Method for compensating detection data of standard component of nitrogen-oxygen sensor

Through nonlinear relationship model training and data compensation technology, the detection differences and performance attenuation between nitrogen oxygen sensor standard parts are solved, which improves detection accuracy and consistency, and reduces the cost of replacing standard parts.

CN118914462APending Publication Date: 2024-11-08ZHENJIANG XINYUAN CHENGDA SENSOR TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410974478.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-19
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

There are detection differences and performance attenuation between the nitrogen oxygen sensor standard parts, resulting in the inability to achieve a rational state of calibration accuracy. The existing methods cannot effectively reduce the impact of differences and increase production costs.

Method used

By designing different sampling atmosphere points, recording the detection values ​​of the standard parts, using nonlinear relationship models (such as BP neural networks) to train the detection value model between the child parts and the master parts, and loading the model regularly for data compensation, realizing the standard parts, The accuracy and consistency of the detection value.

Benefits of technology

Improves detection consistency and accuracy between standard parts, reduces the cost of replacement of standard parts, and achieves improvements in detection values ​​without changing calibration equipment and processes.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118914462A_ABST
    Figure CN118914462A_ABST
Patent Text Reader

Abstract

The invention discloses a method for compensating detection data of a standard component of a nitrogen-oxygen sensor, which comprises the following steps of: 1) selecting different oxygen and nitrogen-oxygen concentrations, and designing different sampling atmosphere points; (2) the mother part and the standard parts are placed under the same gas distribution table, and the designed different atmosphere points are maintained for a certain time in sequence; 3) carrying out data processing on the collected detection values of the standard parts and the mother parts; 4) taking the NOx detection value of the parent part as output, taking the NOx and O2 values of the child part as input, and training a model between the NOx detection values of the child part and the parent part by using a nonlinear relation model; (5) regularly training a model between each standard part and the parent part, and enabling a standard part detection numerical value compensation upper computer to have a model loading function according to the format requirement of an upper computer loading model; and 6) calibrating a dew point instruction sent by the upper computer. Aiming at the problems of deviation and performance degradation between the standard components, the consistency between the standard components and the detection accuracy can be improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of nitrogen oxygen sensor standard component detection, and in particular to a nitrogen oxygen sensor standard component detection data compensation method. Background Art

[0002] Nitrogen oxide sensors are mainly used in diesel engine exhaust treatment systems. Nitrogen oxide sensors transmit the detected NOx concentration signal to the ECU, spray urea through the SCR system, neutralize the NOx in the exhaust gas, and make the NOx concentration in the exhaust gas meet the national standards. On May 26, 2021, the Ministry of Ecology and Environment held a regular press conference to inform that from July, my country will fully implement the National VI emission standard for heavy-duty diesel vehicles. Compared with the National V standard, the National VI nitrogen oxides of heavy-duty vehicles are reduced by 77%, and the nitrogen oxides shall not exceed 35mg / km, which has higher requirements for the detection accuracy of nitrogen oxide sensors. After entering the National VI stage, the number of nitrogen oxide sensors has increased to 2. The additional nitrogen oxide sensors are installed upstream of the oxidation catalytic converter (DOC) to prevent the redox reaction inside the DOC from affecting the concentration measurement of nitrogen oxides. The control unit comprehensively judges the exhaust purification efficiency based on the signals of the two sensors. When the nitrogen oxide sensor detects that the nitrogen oxide content in the exhaust gas exceeds the standard, the engine will report a fault, light up the engine fault light, and even limit the engine output torque. Therefore, the measurement accuracy of the nitrogen oxide sensor is crucial to the vehicle after-treatment system.

[0003] The nitrogen oxide sensor needs to be calibrated before it leaves the factory. By using a nitrogen oxide sensor with high accuracy as a standard part, the NOx and O2 concentration detection values ​​of the standard part and each product at different atmosphere points are obtained, the calibration data is fitted, and the fitted data is written into the control chip through the production equipment to complete the calibration. Therefore, the detection accuracy of the nitrogen oxide sensor depends largely on the accuracy of the standard part, but there are still certain detection differences between different standard parts. At the same time, the detection performance of each nitrogen oxide sensor has a certain attenuation during use, so there are certain differences between the standard parts themselves. The difference and attenuation characteristics between each standard part are strongly nonlinear and cannot be accurately calculated, resulting in the final calibration accuracy cannot reach a rational state. In order to reduce the adverse effects caused by the differences between different standard parts and the attenuation of the standard parts themselves, the method currently commonly used is: before each production, the standard parts used in the production line are compared with the detection values ​​of the master parts at regular intervals. When the error of the standard parts exceeds a certain range, the standard parts will be replaced. This method not only cannot avoid the influence of standard part deviation, but also causes an increase in production costs.

[0004] Therefore, a method for compensating the detection data of nitrogen oxide sensor standard parts is proposed. Summary of the invention

[0005] The purpose of the present invention is to provide a method for compensating the detection data of nitrogen oxide sensor standard parts. In order to improve the consistency and accuracy between each standard part and reduce the cost increase caused by multiple purchases of standard parts due to the performance degradation of standard parts, this patent proposes a method for compensating the detection data of nitrogen oxide sensor standard parts, so that it can improve the consistency and accuracy of standard parts through data processing and reduce the cost of replacing standard parts.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A method for compensating detection data of a nitrogen oxygen sensor standard part comprises the following steps:

[0008] 1) Select different oxygen and nitrogen concentrations and design different sampling atmosphere points;

[0009] 2) Place the mother part and multiple standard parts under the same gas distribution table, and maintain the designed different atmosphere points for a certain period of time in turn, so that each atmosphere point reaches stability, and the sensor detection value reaches a stable state, and record its detection values ​​at different atmosphere points;

[0010] 3) Processing the collected test values ​​of each standard part and the master part;

[0011] 4) Taking the NOx detection value of the mother component as output and the NOx and O2 values ​​of the daughter component as input, a nonlinear relationship model is used to train a model between the NOx detection values ​​of the daughter component and the mother component. Then taking the O2 detection value of the mother component as output and the NOx and O2 values ​​of the daughter component as input, a nonlinear relationship model is used to train a model between the O2 detection values ​​of the daughter component and the mother component.

[0012] 5) Regularly train the model between each standard part and the master part. The standard part detection numerical compensation host has a model loading function. According to the format requirements of the host computer loading model, the trained model is converted into a format so that the loading calculation can be performed through the compensation host computer;

[0013] 6) The calibration host computer sends a dew point command, which is transmitted to the nitrogen oxide sensor standard part through the 1#CAN analyzer and the gateway. After receiving the dew point command, the standard part transmits the message data of the current atmosphere point in real time to the compensation host computer through the gateway and the 2#CAN analyzer. The compensation host computer compensates the standard part detection value through the loaded training model, and sends the compensated message data to the 2#CAN analyzer. When passing through the gateway, the ID of this message is changed back to the original ID, and transmitted to the calibration host computer through the 1#CAN analyzer. The calibration host computer calibrates each nitrogen oxide sensor through the compensated standard part detection value.

[0014] Furthermore, the sampling period of data recording in step 2) should be much shorter than the time that each atmosphere point maintains stability.

[0015] Furthermore, in the process of switching between different atmosphere points in step 3), the atmosphere in the gas distribution station is not stable, and the detection data in which the atmosphere has not yet reached a stable state is deleted.

[0016] Furthermore, in step 4), a model of nonlinear relationship is used to train a model between the O2 detection values ​​of the sub-component and the parent component, and its performance function uses MAE mean absolute error.

[0017] Furthermore, in the step 4), the NOx and O2 models between the remaining sub-components and the parent component are trained according to the model training method of the nonlinear relationship.

[0018] Furthermore, the model of the nonlinear relationship in step 4) adopts BP neural network.

[0019] Furthermore, during the operation of each standard component in step 5), the detection performance will slowly and continuously decay.

[0020] Furthermore, in order to distinguish the compensated message data from the original message data in step 6), the ID of the message data needs to be changed.

[0021] Compared with the prior art, the present invention has the following beneficial effects:

[0022] 1. The present invention aims at the problem of deviation and performance attenuation between standard parts, and can improve the consistency between standard parts and the accuracy of detection;

[0023] 2. The present invention achieves improvement in the test value without changing the internal program parameters and appearance of the existing standard parts;

[0024] 3. After long-term use, the stability and accuracy of the detection values ​​of standard parts will continue to decay over time. The present invention can compensate for the decay on the basis of the original standard parts and reduce the cost of purchasing standard parts.

[0025] 4. There is no need to change the original calibration equipment and calibration process, which is easy to implement. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is a schematic diagram of the principle of the scheme of the present invention;

[0027] Figure 2 It is the original system schematic diagram of the present invention;

[0028] Figure 3 Schematic diagram of the compensation system of the present invention. DETAILED DESCRIPTION

[0029] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0030] Embodiment 1:

[0031] See also Figure 1-3 , the present invention provides a technical solution:

[0032] A method for compensating detection data of a nitrogen oxygen sensor standard part comprises the following steps:

[0033] By selecting different oxygen and nitrogen concentrations, different sampling atmosphere points are designed;

[0034] Place the master part and multiple standard parts under the same gas distribution table, and maintain the designed different atmosphere points for a certain period of time in turn, so that each atmosphere point can reach stability, and the sensor detection value can reach a stable state, and record the detection value at different atmosphere points. The sampling period of data recording should be much shorter than the time each atmosphere point maintains stability.

[0035] Data processing: The collected test values ​​of each standard part and the master part are processed. During the switching process of different atmosphere points, the atmosphere in the gas distribution table is not stable at this time, and the test data where the atmosphere has not reached a stable state is deleted.

[0036] Model training: First, the NOx detection value of the mother component is used as the output, and the NOx and O2 values ​​of the subcomponent are used as the input. The BP neural network can be used to train the model between the NOx detection values ​​of the subcomponent and the mother component. Other model training methods can also be selected. Secondly, the O2 detection value of the mother component is used as the output, and the NOx and O2 values ​​of the subcomponent are used as the input. The BP neural network is used to train the model between the O2 detection values ​​of the subcomponent and the mother component. The performance function uses the MAE mean absolute error. According to the above method, the NOx and O2 models between the remaining subcomponents and the mother component are trained.

[0037] Model loading: Since the detection performance of each standard part will slowly and continuously decay during operation, it is necessary to regularly train the model between each standard part and the master part. The standard part detection numerical compensation host has a model loading function. According to the format requirements of the host computer loading model, the trained model is converted into a format so that it can be loaded and calculated through the compensation host computer.

[0038] Equipment connection: Original calibration system such as Figure 1As shown, the calibration host computer sends a dew point command, which is transmitted to the nitrogen oxide sensor standard part through the CAN analyzer. After receiving the dew point command, the standard part transmits the message data of the current atmosphere point to the calibration host computer through the CAN analyzer in real time. The calibration host computer will collect the detection data of the standard part and each nitrogen oxide sensor in each atmosphere, and calibrate other nitrogen oxide sensors according to the detection values ​​of the standard part.

[0039] The modules of the compensation system of the present invention are connected as follows Figure 2 As shown, the calibration host computer sends a dew point command, which is transmitted to the nitrogen oxide sensor standard part through the 1#CAN analyzer and the gateway. After receiving the dew point command, the standard part transmits the message data of the current atmosphere point to the compensation host computer through the gateway and the 2#CAN analyzer in real time. The compensation host computer compensates the standard part detection value through the loaded training model, but in order to distinguish the compensated message data from the original message data, the message data ID needs to be changed. The compensated message data is sent to the 2#CAN analyzer. When passing through the gateway, the ID of this message is changed back to the original ID and transmitted to the calibration host computer through the 1#CAN analyzer. The calibration host computer calibrates each nitrogen oxide sensor through the compensated standard part detection value.

[0040] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for compensating the detection data of nitrogen and oxygen sensor standard parts, It is characterized in that The following steps are involved: 1) Select different oxygen and nitrogen concentrations and design different sampling atmosphere points; 2) Place the mother part and multiple standard parts under the same gas distribution table, and maintain the designed different atmosphere points for a certain period of time in turn, so that each atmosphere point reaches stability, and the sensor detection value reaches a stable state, and record its detection values ​​at different atmosphere points; 3) Processing the collected test values ​​of each standard part and the master part; 4) Taking the NOx detection value of the mother component as output and the NOx and O2 values ​​of the daughter component as input, a nonlinear relationship model is used to train a model between the NOx detection values ​​of the daughter component and the mother component. Then taking the O2 detection value of the mother component as output and the NOx and O2 values ​​of the daughter component as input, a nonlinear relationship model is used to train a model between the O2 detection values ​​of the daughter component and the mother component. 5) Regularly train the model between each standard part and the master part. The standard part detection numerical compensation host has a model loading function. According to the format requirements of the host computer loading model, the trained model is converted into a format so that the loading calculation can be performed through the compensation host computer; 6) The calibration host computer sends a dew point command, which is transmitted to the nitrogen oxide sensor standard part through the 1#CAN analyzer and the gateway. After receiving the dew point command, the standard part transmits the message data of the current atmosphere point in real time to the compensation host computer through the gateway and the 2#CAN analyzer. The compensation host computer compensates the standard part detection value through the loaded training model, and sends the compensated message data to the 2#CAN analyzer. When passing through the gateway, the ID of this message is changed back to the original ID, and transmitted to the calibration host computer through the 1#CAN analyzer. The calibration host computer calibrates each nitrogen oxide sensor through the compensated standard part detection value.

2. A method for compensating detection data of a nitrogen oxygen sensor standard component according to claim 1, Features: The sampling period of data recording in step 2) should be much shorter than the time that each atmosphere point maintains stability.

3. A method for compensating detection data of a nitrogen oxide sensor standard component according to claim 1, Features: In the process of switching between different atmosphere points in step 3), the atmosphere in the gas distribution station is not stable, and the detection data in which the atmosphere has not yet reached a stable state is deleted.

4. A method for compensating detection data of a nitrogen oxide sensor standard component according to claim 1, Features: In the step 4), a model of nonlinear relationship is used to train a model between the O2 detection values ​​of the sub-component and the parent component, and its performance function uses MAE mean absolute error.

5. A method for compensating detection data of a nitrogen oxygen sensor standard component according to claim 4, Features: In the step 4), the NOx and O2 models between the remaining sub-components and the parent component are trained according to the model training method of the nonlinear relationship.

6. A method for compensating detection data of a nitrogen oxygen sensor standard component according to claim 5, Features: The model of the nonlinear relationship in step 4) adopts BP neural network.

7. A method for compensating detection data of a nitrogen oxide sensor standard component according to claim 1, Features: During the operation of each standard component in step 5), the detection performance will slowly and continuously decay.

8. A method for compensating detection data of a nitrogen oxygen sensor standard component according to claim 1, Features: In the step 6), in order to distinguish the compensated message data from the original message data, the ID of the message data needs to be changed.