Off-gas detection error compensation method

Abstract

The invention discloses a sniffing detection error compensation method, which relates to the technical field of environment detection, and comprises the following steps: S1, data acquisition: collecting detection data and environment parameters of sniffing distinguishers in multiple dimensions, S2, establishment of a sniffing analysis model: extracting features from three aspects of individuals, working states and environments, the method comprises the following steps of: selecting a basic algorithm to construct a model according to personal information, historical detection performance, physiological index statistical values, workload and various environmental parameters, and counting the number of sniffing and distinguishing aiming at a smell adaptation problem, S3, error prediction and coefficient compensation, S4, result correction and S5, data transmission. According to the invention, through data acquisition and establishment of the olfactory analysis model, the physical signs of the olfactory discriminator and the fine parameters of the environment are matched with long-term data accumulation to provide massive and high-quality data for the data-driven olfactory detection error prediction model, and through error prediction, coefficient compensation and result correction, the accuracy of the olfactory detection error prediction model is improved. Therefore, accurate prediction and compensation of the sniff detection error are realized, and the accuracy of sniff detection is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of environmental detection, and in particular to a method for compensating olfactory detection errors. Background Art

[0002] In the field of environmental monitoring, malodors are widely generated in industrial and agricultural production, municipal sewage, sludge treatment, and garbage disposal processes. Many malodorous substances can endanger people's health and even their lives. In recent years, with the increasing awareness of indoor environmental protection, people are eager to have a safe, comfortable and healthy living space. To this end, monitoring of odor pollution in the environment and environmental odor pollution caused by sewage discharge, garbage disposal, etc. has become an important and indispensable industry in people's lives. Due to the large variety of malodorous substances, especially when multiple odors are mixed together, some instruments cannot identify them. Instead, the human sense of smell is needed to judge their environmental impact and pollution level. Although there are instruments to analyze malodorous substances, given the large number of malodorous substances and the complex transformation mechanism in the environment, different malodors bring different physiological sensations to people. "Smell" is more of a sensory evaluation, and it is difficult for instruments to judge human subjective feelings, so odorists need to use their noses to judge. Omnivores mainly use their noses to identify odors such as rotten eggs, mold, feces, and even unusual fragrances, determine the odor level, and confirm whether they are within the prescribed odor emission standards. However, due to individual differences in manual olfactory detection, decreased olfactory sensitivity due to long-term work, weakened adaptability to specific odors, and environmental factors, individual differences among odorists can easily lead to poor consistency in detection data, making it difficult to ensure the stability and reliability of the test results. Therefore, it is necessary to design an olfactory detection error compensation method to solve the above-mentioned problems. Summary of the Invention

[0003] The purpose of the present invention is to address the shortcomings of the prior art and to propose a method for compensating for olfactory detection errors to solve the problem of poor consistency in detection data caused by individual differences among olfactory examiners in the above technical solution.

[0004] To achieve the above object, the present invention is implemented through the following technical solutions: a method for compensating for olfactory detection errors, including the following compensation methods: S1. Data collection: Collect odor detectors' test data and environmental parameters in multiple dimensions. At the same time, the odor detectors' test reports and the use of biosensors to collect odor detectors' physiological indicators during the test process at high frequency are used. The fatigue state is judged based on the preset physiological indicator fluctuation threshold and a rest instruction is automatically issued. With the help of professional equipment, the meteorological parameters and environmental interference information of the test day are accurately collected. S2. Establish an olfactory analysis model: Based on the data collected in step S1, features are extracted from three aspects: individual, work status, and environment. These features cover personal basic information, historical test performance, physiological indicator statistics, workload, and various environmental parameters. A basic algorithm is used to build a model. For the problem of odor adaptation, the number of sniffs is counted. After reaching the threshold, an adaptation change model is constructed by combining time series and deep learning to predict changes in the olfactory threshold. S3. Error prediction and coefficient compensation: Based on the olfactory analysis model established in step S2, the IoT and edge computing are used to transmit and preprocess detection, working status, and environmental data in real time. The preprocessed data is input into the model to predict the detection error, determine the error size and direction, and calculate the compensation coefficient through linear transformation based on the predicted error and the historical maximum error value, in preparation for the correction of the detection results. S4. Result correction: Based on the error prediction and coefficient compensation of the olfactory analysis model in step S3, the compensation coefficient is applied to the original test result to complete the preliminary adjustment. Then, a multi-indicator error evaluation system is established to compare the adjustment result with the error calculated with the standard sample. If the error exceeds the threshold, Bayesian statistical analysis is used to fine-tune the compensation coefficient for secondary correction until the error meets the requirements; S5. Data transmission: Use a secure network protocol to transmit various types of data to the data center or cloud server, and establish an off-site backup mechanism to prevent data loss. On the storage side, use a distributed database to partition storage by data type and time, optimize the database regularly, and clean up expired data to ensure storage and query efficiency.

[0005] Furthermore, the step S1 includes the following specific implementation methods: T1. Multivariate data collection: First, the olfactory reports, environmental parameters, detection time, and detection methods of different odorants who detect the same sample at the same time are collected. At the same time, the working status of the odorants is systematically monitored with the help of biosensors. Then, the working frequency of the odorants is comprehensively analyzed. When the physiological fluctuation value reaches 30%, 60%, and 90% of the fluctuation threshold, a sound prompt and vibration warning are issued. When the physiological fluctuation value reaches 100% of the fluctuation threshold, a rest instruction is automatically sent. Secondly, the weather station and environmental monitoring equipment are used to collect parameters of the weather and working environment at the time, and the wind direction of the interference gas source at the detection site is recorded. T2. Long-term data accumulation: Establish a data storage system to collect multivariate data and classify and store them according to time indexes. At the same time, conduct sample olfactory tests on odor detectors every 7 days.

[0006] Furthermore, the step S2 includes the following specific implementation methods: T3. Feature extraction: Extract the personal information of the odorist as the basic feature, classify and compile the historical detection results of the odorist according to different odor types and detection environments, and calculate indicators such as accuracy and error rate as historical performance features; calculate the mean and variance of the physiological indicator data collected by the biosensor, convert it into work status characteristics, and analyze the relationship between work intensity and physiological indicators in combination with the workload of the day; normalize the meteorological parameters and use them as environmental features together with the interference odor sources and ventilation conditions at the detection site. At the same time, record the distribution of environmental data in different seasons and time periods to analyze the periodic impact of environmental factors, collect multi-dimensional features, help the model adapt to reality, and optimize error compensation; T4. Model establishment and training: A neural network algorithm is selected as the basic model. For the neural network, a three-layer structure is adopted, and the number of input layer nodes is determined according to the number of features; multiple basic models are integrated using the voting method, and different weights are assigned to each model according to its performance on the validation set; the number of times the odor detectors sniff each type of odor reaches 30 times and sets a data volume threshold. When the number of sniffs for each odor reaches the threshold, the time series analysis method is combined with the recurrent neural network of deep learning to construct an adaptive change model. The number of sniffs and time are used as input to predict changes in the olfactory threshold. The models are integrated and an adaptive change model is constructed to improve the error prediction accuracy.

[0007] Furthermore, the step S3 includes the following specific implementation methods: T5. Real-time data input and processing: Using IoT technology, real-time data collected from odor detectors, work status data, and environmental data is transmitted to the analysis system. Edge computing devices are used to perform preliminary cleaning and preprocessing of the data, such as removing outliers and filling missing values. Leveraging IoT and edge computing, preprocessing data improves efficiency and ensures quality. T6. Error Prediction: Input the preprocessed data into the trained olfactory detection error model to predict the olfactory detection error of a specific odorant under the current conditions. The model outputs the absolute value and direction of the error. A positive error indicates that the test result is too high, and a negative error indicates that the test result is too low. The input data predicts the error and clarifies the deviation, providing a key basis for compensation. T7. Compensation coefficient calculation: Based on the prediction error, the compensation coefficient is calculated using linear transformation. If the prediction error is positive, the compensation coefficient = 1-error value / maximum error value. If the prediction error is negative, the compensation coefficient = 1+error value / maximum error absolute value. The maximum error value is determined based on historical data statistics, and the compensation coefficient is calculated based on the error linear transformation to accurately adjust the test results.

[0008] Furthermore, the step S4 includes the following specific implementation methods: T8, Real-time adjustment: Apply the calculated compensation coefficient to the real-time olfactory test data to adjust the original test results. Apply the compensation coefficient to the real-time data, and make dynamic corrections to ensure accurate and timely results. T9. Error assessment: Establish an error assessment index system, including absolute error, relative error, and root mean square error. Compare the adjusted results with the known standard sample results, calculate various error indicators, establish an index system for comparison with the standard sample, and evaluate the accuracy of the test results from multiple dimensions. T10. Secondary correction: If the error assessment result exceeds the preset error threshold and the absolute error is greater than 0.5, the compensation coefficient is fine-tuned based on the error assessment result combined with the Bayesian statistical analysis method, and the test result is corrected again until the error is within an acceptable range. When the threshold is exceeded, Bayesian analysis is combined for fine-tuning to optimize the reliability of the test result.

[0009] Furthermore, the step S5 includes the following specific implementation methods: T11. Transmission: Use secure and reliable network transmission protocols to transmit collected data, model training results, and error assessment reports to a data center or cloud server. At the same time, establish a data backup mechanism to regularly back up data to an off-site server. Use secure protocols to transmit data and back it up off-site to ensure data security and integrity. T12. Storage: In data centers or cloud servers, use a distributed database to store data, partitioning storage based on data type and time for quick query and analysis. At the same time, regularly optimize the database to clean up expired data and improve data storage and query efficiency. Distributed storage and regular optimization improve data management and usage efficiency.

[0010] In summary, the present invention provides a method for compensating for olfactory detection errors, which has the following beneficial effects: 1. Through data collection and the establishment of an olfactory analysis model, the physiological indicators of the odorist and the subtle parameters of the environment, combined with long-term data accumulation, provide massive and high-quality data for the data-driven olfactory detection error prediction model. Through error prediction, coefficient compensation and result correction, accurate prediction and compensation of olfactory detection errors are achieved from multiple dimensions of individual, working status and environment. It can effectively quantify the individual differences of different odorists, offset these inconsistencies through compensation coefficient calculation and result correction, make the test results closer to the true value, and greatly improve the overall accuracy of olfactory detection.

[0011] 2. Biosensors track heart rate, respiratory rate, and sweat secretion in real time. Based on pre-set fluctuation thresholds, early warnings are issued and rest periods are automatically scheduled, effectively preventing detection errors caused by fatigue. Furthermore, an odor adaptation fluctuation model is constructed. By statistically analyzing sniff counts and data, the system accurately assesses changes in the detective's sensitivity to various odors and allows for timely adjustments to detection strategies.

[0012] 3. Real-time data input and processing, combined with efficient IoT and edge computing technologies, quickly complete data transmission and preprocessing, saving time for subsequent links. By establishing an error assessment indicator system and a secondary correction mechanism, it ensures that the test results are within an acceptable range, reduces the cost of repeated testing, and improves testing efficiency. Whether facing testing needs in different environments or responding to changes in the odor detector's own state, this solution demonstrates greater adaptability and effectively guarantees the reliability and effectiveness of olfactory testing. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 Schematic diagram of an error compensation architecture of an olfactory detection error compensation method according to the present invention. DETAILED DESCRIPTION

[0014] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0015] Example 1: See also Figure 1 As shown, the present invention provides a method for compensating for olfactory detection errors, including the following compensation methods: S1. Data collection: Collect odor detectors' test data and environmental parameters in multiple dimensions. At the same time, the odor detectors' test reports and the use of biosensors to collect odor detectors' physiological indicators during the test process at high frequency are used. The fatigue state is judged based on the preset physiological indicator fluctuation threshold and a rest instruction is automatically issued. With the help of professional equipment, the meteorological parameters and environmental interference information of the test day are accurately collected. S2. Establish an olfactory analysis model: Based on the data collected in step S1, features are extracted from three aspects: individual, work status, and environment. These features cover personal basic information, historical test performance, physiological indicator statistics, workload, and various environmental parameters. A basic algorithm is used to build a model. For the problem of odor adaptation, the number of sniffs is counted. After reaching the threshold, an adaptation change model is constructed by combining time series and deep learning to predict changes in the olfactory threshold. S3. Error prediction and coefficient compensation: Based on the olfactory analysis model established in step S2, the IoT and edge computing are used to transmit and preprocess detection, working status, and environmental data in real time. The preprocessed data is input into the model to predict the detection error, determine the error size and direction, and calculate the compensation coefficient through linear transformation based on the predicted error and the historical maximum error value, in preparation for the correction of the detection results. S4. Result correction: Based on the error prediction and coefficient compensation of the olfactory analysis model in step S3, the compensation coefficient is applied to the original test result to complete the preliminary adjustment. Then, a multi-indicator error evaluation system is established to compare the adjustment result with the error calculated with the standard sample. If the error exceeds the threshold, Bayesian statistical analysis is used to fine-tune the compensation coefficient for secondary correction until the error meets the requirements; S5. Data transmission: Use a secure network protocol to transmit various types of data to the data center or cloud server, and establish an off-site backup mechanism to prevent data loss. On the storage side, use a distributed database to partition storage by data type and time, optimize the database regularly, and clean up expired data to ensure storage and query efficiency.

[0016] Step S1 includes the following specific implementation methods: T1. Multivariate data collection: First, the olfactory reports, environmental parameters, detection time and detection methods of different olfactory experts who detect the same sample at the same time are collected. At the same time, the working status of the olfactory experts is systematically monitored with the help of biosensors. The olfactory reports of different olfactory experts on the same sample at the same time are collected. Through multi-data comparison, the errors caused by individual differences can be effectively reduced, making the detection results more reliable and representative. At the same time, combined with environmental parameters, detection time and detection methods, a comprehensive analysis of the impact of environmental factors and detection processes on the olfactory results is conducted, providing rich data support for subsequent error correction, thereby significantly improving the accuracy of the detection data. Then, a comprehensive analysis of the working frequency of the olfactory experts is conducted. When the physiological fluctuation value reaches 30%, 60% and 90% of the fluctuation threshold, a sound prompt and vibration warning are issued. When the physiological fluctuation value reaches 100% of the fluctuation threshold, a rest instruction is automatically sent. The working status of the olfactory experts is systematically monitored with the help of biosensors. According to the graded warning of the physiological fluctuation value, a sound prompt and vibration warning are issued when the physiological fluctuation value reaches 30%, 60% and 90% of the fluctuation threshold, which can be provided in advance. The odor detector is reminded to pay attention to his or her own state and adjust the work rhythm in time to avoid detection errors due to fatigue or physical discomfort; and when the value reaches 100%, a rest instruction is automatically sent to force the odor detector to rest, which can effectively prevent the decrease in olfactory sensitivity due to excessive fatigue, thereby ensuring that the odor detector always maintains a good working state during the detection process and guarantees the stability of the detection quality. Secondly, the weather station and environmental monitoring equipment are used to collect parameters of the weather and working environment at the time, and record the wind direction of whether there is an interfering gas source at the detection site. The weather station and environmental monitoring equipment are used to collect weather and working environment parameters, and record the wind direction of whether there is an interfering gas source at the detection site. The complex impact of environmental factors on olfactory detection can be analyzed in detail. Clarifying the mechanism of the effect of different meteorological conditions (such as temperature, humidity, air pressure, etc.) and interfering gas sources on olfactory identification results will help to more accurately consider environmental factors during data processing and model construction, improve the pertinence and effectiveness of olfactory detection error compensation, and make the detection results more in line with actual conditions; T2. Long-term data accumulation: Establish a data storage system to collect multivariate data and classify and store them according to time indexes. Classification and storage according to time indexes facilitate data retrieval and improve management efficiency. At the same time, accumulate data over a long period of time to help discover the changing patterns of olfactory senses and environmental factors of odorants, optimize error compensation models, and conduct sample olfactory tests on odorants every 7 days to maintain data consistency, ensure that data at each stage are comparable, and make analysis more scientific and reliable.

[0017] Step S2 includes the following specific implementation methods: T3. Feature extraction: Extract the personal information of the odorist as the basic feature, classify and count the historical detection results of the odorist according to different odor types and detection environments, and calculate indicators such as accuracy and error rate as historical performance features; personal information and historical detection indicators reflect the basic characteristics and past performance of the odorist; calculate the mean and variance of the physiological indicator data collected by the biosensor, convert it into work status characteristics, and analyze the relationship between work intensity and physiological indicators in combination with the workload of the day; physiological indicators combined with workload can accurately grasp the work status; normalize the meteorological parameters and use them together with the interference odor sources and ventilation conditions at the detection site as environmental features, and record the distribution of environmental data in different seasons and time periods to analyze the periodic impact of environmental factors; environmental features cover meteorological and on-site conditions, taking into account seasonal time distribution, which helps to analyze the impact of environmental factors, improve the adaptability of the model to complex reality, and make error compensation more scientific and effective; T4. Model establishment and training: A neural network algorithm is selected as the basic model. For the neural network, a three-layer structure is adopted, and the number of input layer nodes is determined according to the number of features; a voting method is used to integrate multiple basic models, and different weights are assigned to each model according to its performance on the validation set; the number of times the odor discerner sniffs each type of odor is counted, and a data volume threshold is set. When the number of sniffs for each odor reaches 30 times and reaches the threshold, a time series analysis method is combined with a deep learning recurrent neural network to construct an adaptive change model, which uses the number of sniffs and time as input to predict changes in the olfactory threshold; a neural network is selected as the basic model, combined with an integrated learning voting method, to give full play to the advantages of each model and improve the model's accuracy and generalization ability.

[0018] Step S3 includes the following specific implementation methods: T5. Real-time data input and processing: Through the Internet of Things (IoT) technology, the real-time data collected from odor detectors, work status data, and environmental data is transmitted to the analysis system. Edge computing devices are used to perform preliminary cleaning and preprocessing of the data, such as removing outliers and filling missing values. Using IoT and edge computing technologies, real-time data transmission and preprocessing are achieved. Various data types are quickly transmitted to the analysis system. Edge computing devices pre-clean the data, remove outliers, fill missing values, reduce invalid data interference, and ensure data quality. T6. Error Prediction: Input the preprocessed data into the trained olfactory detection error model to predict the olfactory detection error of a specific odorant under the current conditions. The model outputs the absolute value and direction of the error. A positive error indicates that the test result is too high, and a negative error indicates that the test result is too low. Input the preprocessed data into the model prediction error, and output the absolute value and direction of the error. This will clearly indicate the deviation between the test result and the true value, and accurately determine whether the test result is too high or too low, providing a key basis for subsequent compensation. T7. Compensation coefficient calculation: Based on the prediction error, the compensation coefficient is calculated using linear transformation. If the prediction error is positive, the compensation coefficient = 1-error value / maximum error value. If the prediction error is negative, the compensation coefficient = 1+error value / maximum error absolute value. The maximum error value is determined based on historical data statistics. The compensation coefficient is calculated using linear transformation based on the prediction error, and the maximum error value is determined based on historical data. Through reasonable mathematical calculation methods, corresponding compensation coefficients are generated for errors of different directions and sizes, providing a quantitative basis for real-time detection data correction.

[0019] Step S4 includes the following specific implementation methods: T8. Real-time adjustment: Apply the calculated compensation coefficient to the real-time olfactory test data to adjust the original test results. Apply the compensation coefficient to the real-time olfactory test data to make immediate adjustments to the original results. Dynamically correct the data during the test process to make the test results closer to the true value, avoid excessive deviation in the results due to error accumulation, and ensure the timeliness and accuracy of the test results. T9. Error evaluation: Establish an error evaluation index system, including absolute error, relative error, and root mean square error. Compare the adjusted results with the known standard sample results and calculate various error indicators. T10. Secondary correction: If the error assessment result exceeds the preset error threshold and the absolute error is greater than 0.5, the compensation coefficient is fine-tuned based on the error assessment result and combined with the Bayesian statistical analysis method, and the test result is corrected again until the error is within an acceptable range; an evaluation index system including absolute error, relative error and root mean square error is established, and the error index is calculated by comparison with the standard sample to comprehensively evaluate the accuracy of the test results and measure the error size from multiple angles.

[0020] The S5 step includes the following specific implementation methods: T11. Transmission: Use secure and reliable network transmission protocols to transmit collected data, model training results, and error assessment reports to a data center or cloud server. At the same time, establish a data backup mechanism to regularly back up data to an off-site server. Use secure network protocols to transmit data, model results, and assessment reports to a data center or cloud server, and establish an off-site backup mechanism to ensure data transmission security and prevent data from being stolen or tampered with during transmission. T12. Storage: In data centers or cloud servers, a distributed database is used to store data, and partitioned storage is performed based on data type and time for quick query and analysis. At the same time, the database is regularly optimized to clean up expired data and improve data storage and query efficiency. In data centers or cloud servers, a distributed database is used to partition storage based on data type and time to facilitate quick retrieval and analysis of specific data, reduce data redundancy, and save storage space.

[0021] Example 2: This embodiment is a method for compensating for olfactory detection errors, including a specific monitoring method during the multivariate acquisition process in step T1 and a specific capture method for model building and training in step T4: In step T1, to address measurement fluctuations caused by changes in olfactory sensitivity over time, an odorant work status monitoring system was introduced. Wearable biosensors continuously track the odorant's physiological indicators, such as heart rate, respiratory rate, and sweat secretion, to assist in determining their level of olfactory fatigue. When the monitoring data indicates a high level of fatigue, a rest command is automatically triggered to prevent a decrease in detection accuracy due to overwork. This system accurately determines the level of olfactory fatigue and effectively prevents a decrease in olfactory sensitivity due to overwork, ensuring the accuracy and reliability of olfactory detection results.

[0022] In step T4, since odor discerners often sniff the same odor, they tend to adapt to it and reduce their sensitivity to it. The number of times the odor discerners sniff is counted and combined with the odor data records. When the amount of data reaches a certain level, data modeling is used to build a model of the odor discerners' adaptation and change to various odors. Within a certain period of time, after a certain number of times of sniffing a certain odor, the odor threshold will fluctuate significantly. By capturing the associated influencing factors through data analysis, the odor discerners' latest thresholds for various odors can be determined. By counting the number of times the odor discerners sniff various odors and combining it with the data records, data modeling is carried out after the data volume reaches the standard to build an adaptation and change model. This model can capture the fluctuation of the odor threshold after a specific number of sniffs within a certain period of time, accurately determine the odor discerners' latest thresholds for various odors, and effectively deal with olfactory adaptability.

[0023] Example 3: This embodiment is a method for compensating for olfactory detection errors. In step T2, an index table of a data storage system is established: Insert links in test data, physiological indicators, weather parameters and environmental parameters to view their detailed parameter reports.

[0024] Through a time-indexed storage system, massive amounts of data can be collected and organized in an orderly manner, facilitating rapid retrieval and call-up, providing specific test data for establishing olfactory analysis models, error prediction, and coefficient compensation. At the same time, it can clearly analyze the changing trends of the olfactory detective's olfactory over working hours, as well as the impact of environmental factors on test data.

[0025] The above description is merely a preferred embodiment of the present invention and does not constitute any other form of limitation to the present invention. Any person skilled in the art may utilize the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes for application in other fields. However, any simple modification, equivalent change, and modification of the above embodiments made in accordance with the technical essence of the present invention without departing from the technical solution of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. A method for compensating for olfactory detection errors, comprising the following compensation methods: S1. Data Collection: Multi-dimensional collection of odorant detection data and environmental parameters. Meanwhile, the odorant's detection report is compiled. Biosensors are used to collect the odorant's physiological indicators during the detection process at high frequency. Fatigue status is determined based on preset physiological indicator fluctuation thresholds and rest instructions are automatically issued. Professional equipment is used to accurately collect meteorological parameters and environmental interference information for the day of detection. S2. Build an olfactory analysis model: Based on the data collected in step S1, extract features from three aspects: individual, work status, and environment. These features include basic personal information, historical test performance, physiological indicator statistics, workload, and various environmental parameters. A basic algorithm is used to build a model. To address odor adaptation, the number of sniffs is counted. Once the threshold is reached, an adaptation change model is constructed by combining time series and deep learning to predict changes in olfactory thresholds. S3. Error prediction and coefficient compensation: Based on the olfactory analysis model established in step S2, the IoT and edge computing are used to transmit and preprocess detection, working status, and environmental data in real time. The preprocessed data is input into the model to predict the detection error, determine the error size and direction, and calculate the compensation coefficient through linear transformation based on the predicted error and the historical maximum error value, in preparation for the correction of the detection results. S4. Result correction: Based on the error prediction and coefficient compensation of the olfactory analysis model in step S3, the compensation coefficient is applied to the original test results to complete preliminary adjustment. Then, a multi-index error evaluation system is established to compare the adjusted results with the error calculated from the standard sample. If the error exceeds the threshold, Bayesian statistical analysis is used to fine-tune the compensation coefficient for secondary correction until the error meets the requirements; S5. Data transmission: Use secure network protocols to transmit various types of data to data centers or cloud servers, and establish off-site backup mechanisms to prevent data loss. On the storage side, use distributed databases to partition storage by data type and time, regularly optimize the database, and clean up expired data to ensure storage and query efficiency.

2. The method for compensating for olfactory detection errors according to claim 1, wherein: The S1 step includes the following specific implementation methods: T1. Multivariate data collection: First, the olfactory reports, environmental parameters, detection time, and detection methods of different odorants who detect the same sample at the same time are collected. At the same time, the working status of the odorants is systematically monitored with the help of biosensors. Then, the working frequency of the odorants is comprehensively analyzed. When the physiological fluctuation value reaches 30%, 60%, and 90% of the fluctuation threshold, a sound prompt and vibration warning are issued. When the physiological fluctuation value reaches 100% of the fluctuation threshold, a rest instruction is automatically sent. Secondly, the weather station and environmental monitoring equipment are used to collect parameters of the weather and working environment at the time, and the wind direction of the interference gas source at the detection site is recorded. T2. Long-term data accumulation: Establish a data storage system to collect multivariate data and classify and store them according to time indexes. At the same time, conduct sample olfactory tests on odor detectors every 7 days.

3. The method for compensating for olfactory detection errors according to claim 1, wherein: The S2 step includes the following specific implementation methods: T3. Feature extraction: Extract the personal information of the odorant as the basic features, classify and compile the odorant's historical detection results according to different odor types and detection environments, and calculate indicators such as accuracy and error rate as historical performance features; The mean and variance of the physiological indicator data collected by the biosensor are calculated and converted into work status characteristics. The relationship between work intensity and physiological indicators is analyzed in combination with the daily workload. Meteorological parameters are normalized and used as environmental characteristics along with the interference odor sources and ventilation conditions at the detection site. The distribution of environmental data across different seasons and time periods is recorded to analyze the cyclical impact of environmental factors. T4. Model establishment and training: A neural network algorithm is selected as the basic model. For the neural network, a three-layer structure is adopted, and the number of input layer nodes is determined according to the number of features; multiple basic models are integrated using the voting method, and different weights are assigned to each model according to its performance on the validation set; the number of times the odor detectors sniff each type of odor reaches 30 times and sets a data volume threshold. When the number of sniffs for each odor reaches the threshold, a time series analysis method is combined with a deep learning recurrent neural network to construct an adaptive change model. The number of sniffs and time are used as input to predict changes in the olfactory threshold.

4. The method for compensating olfactory detection errors according to claim 1, wherein: The S3 step includes the following specific implementation methods: T5. Real-time data input and processing: Using IoT technology, the real-time data collected from odor detectors, work status data, and environmental data is transmitted to the analysis system. Edge computing devices are used to perform preliminary cleaning and preprocessing of the data, such as removing outliers and filling in missing values. T6. Error prediction: The preprocessed data is input into the trained olfactory detection error model to predict the olfactory detection error of a specific odorant under the current conditions. The model outputs the absolute value and direction of the error. A positive error indicates that the test result is too high, and a negative error indicates that the test result is too low. T7. Compensation coefficient calculation: Based on the forecast error, the compensation coefficient is calculated using linear transformation. If the forecast error is positive, the compensation coefficient = 1 - error value / maximum error value. If the forecast error is negative, the compensation coefficient = 1 + error value / maximum error absolute value. The maximum error value is determined based on historical data statistics.

5. The method for compensating olfactory detection errors according to claim 1, wherein: The S4 step includes the following specific implementation methods: T8, real-time adjustment: applying the calculated compensation coefficient to the real-time olfactory detection data to adjust the original detection results; T9. Error evaluation: Establish an error evaluation index system, including absolute error, relative error, and root mean square error. Compare the adjusted results with the known standard sample results and calculate various error indicators. T10. Secondary correction: If the error assessment result exceeds the preset error threshold and the absolute error is greater than 0.5, the compensation coefficient is fine-tuned based on the error assessment result and combined with the Bayesian statistical analysis method, and the test result is corrected again until the error is within an acceptable range.

6. The method for compensating olfactory detection errors according to claim 1, wherein: The S5 step includes the following specific implementation methods: T11. Transmission: Use secure and reliable network transmission protocols to transmit collected data, model training results, and error assessment reports to a data center or cloud server. At the same time, establish a data backup mechanism to regularly back up data to an off-site server. T12. Storage: In data centers or cloud servers, data is stored in a distributed database, partitioned by data type and time for quick query and analysis. At the same time, the database is regularly optimized to clean up expired data and improve data storage and query efficiency.

Images