples (samples that meet the targets) and negative example samples (samples that do not meet the targets) are labeled to construct a training dataset. Specifically, positive example samples are samples that meet the emission standards or optimization targets, and negative example samples are samples that do not meet the standards. The labeling process is based on whether the samples meet these preset conditions. Next, a neural network framework for the exhaust analysis model is built. The model includes an input layer, a hidden layer, an output layer, and a fully connected layer. The labeled training data is input into the model through the input layer, and the data includes exhaust gas components and engine state parameters. Through the hidden layer, the convolutional neural network (CNN) in deep learning is used to extract and learn data features, and the implicit associations between each state and the exhaust are extracted through inter-layer operations. Through the output layer, the exhaust analysis result is output according to the input data. The output of the model is the recognition result of the exhaust gas components, including the concentration values of each component and whether the emission status meets the standards.

[0053] Furthermore, set the convergence target for model training, such as the classification accuracy reaching 90%, or the loss function value dropping below 0.01. Input the training dataset into the model for training, and gradually adjust the model parameters. During the training process, the model is iteratively trained multiple times according to the loss function and the set convergence target. After each iteration, the model is evaluated through the validation set, and the accuracy and loss function value of the model are calculated. When the model reaches the convergence target, the training ends, and the final model is saved.

[0054] P50: Based on the preset processing target, configure the exhaust gas detection device, and perform target detection on the vehicle exhaust through the exhaust gas detection device to obtain the target detection result;

[0055] Furthermore, as Figure 3 shown, step P50 of the embodiment of the present application further includes:

[0056] P51: Perform target recognition and extraction according to the preset processing target to obtain target features, and the target features include exhaust gas component features and engine state features;

[0057] P52: When the target feature is the exhaust gas component feature, configure the detection component parameters of the exhaust gas detection device based on the exhaust gas component;

[0058] P53: When the target feature is the engine state feature, configure the activation instruction of the exhaust gas detection device, and the activation instruction is used to start the exhaust gas detection device when the engine operating state parameter meets the activation target feature.

[0059] Exemplarily, the tail gas detection device is configured according to the preset processing target, and the target detection of the vehicle tail gas is carried out by the tail gas detection device to obtain the target detection result. Specifically, target recognition and extraction are carried out according to the preset processing target to obtain target features related to tail gas detection and engine status. The target features include tail gas component features, such as the content of harmful substances such as carbon monoxide and nitrogen oxides, and engine status features, such as speed, load, temperature, etc.

[0060] Further, when the target feature is the tail gas component feature, the detection component parameters of the tail gas detection device are configured according to the tail gas components, that is, according to the tail gas components specified in the preset processing target, the detection range and sensitivity of the tail gas detection device are adjusted to ensure that the content of the specified tail gas components can be accurately measured, and the actual emission data of specific components in the vehicle tail gas can be obtained in real time.

[0061] Further, when the target feature is the engine status feature, an activation instruction for the tail gas detection device is configured. The activation instruction is used to automatically start the tail gas detection device when the engine operation state parameters meet the activation target feature. For example, it can be set that when the engine speed reaches the speed threshold or the load exceeds the load range, the tail gas detection device automatically starts to work, so as to ensure that the tail gas detection device can capture tail gas data in time when the engine is in a critical operation state, so as to more comprehensively understand the impact of engine performance on tail gas emissions.

[0062] P60: Input the engine monitoring parameters and the target detection result into the exhaust analysis model to analyze the engine exhaust state and determine the exhaust recognition result.

[0063] Optionally, engine monitoring parameters, such as engine speed, load, fuel consumption rate, intake and exhaust temperature, etc., and the target detection result are input into the exhaust analysis model. The exhaust analysis model analyzes the engine exhaust state, identifies key information related to the exhaust state in the input data, and accordingly obtains the exhaust recognition result. The exhaust recognition result can reflect the engine exhaust state, and may include specific content of tail gas components, whether the emission level meets the standard, whether there are potential faults or abnormalities, etc. It has important guiding significance for engine optimization, formulation of tail gas treatment strategies, and fault prevention.

[0064] Further, the embodiment of the present application further includes step P70, and step P70 further includes:

[0065] P71: Determine the exhaust relationship between the engine monitoring parameters and the tail gas components according to the exhaust recognition result;

[0066] P72: Collect the running speed and road conditions of the vehicle, analyze the matching relationship based on the engine monitoring parameters, the running speed of the vehicle, and the road conditions, and determine the useless performance loss.

[0067] P73: Trace back the engine based on the useless performance loss to determine the power control variable; analyze the exhaust relationship between the engine monitoring parameters and the exhaust gas components to determine the exhaust gas control variable.

[0068] P74: Optimize the power loss and exhaust gas components according to the power control variable and the exhaust gas control variable, obtain the optimized engine control strategy for feedback, and use it to guide the driver to perform engine operation and fuel injection adjustment control.

[0069] Specifically, based on the exhaust recognition result, deeply analyze the exhaust relationship between the engine monitoring parameters and the exhaust gas components, including analyzing the specific impact of different engine monitoring parameters on the exhaust gas components, as well as the interaction and correlation between different engine monitoring parameters and the exhaust gas components, to accurately understand how the engine operating state affects the exhaust emissions.

[0070] Furthermore, collect information such as the running speed and road conditions of the vehicle, perform a matching relationship analysis with the engine monitoring parameters, and identify the useless performance loss caused by poor engine operation or road conditions by comparing the actual operating parameters with the parameters under the ideal state. The useless performance loss may include power waste, increased fuel consumption, etc., which have a negative impact on both the engine performance and the exhaust emissions.

[0071] Furthermore, trace back the engine according to the useless performance loss, analyze the specific reasons for the loss, and determine the power control variable. The power control variable may include adjusting the engine intake and exhaust system, optimizing the fuel injection strategy, etc., aiming to improve the efficiency and performance of the engine. At the same time, analyze the exhaust gas control variable according to the exhaust relationship between the engine monitoring parameters and the exhaust gas components, and determine the exhaust gas control variable, that is, determine the engine operating parameter adjustment amount to reduce harmful exhaust emissions.

[0072] Furthermore, based on the power control variables and exhaust gas control variables, power loss and exhaust gas composition are optimized, and corresponding engine optimization control strategies are formulated. Exemplarily, the linear regression method is used to fit the relationship between engine power loss and control variables, and a fitness function is set. With the goal of minimizing power loss, the control variables are adjusted to achieve optimal power output. Through the fitting relationship between the exhaust gas identification result and the optimization of exhaust gas composition, the operating parameters of the exhaust gas treatment equipment (such as the EGR opening) are adjusted to ensure that the exhaust gas composition meets the standards. For example, the engine speed, load, fuel injection volume, etc. are adjusted to minimize power loss and optimize exhaust gas composition, and the engine optimization control strategy is fed back to guide the driver to perform engine operation and fuel injection adjustment control, so as to improve engine performance and reduce exhaust gas emissions during actual operation.

[0073] Furthermore, step P74 of the embodiment of the present application further includes:

[0074] P74-1: Based on the running speed and road condition status, a sample set is collected, and a relationship fitting is performed with the power control variable as the independent variable and power loss as the dependent variable to determine the first fitting relationship;

[0075] P74-2: Based on the exhaust gas analysis model, model parameters are extracted to determine the second fitting relationship between the exhaust gas control variable and exhaust gas composition;

[0076] P74-3: Set the optimization coefficients for power loss and exhaust gas composition optimization, synthesize the first fitting relationship and the second fitting relationship, construct a fitness function, and establish an optimization space with the maximum total target as the optimization goal;

[0077] P74-4: Set the exhaust gas composition optimization target as a constraint condition and add it to the optimization space. Randomly select a combination of power control variables and exhaust gas control variables. Where there are overlapping variables in the power control variables and exhaust gas control variables, unified values are taken. Based on the fitness function, evaluate the random combination as the current optimization strategy;

[0078] P74-5: Randomly select a second combination of power control variables and exhaust gas control variables, perform fitness evaluation, compare it with the evaluation result of the current optimization strategy, select the combination with the larger evaluation result as the current optimization strategy, and repeat the iteration until the optimization goal is met or the iteration times are reached, and obtain the optimal optimization strategy for feedback.

[0079] Optionally, based on the running speed and road condition status, a sample set is collected, and a relationship fitting is performed with the power control variable as the independent variable and the power loss as the dependent variable. Through statistical or machine learning methods, the fitting relationship between the power control variable and the power loss, that is, the first fitting relationship, is determined. Further, based on the exhaust analysis model, model parameters are extracted to quantify the impact of the exhaust gas control variable on exhaust emissions, and the second fitting relationship between the exhaust gas control variable and the exhaust gas components is determined.

[0080] Further, optimization coefficients for power loss and exhaust gas component optimization are set. The optimization coefficients represent the degree of emphasis on different optimization objectives. Further, the first fitting relationship and the second fitting relationship are integrated to construct a fitness function, which is used to evaluate the exhaust emission optimization effect under different combinations of control variables. Finally, an optimization space is established with the maximum of the target sum as the optimization objective, that is, a parameter optimization range is established with the power loss and the exhaust gas components reaching the optimal simultaneously as the optimization objective.

[0081] Further, the exhaust gas component optimization objective is set as a constraint condition and added to the optimization space to ensure that when searching for the optimal control strategy, the selected strategy can meet the requirements of exhaust emissions. Randomly select a combination of the power control variable and the exhaust gas control variable as the initial optimization strategy. If there are overlapping variables among the power control variable and the exhaust gas control variable, it is necessary to ensure that these variables remain unified when randomly taking values to ensure the consistency of the strategy. Finally, evaluate the random combination according to the previously constructed fitness function, and use this evaluation result as the current optimization strategy.

[0082] Further, randomly select a second combination of the power control variable and the exhaust gas control variable, perform a fitness evaluation, and compare it with the evaluation result of the current optimization strategy. If the evaluation result of the new strategy is better than that of the current optimization strategy, replace the current optimization strategy with the new strategy; otherwise, retain the current optimization strategy and continue the next iteration. And so on, continuously perform optimization iterations until the optimization objective is met, that is, the power loss reaches the minimum and the exhaust gas components meet the optimization requirements, or until the preset number of iterations is reached. An optimal optimization strategy that meets all constraint conditions and can minimize the power loss is obtained. Feed the optimization strategy back to the driver or the engine control system to improve the engine performance and reduce exhaust emissions.

[0083] In summary, the embodiments of the present application have at least the following technical effects:

[0084] Based on the operating state of the engine, this application constructs a monitoring dataset of exhaust gas components, conducts category uniformity analysis and dataset expansion on it, and then uses it as a training dataset for model training convergence to obtain an exhaust gas analysis model. Based on a preset processing objective, exhaust gas detection equipment is configured for target detection, and the engine monitoring parameters and target detection results are input into the exhaust gas analysis model to determine the exhaust gas identification result.

[0085] It achieves the technical effect of improving the engine performance and reducing exhaust gas emissions through dynamic monitoring of exhaust gas emissions and engine control optimization.

[0086] It should be noted that the above sequence of embodiments of this application is only for description and does not represent the superiority or inferiority of the embodiments. And the above specific embodiments of this specification have been described. Additionally, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0087] The above are only the preferred embodiments of this application and are not intended to limit this application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of this application shall be included within the protection scope of this application.

[0088] This specification and the drawings are only exemplary descriptions of this application and are considered to have covered any and all modifications, variations, combinations, or equivalents within the scope of this application. Obviously, those skilled in the art can make various changes and modifications to this application without departing from the scope of this application. Thus, if these modifications and variations of this application fall within the scope of this application and its equivalent technologies, this application is intended to include these changes and modifications.

Claims

1. A method for analyzing automobile exhaust by monitoring the composition of automobile exhaust, characterized in that: include: Based on the engine running state, constructing an exhaust gas component monitoring data set, wherein the exhaust gas component monitoring data set includes exhaust gas components and engine state parameters; Performing a category uniformity analysis on the exhaust gas component monitoring data set, and expanding the data set based on the category uniformity analysis result to obtain a target data set; Annotate samples of the target data set according to preset processing objectives to construct a training data set; Building a model framework, performing model training convergence through the training data set, and obtaining an exhaust analysis model; Based on the preset processing target, an exhaust gas detection device is configured, and target detection of automobile exhaust is performed by the exhaust gas detection device to obtain a target detection result; Inputting the engine monitoring parameters and the target detection results into the exhaust analysis model to analyze the engine exhaust state and determine the exhaust recognition result; Determining an exhaust relationship between engine monitoring parameters and exhaust gas components according to the exhaust gas identification result; The running speed and road condition of the vehicle are collected, and matching relationship analysis is performed according to the engine monitoring parameters, the running speed of the vehicle, and the road condition to determine useless performance loss; Perform engine tracing according to the useless performance loss to determine the power control variable; perform exhaust control variable analysis according to the exhaust relationship between the engine monitoring parameters and exhaust components to determine the exhaust control variable; According to the power control variable and the exhaust gas control variable, the power loss and the exhaust gas composition are optimized, and the engine optimization control strategy is obtained for feedback to guide the driver to perform engine operation and refueling adjustment control; Wherein, before constructing the exhaust gas component monitoring data set based on the engine running state, the following steps are included: According to the instantaneous, intermittent and continuous emission states of the exhaust gas, the first engine operation state is configured; According to the engine speed and torsional strength, the relationship between exhaust and power is analyzed, and the second engine operation state is configured; Extracting operation state features from the first engine operation state and the second engine operation state, and performing random combination of features to construct a reconstructed state, performing engine operation evaluation on the reconstructed state, taking the reconstructed state that satisfies the operation evaluation as a newly added operation state, and taking the reconstructed state that does not satisfy the operation evaluation as a terminated operation state; Searching and deriving based on the newly added operating state according to a set step length, continuing to evaluate the engine operation, and so on, until there is no newly added operating state; All newly added operating states, the first engine operating state, and the second engine operating state are aggregated to obtain the engine operating state.

2. The method according to claim 1, characterized in that Performing a category uniformity analysis on the exhaust gas component monitoring data set, and expanding the data set based on the category uniformity analysis result to obtain a target data set, including: Based on the engine running state, classifying the exhaust gas component monitoring data set to obtain clusters of each state category; Decentralizing the number of each state category cluster to determine the sample deviation; For those that do not meet the preset uniformity requirements, state category cluster supplementation is performed to obtain the target data set, and the state category cluster supplementation includes: performing genetic fission based on the original sample data and obtaining a similar sample set of the same family for correction adjustment.

3. The method according to claim 1, characterized in that The target data set is labeled with samples according to the preset processing objectives to construct a training data set, including: The preset processing target includes exhaust target components and engine target state parameters, and samples are screened in the target data set according to the preset processing target, and the preset processing target is used as an identification target for sample labeling; The positive sample and the negative sample categories are set according to the preset processing target, and the positive sample and the negative sample of the target data set are screened according to the category setting requirements. The positive sample and the preset processing target have a direct identification relationship, and the negative sample and the positive sample have an inverse identification relationship.

4. The method according to claim 3, characterized in that The building of the model framework, performing model training convergence through the training data set, and obtaining the exhaust analysis model includes: Building the model framework includes identifying and filtering layers, input layers, hidden layers, output layers, and fully connected layers; Setting a recognition and screening target based on a target data set, obtaining samples whose annotations match the target data set through the input layer; The hidden layer is trained using the training data entering the input layer, and the analysis results are obtained and output through the output layer. The fully connected layer connects all neurons to obtain exhaust recognition results, and the exhaust recognition results are used to describe the exhaust results caused by the engine operating parameters. A convergence target is set based on the training data set, and a target test is performed according to the output result. When the convergence target is met, the training is completed.

5. The method according to claim 3, characterized in that Based on the preset processing target, the exhaust gas detection equipment is configured, including: Performing target recognition and extraction according to the preset processing target to obtain target features, wherein the target features include exhaust gas composition features and engine state features; When the target feature is the exhaust gas component feature, configuring the detection component parameters of the exhaust gas detection device based on the exhaust gas component; When the target feature is the engine state feature, an activation instruction for the exhaust gas detection device is configured, and the activation instruction is used to start the exhaust gas detection device when the engine operating state parameter meets the activation target feature.

6. The method according to claim 1, characterized in that According to the power control variable and the exhaust gas control variable, the power loss and the exhaust gas composition are optimized, and the engine optimization control strategy is obtained for feedback, including: Based on the sample set collected from the running speed and road condition, a relationship fitting is performed with the power control variable as an independent variable and the power loss as a dependent variable to determine a first fitting relationship; Based on the exhaust gas analysis model, extracting model parameters, and determining a second fitting relationship between the exhaust gas control variable and the exhaust gas component; Setting optimization coefficients for power loss and exhaust gas composition optimization, integrating the first fitting relationship and the second fitting relationship, constructing a fitness function, and establishing an optimization space with the maximum target sum as the optimization target; Setting the exhaust gas composition optimization target as a constraint condition and adding it to the optimization space, randomly selecting a combination of power control variables and exhaust gas control variables, wherein the power control variables and exhaust gas control variables have a unified value of overlapping variables, and evaluating the random combination based on the fitness function as the current optimization strategy; A second combination of power control variables and exhaust gas control variables is randomly selected for fitness evaluation, and compared with the evaluation result of the current optimization strategy. The combination with the larger evaluation result is selected as the current optimization strategy, and it is iterated repeatedly until the optimization goal is met or the number of iterations is reached, and the optimal optimization strategy is obtained for feedback.

Images