Automobile emission monitoring method and system based on multi-source emission data fusion
By constructing a vehicle emission monitoring method that integrates multi-source emission data, using physical information neural networks and recursive Bayesian filtering algorithms to separate the background concentration of the environment, and combining it with the causal graph of the emission system for fault diagnosis, the problem of environmental interference in vehicle exhaust gas monitoring is solved, and the accurate acquisition of real emissions and intelligent fault diagnosis are achieved.
Patent Information
- Application Number
- CN202511413805.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-09-29
- Publication Date
- 2026-02-13
AI Technical Summary
Existing vehicle exhaust emission monitoring technologies cannot accurately and in real time remove background environmental interference in complex environments, leading to false alarms or missed alarms and failing to obtain the true emission levels.
By constructing a vehicle emission monitoring method that integrates multi-source emission data, an emission simulation model is built using a physical information neural network. The background concentration of the environment is separated in real time by combining a recursive Bayesian filtering algorithm, and fault diagnosis is performed by combining the cause-effect graph of the emission system.
It enables precise identification of the true emission concentration of vehicles in complex environments, reduces false alarm rates, and improves the accuracy of emission compliance determination and fault diagnosis.
Smart Images
Figure CN121525433A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of motor vehicle exhaust emission monitoring, in particular to a multi-source emission data fusion automobile emission monitoring method and system. BACKGROUND
[0002] At present, the existing vehicle-mounted exhaust monitoring technology mostly directly uses sensors to measure the concentration value near the exhaust port. However, the measured value is essentially the mixed concentration of vehicle exhaust gas and the background gas in the environment air, and is not the true emission level of the vehicle. The environmental background concentration fluctuates greatly due to factors such as region and weather, and directly using the mixed concentration value for emission compliance judgment or fault diagnosis will introduce significant errors, leading to false positives or false negatives. Therefore, there is currently a lack of an effective technical solution that can strip the environmental background interference in real time and accurately in a complex environment during vehicle driving to obtain the true emission source strength. SUMMARY
[0003] In view of the above prior art, the present application provides a multi-source emission data fusion automobile emission monitoring method and system, which mainly solves the technical problems existing in the background art.
[0004] To achieve the above purpose, the technical scheme of the embodiment of the present application is as follows: In a first aspect, the present application provides a multi-source emission data fusion automobile emission monitoring method, which comprises the following steps: step S1: constructing an emission simulation model according to the multi-source data fusion of historical vehicle road data, weather data, driving state and engine operating state; Step S2: real-time acquisition of current live data and input into the emission simulation model, output of the qualified value range of each emission gas content in the exhaust emission; Step S3: real-time acquisition of the measured value of each gas content in the exhaust emission under the current live data, and comparison and analysis with the qualified value range of each emission gas content in the exhaust emission output according to the emission simulation model, to judge whether each emission gas content is qualified, and if there is an unqualified or abnormal item, it is labeled and converted into a structured text description; Step S4: constructing an emission system causal diagram, and positioning the corresponding abnormal reason in the emission system causal diagram according to the structured text description, and then analyzing the potential faults that may exist in the vehicle.
[0005] As a preferred scheme of the present application, the emission simulation model constructed in step S1 is specifically as follows: preprocessing the historical data, and adding the engine universal characteristic and the emission formation mechanism as constraints to the loss function according to the physical information neural network and training the model; The probability distribution of each exhaust gas emission value in the exhaust emission under the influence of different road surface data, meteorological data, driving state and engine operating state is analyzed and predicted, so as to construct an emission simulation model.
[0006] As a preferred scheme of the present application, the qualified value range in step S2 is specifically: According to the parameters of the probability distribution output by each prediction point, the mean and standard deviation of the distribution are calculated; According to the mean and standard deviation, a qualified range is set: , wherein, is the mean of the distribution, is the confidence level, is the standard deviation of the distribution.
[0007] In the present embodiment, the dynamic qualified range is a probability interval, the upper limit of which is , and the lower limit of which is , wherein is a constant determined according to a preset confidence level, and the range can be dynamically adjusted according to the real-time working condition of the vehicle, and the tolerance is automatically relaxed in high-load working conditions.
[0008] As a preferred scheme of the present application, in step S3, the measured value of the content of each gas in the exhaust emission currently collected is separated from the content of the same gas in the air, specifically as follows: The mixed gas concentration measurement value containing exhaust gas and ambient air is collected, and the pure ambient background gas concentration measurement value is collected; In the embodiment, the pure ambient background gas concentration measurement value is collected through an independent upwind or far-field sensor channel. The sensor is arranged outside the influence range of the vehicle's own exhaust plume to ensure that its measurement value can truly reflect the instantaneous change of the environmental background concentration. The mixed gas concentration measurement value containing exhaust gas and ambient air is collected by a sensor of the same type arranged at the near field (for example, 10-50 cm) downstream of the exhaust pipe. Strictly synchronized sampling of the two channels is the basis for subsequent difference calculation and model accuracy.
[0009] Using the sensor calibration parameters and cross-sensitivity coefficients pre-stored in the system, the mixed gas concentration measurement value and the ambient background gas concentration measurement value are respectively digitally compensated to eliminate the influence of environmental temperature, humidity and pressure changes on the measurement accuracy of the sensor, and calibrated near-field concentration values and far-field concentration values are obtained; A state space model for separation calculation is constructed, and the real exhaust concentration of the vehicle and the ambient background concentration are defined as two state variables to be solved; a measurement equation is established, which expresses the difference between the two calibrated concentration values as a linear combination of the two calibrated concentration values weighted by a dilution factor; A recursive Bayesian filtering algorithm is used for real-time state estimation, and the state estimation value at the previous time is used as the basis to predict the state at the current time, obtain the latest difference observation value, and calculate an optimal gain coefficient; according to the optimal gain coefficient, the difference between the current prediction value and the latest observation value is fused, so as to update and obtain the two state variables at the current time; Then, the optimal estimation values of the two state variables are calculated recursively, including the separated real emission gas concentration values of the vehicle and the current environmental background concentration value.
[0010] In this embodiment, the combination of the state space model and the recursive Bayesian filtering algorithm realizes accurate and real-time separation of the real emission concentration of the vehicle in a complex driving environment, effectively overcomes the interference of the environmental background, and provides a highly reliable data basis for emission compliance determination and fault diagnosis.
[0011] As a preferred scheme of the present application, the step S3 further specifically comprises: comparing and analyzing the separated real emission gas concentration values of the vehicle according to the qualified value range of the content of each emission gas in the tail gas emission amount output by the emission simulation model, including: Each separated real emission gas concentration value is compared and analyzed one by one with the qualified value range of the content of each emission gas in the tail gas emission amount output by the emission simulation model, to determine whether the content of each emission gas in all tail gas emission amounts is in the qualified value range; If the content of a certain emission gas is not in the corresponding qualified value range, the emission gas is identified as an unqualified or abnormal item, and it is identified whether the emission gas concentration value is higher than the qualified value range or lower than the qualified value range.
[0012] As a preferred scheme of the present application, the step S3 further specifically comprises: The emission gas identified as an unqualified or abnormal item is converted into a structured text description, specifically including: an abnormal gas type field including a specific abnormal gas name, a deviation direction field including "high" or "low" state, an over-standard amplitude field including a specific deviation percentage value, and a live context field including key working condition parameters when the abnormality occurs; and the fields are spliced in order to form a search string.
[0013] As a preferred scheme of the present application, the step S4 of constructing the emission system causal diagram specifically comprises: The target vehicle emission system is analyzed as a whole and divided into multiple functional subsystems, including an engine combustion system, an exhaust aftertreatment system, a sensor monitoring system and an electronic control system; For each subsystem, a causal diagram node representing its specific working state is defined; The nodes are divided into three types: root cause nodes for indicating specific component failure or performance degradation states, intermediate nodes for indicating system performance parameter abnormalities, and observation nodes for indicating directly measurable tail gas emission abnormalities; Based on engine combustion principles, catalytic chemical reaction mechanisms and fluid dynamics theories, the causal relationship connections between the nodes are established, and the nodes with causal correlations are connected to each other by directed connection lines, with the connection direction representing the action direction of the causal relationship; For each node, a corresponding conditional probability parameter is defined to quantitatively represent the influence of parent node state changes on the node state, and the nodes, connection relationships and conditional probability parameters are integrated to form a complete emission system causal diagram.
[0014] As a preferred scheme of the present application, the step S4 specifically includes: The string is parsed grammatically and semantically, and four key field contents are identified and extracted through predefined delimiters; The extracted key field contents are matched with observation nodes in the emission system causal diagram in terms of similarity; starting from the matched observation nodes, path traversal is performed in the reverse direction of the causal relationship; The causal paths corresponding to different abnormal gases are cross-verified to find common parent nodes or associated nodes; The posterior probability of each potential fault hypothesis is calculated through the Bayes update rule, and the fault hypothesis list is sorted in descending order of probability value to generate a fault hypothesis list arranged in confidence; The most likely root cause node is selected from the fault hypothesis list, and is mapped to a specific component failure or system failure mode, and finally a diagnostic report containing the following contents is generated: the most likely root cause, the evidence supporting the conclusion, the complete fault propagation path analysis and the recommended repair measures.
[0015] In a second aspect, the present application also provides an automobile emission monitoring system, which comprises: An emission model construction module: an emission simulation model is constructed after multi-source data fusion according to historical vehicle road data, meteorological data, driving state and engine operating state; A model analysis module: real-time current live data is collected and input into the emission simulation model, and the qualified value range of each emission gas content in the tail gas emission is output; The comparison module: real-time acquisition of the measured values of the content of each gas in the exhaust emission under the current live, and comparison and analysis with the qualified value range of the content of each emission gas in the exhaust emission output according to the emission simulation model, to judge whether the content of each emission gas is qualified, if there is an unqualified or abnormal item, it will be marked and converted into a structured text description; The cause and effect analysis module: build an emission system cause and effect diagram, and then locate the corresponding abnormal reason in the emission system cause and effect diagram according to the structured text description, and further analyze the potential faults that the vehicle may have.
[0016] The beneficial effects of the present application are: the present application builds an emission simulation model and compares and analyzes the qualified value range of the content of each emission gas in the exhaust emission output by the model, marks the unqualified or abnormal items, and then locates the corresponding abnormal reason through the emission system cause and effect diagram, and further analyzes the potential faults that the vehicle may have, effectively solving the problems of false alarm and missed alarm caused by the use of fixed threshold in traditional automobile emission monitoring, and unable to distinguish environmental background interference; the present application also separates the environmental background concentration interference and obtains the true emission value; combined with the cause and effect diagram reasoning, the intelligent diagnosis from abnormal phenomenon to fault root cause is realized. The monitoring accuracy is improved and the system false alarm rate is reduced. BRIEF DESCRIPTION OF DRAWINGS
[0017] Figure 1 The step schematic diagram of the multi-source emission data fusion automobile emission monitoring method provided by the present application; Figure 2 The structure schematic diagram of the automobile emission monitoring system provided by the present application. DETAILED DESCRIPTION
[0018] The technical solutions of the present application are further described in detail below in combination with the drawings and specific embodiments. Unless otherwise defined, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present application belongs. The terms used in the specification of the present application are only for the purpose of describing the specific embodiments and are not intended to limit the present application. In the following description, the expression "some embodiments" describes a subset of all possible embodiments, but it should be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict.
[0019] In the following description, a large number of specific details are given in order to provide a more thorough understanding of the present application. However, it is obvious to those skilled in the art that the present application can be implemented without one or more of these details. In other examples, some technical features known in the art are not described in order to avoid obscuring the present application.
[0020] It should be understood that the present invention can be embodied in various forms and should not be construed as being limited to the embodiments set forth herein. Rather, providing these embodiments will make the disclosure thorough and complete, and will fully convey the scope of the invention to those skilled in the art. Furthermore, the terminology used herein is intended only to describe particular embodiments and is not intended to limit the invention. When used herein, the singular forms “a,” “an,” and “the” are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the terms “compose” and / or “comprising,” when used in this specification, identify the presence of the stated features, integers, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups. When used herein, the term “and / or” includes any and all combinations of the associated listed items.
[0021] It should also be noted that when an element is referred to as being "fixed to" another element, it can be directly attached to the other element or there may be an intervening element. When an element is referred to as being "connected to" another element, it can be directly connected to the other element or there may be an intervening element. The terms "vertical," "horizontal," "inner," "outer," "left," "right," and similar expressions used herein are for illustrative purposes only and do not represent the only possible implementation.
[0022] To fully understand this invention, a detailed structure will be presented in the following description to illustrate the technical solution proposed by this invention. Optional embodiments of the invention are described in detail below; however, in addition to these detailed descriptions, the invention may have other embodiments.
[0023] Firstly, this invention provides a method for monitoring vehicle emissions by fusing multi-source emission data, please refer to the appendix. Figure 1 The method includes the following steps: Step S1: Construct an emission simulation model by fusing multi-source data based on historical vehicle road surface data, meteorological data, driving status, and engine operating status; Step S2: Collect real-time data and input it into the emission simulation model, outputting the acceptable range of each emission gas content in the exhaust gas emission. Step S3: Real-time acquisition of the measured values of each gas content in the exhaust gas emission under the current actual conditions, and comparison and analysis with the qualified value range of each emission gas content in the exhaust gas emission output according to the emission simulation model, to determine whether each emission gas content is qualified. If there are unqualified or abnormal items, they are marked and converted into structured text descriptions. Step S4: Construct a cause-effect graph of the emission system, then locate the corresponding abnormal causes in the cause-effect graph of the emission system based on the structured text description, and then analyze the potential faults that may exist in the vehicle.
[0024] As a preferred embodiment of the present invention, the construction of the emission simulation model in step S1 is specifically as follows: historical data is preprocessed, and the engine universal characteristics and emission formation mechanism are added as constraints to the loss function according to the physical information neural network and the model is trained. The probability distribution of each emission gas in exhaust emissions is analyzed and predicted under the influence of different road surface data, meteorological data, driving conditions, and engine operating conditions, thereby constructing an emission simulation model.
[0025] In this embodiment, the input layer of the emission simulation model receives a preprocessed multi-source feature vector, which includes road surface data such as slope and road type; meteorological data such as ambient temperature, pressure, and humidity; driving status such as vehicle speed and acceleration; and engine operating status such as engine speed, load, air-fuel ratio, and coolant temperature.
[0026] In this embodiment, the loss function consists of two parts:
[0027] in, This represents the data fit term, used to measure the difference between model predictions and actual emission measurements. It is typically expressed as mean squared error (MSE) or negative log-likelihood (NLL) loss. This term ensures that the model is consistent with historical observation data.
[0028] This represents a physical constraint term, used to embed the engine's universal characteristics and emission formation mechanism as mandatory soft constraints into the model training process, forcing the network's learning results to conform to basic physical laws. The parameters... It is used to adjust the strength of physical constraints.
[0029] The physical constraint term can be further decomposed into multiple sub-terms, for example: Constraints based on the universal characteristics of the engine: The universal characteristics of the engine describe the inherent relationship between its emissions and its speed and torque. This constraint can be expressed as:
[0030] in, For the sample size, Indicates the neural network for the first... The predicted emissions values for each sample This indicates that after consulting the engine's universal characteristic MAP diagram and performing interpolation calculations, the result was obtained at a speed of... Torque is The typical concentration value of this emission under operating conditions.
[0031] A constraint based on the chemical relationships between emissions: There are stoichiometric relationships between the components in the exhaust gas. For example, when the air-fuel ratio (AFR) becomes leaner, NOx emissions typically increase, while CO and HC emissions decrease. This constraint can be expressed as:
[0032] in, This represents the partial derivative of the NOx emission value predicted by the neural network model with respect to the air-fuel ratio. This represents the partial derivative of the sum of CO and HC emissions predicted by the neural network model with respect to the air-fuel ratio. ( ) denotes the rectified linear function. This constraint ensures that the emission trends predicted by the model are consistent with basic chemical principles.
[0033] Constraints based on mass or energy conservation: For example, the carbon balance principle requires that the carbon produced by fuel consumption should be equal to the total carbon contained in CO, CO2, and HC in the exhaust gas. This constraint can be expressed as:
[0034] in, Indicates the first Fuel mass flow rate for each sample This indicates the mass fraction of carbon in the fuel. , , These represent the mass emission rates of CO, CO2, and HC pollutants predicted by the neural network model, respectively. , , These represent the molar masses of a carbon atom, a carbon monoxide molecule, and a carbon dioxide molecule, respectively. This represents the mass fraction of carbon in the HC molecule, and this constraint ensures the physical plausibility of the model's predictions.
[0035] As a preferred embodiment of the present invention, the qualified numerical range in step S2 is specifically as follows: Calculate the mean and standard deviation of the probability distribution based on the parameters of the probability distribution output for each prediction point. The acceptable range is set based on its mean and standard deviation: ,in, Let be the mean of the distribution. For confidence level, denoted as the standard deviation of the distribution.
[0036] As a preferred embodiment of the present invention, in step S3, the measured values of the content of each gas in the exhaust gas emission under the current conditions are separated from the content of the same gas in the air, as follows: Collect concentration measurements of mixed gases including exhaust gas and ambient air, as well as pure background gas concentration measurements. Digital compensation processing is performed on the mixed gas concentration measurement value and the ambient background gas concentration measurement value to eliminate the influence of changes in ambient temperature, humidity and pressure on the sensor measurement accuracy, and to obtain the calibrated near-field concentration value and far-field concentration value. A state-space model for separate computation is constructed, defining the actual vehicle emission concentration and the environmental background concentration as two state variables to be solved. Then establish an observation equation, which expresses the difference between the two calibrated concentration values as a combination of the two calibrated concentration values linearly weighted by a dilution factor; A recursive Bayesian filtering algorithm is used for real-time state estimation. Based on the state estimate of the previous time step, the state at the current time step is predicted, the latest difference observation is obtained, and an optimal gain coefficient is calculated. According to the optimal gain coefficient, the difference between the current prediction value and the latest observation value is fused to update and adjust the two state variables at the current time step. Then, the optimal estimates of two state variables are output through recursive calculation. The optimal estimates include the actual emission gas concentration values of each vehicle separated from the data, and the current environmental background concentration value.
[0037] As a preferred embodiment of the present invention, step S3 further includes: comparing and analyzing the separated actual emission gas concentration values of the vehicle with the acceptable numerical ranges of each emission gas content in the exhaust gas emission output by the emission simulation model, including: The concentration values of each isolated real emission gas are compared and analyzed one by one with the acceptable range of each emission gas content in the exhaust gas output by the emission simulation model to determine whether the content of each emission gas in the total exhaust gas output is within the acceptable range. If the content of a certain emitted gas is outside the corresponding acceptable value range, the emitted gas will be marked as unacceptable or abnormal, and it will be indicated whether the concentration of the emitted gas is higher or lower than the acceptable value range.
[0038] For example, when the measured value of a certain gas content continuously exceeds the dynamic acceptable range output by the emission simulation model, the anomaly identification procedure is immediately initiated. First, the system accurately records the specific type of the abnormal gas, including but not limited to nitrogen oxides, carbon monoxide, or hydrocarbons. Second, the system determines the direction of deviation and records the deviation attribute by comparing the measured value with the upper and lower limits of the acceptable range: when the measured value is greater than the upper limit of the acceptable range, it is identified as "concentration too high"; when the measured value is less than the lower limit of the acceptable range, it is identified as "concentration too low".
[0039] Next, the system quantifies the degree of anomaly, generating data on the extent of deviation. This data is expressed as a percentage, indicating the degree to which the measured value deviates from the acceptable range. Simultaneously, the system captures and records the operating context information at the time of the anomaly, including but not limited to engine speed range, load status, exhaust temperature characteristics, and aftertreatment system operating parameters. These contextual parameters provide crucial environmental clues for subsequent fault diagnosis.
[0040] In this embodiment, by comparing the filtered actual emission values with the dynamically changing acceptable range, the emission monitoring is contextualized. Through the refined identification of anomalies, causal clues are provided for subsequent root cause analysis of faults, thereby improving the intelligence level, diagnostic efficiency, and accuracy of the entire emission monitoring system.
[0041] As a preferred embodiment of the present invention, step S3 further includes: Emissions identified as non-compliant or abnormal are converted into structured text descriptions, including: an abnormal gas type field (including the specific abnormal gas name), a deviation direction field (including "high" or "low" status), an exceedance range field (including the specific deviation percentage value), and a real-time context field (including key operating parameters at the time of the abnormality). These fields are then concatenated in sequence to form a search string.
[0042] As a preferred embodiment of the present invention, the step S4 of constructing the emission system cause-effect graph specifically includes: The target vehicle's emission system is analyzed as a whole and divided into multiple functional subsystems, including the engine combustion system, exhaust aftertreatment system, sensor monitoring system, and electronic control system. For each subsystem, define a cause-effect graph node representing its specific operating state; The nodes are divided into three types: root cause nodes, which are used to indicate the failure or performance degradation of specific components; intermediate nodes, which are used to indicate abnormal system performance parameters; and observation nodes, which are used to indicate abnormal exhaust emissions that can be directly measured. Based on engine combustion principles, catalytic chemical reaction mechanisms, and fluid dynamics theory, causal relationships between various nodes are established. Directed connecting lines are used to link nodes with causal connections, with the direction of connection indicating the direction of the causal relationship. For example, the "injector carbon buildup" node points to the "air-fuel ratio imbalance" node, and the "air-fuel ratio imbalance" node points to the "high nitrogen oxide concentration" node. Each causal relationship is established based on a clear physical mechanism or chemical principle.
[0043] In this embodiment, by integrating engine control unit calibration data, fault mode analysis data, domain expert experience and knowledge, and historical maintenance case data, a causal discovery algorithm based on statistical analysis is used to further verify and supplement causal relationships that are uncertain or controversial. Next, define corresponding conditional probability parameters for each node to quantify the impact of parent node state changes on the node's state. Integrate the nodes, connections, and conditional probability parameters to form a complete causal graph of the emission system. Finally, use historical case data to verify the constructed causal graph, calculate its diagnostic accuracy and recall, and iteratively optimize the network structure based on the verification results.
[0044] In this embodiment, the system initially filters a set of potentially related fault nodes in the emission system causal graph based on the combination of abnormal gas type and deviation direction. The causal graph nodes contain various possible fault modes, and each node has a predefined associated gas abnormality characteristic mode.
[0045] Further utilizing the operating condition context information, a credibility weighting calculation is performed on the initially screened fault nodes. Fault nodes with a high degree of matching with the current operating condition characteristics receive higher weights, while nodes that do not match are downweighted or excluded.
[0046] As a preferred embodiment of the present invention, step S4 specifically includes: Perform syntax parsing and semantic analysis on the string, and identify and extract the content of four key fields using predefined delimiters; The extracted key field content is matched with the observation nodes in the causal graph of the emission system; starting from the matched observation node, the path is traversed in the reverse direction of the causal relationship; Cross-validate the causal paths corresponding to different anomalous gases to find common parent nodes or related nodes; The posterior probability of each potential failure hypothesis is calculated using the Bayesian update rule, and the hypotheses are sorted from high to low according to their probability values to generate a list of failure hypotheses sorted by confidence level. The most likely root cause node is selected from the list of fault hypotheses and mapped to a specific component failure or system failure mode. Finally, a diagnostic report is generated that includes the following: the most likely root cause, details of the evidence supporting the conclusion, a complete fault propagation path analysis, and recommended maintenance measures.
[0047] In this embodiment, the "observation nodes" in the emission system causal graph represent directly measurable end phenomena, such as "high NOx reading". The system matches the gas_type and deviation_direction in the query with the attributes of all observation nodes. For example, the query {gas_type: NOx, deviation: high} will match an observation node named "high NOx concentration". The higher the matching degree, the greater the weight of that node as the starting point for inference.
[0048] Starting from the matched observation node, the process traverses its parent nodes in reverse causal order. For example, starting from the "high NOx concentration" node, it might traverse backwards to its parent node "decreased SCR catalytic efficiency," and then continue to the "urea injection failure" node. This process systematically explores all potential fault paths that could lead to this abnormal symptom.
[0049] When multiple anomalies exist, such as reporting "NOx too high" and "NH3 leak" at the same time, the system will perform reverse traversal and look for the common parent node that intersects in the path. This common node is a strong indication of the common cause of multiple anomalies. For example, "excessive urea injection" may simultaneously lead to incomplete NOx conversion and NH3 leakage.
[0050] Each node in the causal graph is associated with a prior probability, representing the prevalence of the fault, and a conditional probability table, representing the strength of the influence of the parent node's state on the current node's state. At the start of inference, the probabilities of all fault hypotheses are their prior probabilities. When new evidence, such as "high NOx concentration," is introduced, the system recursively updates the posterior probabilities of all relevant nodes in the graph using Bayesian rules, gradually narrowing down the range of fault hypotheses. For example, let's assume the most likely root cause is a failure of the urea injection pump in the selective catalytic reduction (SCR) system.
[0051] Supporting evidence: Nitrogen oxide (NOx) emission concentrations consistently exceeded the dynamic compliance range upper limit by 65%.
[0052] This anomaly is particularly pronounced under medium and high engine load conditions. Therefore, the diagnostic system calculates the posterior probability of this fault hypothesis to be 82.5%.
[0053] The fault propagation path analysis is as follows: urea injection pump failure → insufficient injection volume → decreased catalytic reduction efficiency of SCR system → ineffective purification of original engine NOx emissions → high NOx concentration reading downstream of exhaust aftertreatment.
[0054] Finally, the recommended maintenance measures are as follows: Check the urea level and quality.
[0055] Use a diagnostic tool to read the relevant fault codes of the SCR system.
[0056] Check if the urea pump inlet filter is clogged.
[0057] Secondly, the present invention also provides an automotive emissions monitoring system, please refer to the appendix. Figure 2 The system includes: Emissions model building module: An emissions simulation model is built by fusing multi-source data, including historical vehicle road data, meteorological data, driving status, and engine operating status. Model Analysis Module: Collects real-time data and inputs it into the emission simulation model, outputting the acceptable range of values for each emission gas content in the exhaust gas emission. Comparison module: Real-time acquisition of measured values of each gas content in exhaust emissions under current conditions, and comparison analysis with the qualified value range of each gas content in exhaust emissions output by the emission simulation model to determine whether each gas content is qualified. If there are unqualified or abnormal items, they are marked and converted into structured text descriptions. Causal analysis module: Constructs a causal graph of the emission system, then locates the corresponding abnormal causes in the causal graph based on the structured text description, and then analyzes the potential faults that may exist in the vehicle.
[0058] Thirdly, embodiments of this application also provide an electronic device, including a processor and a memory, wherein the memory stores a plurality of instructions; the processor loads instructions from the memory to execute steps in any of the multi-source emission data fusion vehicle emission monitoring methods provided in embodiments of this application.
[0059] Fourthly, embodiments of this application also provide a computer-readable storage medium storing a plurality of instructions adapted for loading by a processor to execute steps in any of the multi-source emission data fusion vehicle emission monitoring methods provided in embodiments of this application.
[0060] Fifthly, embodiments of this application also provide a computer program product, including a computer program / instructions, which, when executed by a processor, implement the steps in any of the multi-source emission data fusion vehicle emission monitoring methods provided in embodiments of this application.
[0061] In this embodiment, those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be accomplished by instructions, or by instructions controlling related hardware. These instructions can be stored in a computer-readable storage medium and loaded and executed by a processor.
[0062] Therefore, embodiments of this application provide a computer-readable storage medium storing a plurality of instructions that can be loaded by a processor to execute steps in any of the multi-source emission data fusion vehicle emission monitoring methods provided in embodiments of this application.
[0063] The storage medium may include: read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.
[0064] According to one aspect of this application, a computer program product or computer program is provided, comprising a computer program / instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer program / instructions from the computer-readable storage medium and executes the computer program / instructions, causing the electronic device to perform the method provided in various optional implementations of the multi-source emission data fusion vehicle emission monitoring method steps provided in the above embodiments.
[0065] Since the instructions stored in the storage medium can execute the steps in any of the multi-source emission data fusion vehicle emission monitoring methods provided in the embodiments of this application, the beneficial effects that any of the multi-source emission data fusion vehicle emission monitoring methods provided in the embodiments of this application can achieve can be realized. For details, please refer to the previous embodiments, which will not be repeated here.
[0066] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. The scope of protection of the present invention should be determined by the scope of the claims.
Claims
1. A method for monitoring vehicle emissions by fusing multi-source emission data, characterized in that, The method includes the following steps: Step S1: Construct an emission simulation model by fusing multi-source data based on historical vehicle road surface data, meteorological data, driving status, and engine operating status; Step S2: Collect real-time data and input it into the emission simulation model, outputting the acceptable range of each emission gas content in the exhaust gas emission. Step S3: Real-time acquisition of the measured values of each gas content in the exhaust gas emission under the current actual conditions, and comparison and analysis with the qualified value range of each emission gas content in the exhaust gas emission output according to the emission simulation model, to determine whether each emission gas content is qualified. If there are unqualified or abnormal items, they are marked and converted into structured text descriptions. Step S4: Construct a cause-effect graph of the emission system, then locate the corresponding abnormal causes in the cause-effect graph of the emission system based on the structured text description, and then analyze the potential faults that may exist in the vehicle.
2. The vehicle emission monitoring method based on multi-source emission data fusion according to claim 1, characterized in that, The specific steps for constructing the emission simulation model in step S1 are as follows: historical data are preprocessed, and the engine's universal characteristics and emission formation mechanism are added as constraints to the loss function based on the physical information neural network, and the model is trained. The probability distribution of each emission gas in exhaust emissions is analyzed and predicted under the influence of different road surface data, meteorological data, driving conditions, and engine operating conditions, thereby constructing an emission simulation model.
3. The vehicle emission monitoring method based on multi-source emission data fusion according to claim 2, characterized in that, The specific range of acceptable values in step S2 is as follows: Calculate the mean and standard deviation of the probability distribution based on the parameters of the probability distribution output for each prediction point. The acceptable range is set based on its mean and standard deviation: ,in, Let be the mean of the distribution. For confidence level, denoted as the standard deviation of the distribution.
4. The vehicle emission monitoring method based on multi-source emission data fusion according to claim 3, characterized in that, In step S3, the measured values of each gas content in the currently collected exhaust gas emission are separated from the content of the same gas in the air, as follows: Collect concentration measurements of mixed gases including exhaust gas and ambient air, as well as pure background gas concentration measurements. Digital compensation processing is performed on the mixed gas concentration measurement value and the ambient background gas concentration measurement value to eliminate the influence of changes in ambient temperature, humidity and pressure on the sensor measurement accuracy, and to obtain the calibrated near-field concentration value and far-field concentration value. A state-space model for separate computation is constructed, defining the actual vehicle emission concentration and the environmental background concentration as two state variables to be solved. Then establish an observation equation, which expresses the difference between the two calibrated concentration values as a combination of the two calibrated concentration values linearly weighted by a dilution factor; A recursive Bayesian filtering algorithm is used for real-time state estimation. Based on the state estimate of the previous time step, the state at the current time step is predicted, the latest difference observation is obtained, and an optimal gain coefficient is calculated. According to the optimal gain coefficient, the difference between the current prediction value and the latest observation value is fused to update and adjust the two state variables at the current time step. Then, the optimal estimates of two state variables are output through recursive calculation. The optimal estimates include the actual emission gas concentration values of each vehicle separated from the data, and the current environmental background concentration value.
5. The vehicle emission monitoring method based on multi-source emission data fusion according to claim 4, characterized in that, Step S3 further includes: comparing and analyzing the separated actual emission gas concentration values of the vehicle against the acceptable range of each emission gas content in the exhaust gas emission output by the emission simulation model, including: The concentration values of each isolated real emission gas are compared and analyzed one by one with the acceptable range of each emission gas content in the exhaust gas output by the emission simulation model to determine whether the content of each emission gas in the total exhaust gas output is within the acceptable range. If the content of a certain emitted gas is outside the corresponding acceptable value range, the emitted gas will be marked as unacceptable or abnormal, and it will be indicated whether the concentration of the emitted gas is higher or lower than the acceptable value range.
6. The vehicle emission monitoring method based on multi-source emission data fusion according to claim 5, characterized in that, Step S3 further includes: Emissions identified as non-compliant or abnormal are converted into structured text descriptions, including: an abnormal gas type field (including the specific abnormal gas name), a deviation direction field (including "high" or "low" status), an exceedance range field (including the specific deviation percentage value), and a real-time context field (including key operating parameters at the time of the abnormality). These fields are then concatenated in sequence to form a search string.
7. The vehicle emission monitoring method based on multi-source emission data fusion according to claim 6, characterized in that, The specific steps in step S4 of constructing the cause-effect graph of the emission system include: The target vehicle's emission system is analyzed as a whole and divided into multiple functional subsystems, including the engine combustion system, exhaust aftertreatment system, sensor monitoring system, and electronic control system. For each subsystem, define a cause-effect graph node representing its specific operating state; The nodes are divided into three types: root cause nodes, which are used to indicate the failure or performance degradation of specific components; intermediate nodes, which are used to indicate abnormal system performance parameters; and observation nodes, which are used to indicate abnormal exhaust emissions that can be directly measured. Based on the engine combustion principle, catalytic chemical reaction mechanism and fluid dynamics theory, the causal relationship between each node is established. Directed connecting lines are used to connect nodes with causal relationships, and the connection direction indicates the direction of the causal relationship. Then, define corresponding conditional probability parameters for each node to quantify the degree of influence of the parent node's state change on the node's state. Integrate the nodes, connections, and conditional probability parameters to form a complete causal graph of the emission system.
8. The vehicle emission monitoring method based on multi-source emission data fusion according to claim 7, characterized in that, Step S4 specifically includes: Perform syntax parsing and semantic analysis on the string, and identify and extract the content of four key fields using predefined delimiters; The extracted key field content is matched with the observation nodes in the causal graph of the emission system; starting from the matched observation node, the path is traversed in the reverse direction of the causal relationship; Cross-validate the causal paths corresponding to different anomalous gases to find common parent nodes or related nodes; The posterior probability of each potential failure hypothesis is calculated using the Bayesian update rule, and the hypotheses are sorted from high to low according to their probability values to generate a list of failure hypotheses sorted by confidence level. The most likely root cause node is selected from the list of fault hypotheses and mapped to a specific component failure or system failure mode. Finally, a diagnostic report is generated that includes the following: the most likely root cause, details of the evidence supporting the conclusion, a complete fault propagation path analysis, and recommended maintenance measures.
9. A vehicle emission monitoring system, characterized in that, The system includes: Emissions model building module: An emissions simulation model is built by fusing multi-source data, including historical vehicle road data, meteorological data, driving status, and engine operating status. Model Analysis Module: Collects real-time data and inputs it into the emission simulation model, outputting the acceptable range of values for each emission gas content in the exhaust gas emission. Comparison module: Real-time acquisition of measured values of each gas content in exhaust emissions under current conditions, and comparison analysis with the qualified value range of each gas content in exhaust emissions output by the emission simulation model to determine whether each gas content is qualified. If there are unqualified or abnormal items, they are marked and converted into structured text descriptions. Causal analysis module: Constructs a causal graph of the emission system, then locates the corresponding abnormal causes in the causal graph based on the structured text description, and then analyzes the potential faults that may exist in the vehicle.