Real-time data acquisition and analysis method in industrial intelligent manufacturing

Abstract

The invention discloses a real-time data acquisition and analysis method in industrial intelligent manufacturing, and relates to the technical field of industrial automation and intelligent manufacturing, and the method comprises the following steps: air quality monitoring equipment carries out conventional detection on the concentration of harmful gas in the environment according to a pre-configured time sequence resolution, and obtains air quality parameter information in real time; according to the invention, the response sensitivity and intelligent identification capability of the air quality monitoring equipment to instantaneous high-concentration harmful gas leakage are improved, and the monitoring limitation of traditional fixed time sequence resolution is broken through. And through fusion of feature engineering and a machine learning model, accurate judgment of dynamic change of gas is realized. After the severe fluctuation state is identified, the system dynamically adjusts the sampling frequency and starts the cache, the concentration peak value is effectively captured, meanwhile, the storage pressure is relieved, the stability and the real-time performance of the system are enhanced, and the safety protection capability of the industrial environment is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of industrial automation and intelligent manufacturing, and specifically to a real-time data acquisition and analysis method in industrial intelligent manufacturing. BACKGROUND

[0002] Real-time data acquisition and analysis in industrial intelligent manufacturing refers to the continuous and dynamic collection of operational data from key links such as production lines, equipment, process parameters, and environmental monitoring (such as air quality) in an intelligent manufacturing environment using sensors, Industrial Internet of Things (IIoT) devices, and other technical means. Through edge computing, cloud computing, or artificial intelligence algorithms, data is analyzed and processed in real time. This process aims to achieve rapid perception of production status, prediction and early warning of equipment failure, optimization and adjustment of process parameters, and improvement of overall manufacturing efficiency. This process emphasizes "real-time", i.e. the process from data collection to analysis has extremely low delay, ensuring that the manufacturing system can respond quickly to unexpected situations, and is the foundation for achieving efficient, intelligent, and adaptive industrial production.

[0003] Air quality monitoring equipment refers to instruments and systems used to detect and assess the concentration of various pollutants in the air. It integrates sensors and analysis devices to monitor harmful gases (such as carbon dioxide, carbon monoxide, nitrogen oxides, volatile organic compounds, etc.), particulate matter (such as PM2.5, PM10), and temperature and humidity in real time. Its role is to continuously collect data to assess air quality conditions, help identify potential air pollution sources and hazards, and provide data support for environmental protection, industrial safety, and health monitoring. In industrial environments, air quality monitoring equipment is particularly important as it can monitor harmful gas concentrations in workshops or production environments in real time, ensuring compliance with safety and environmental standards, and preventing health problems or production accidents caused by air pollution. At the same time, it can also serve as a tool for optimizing production environments and reducing energy waste, promoting green manufacturing and sustainable development for enterprises.

[0004] The existing technology has the following shortcomings: The air quality monitoring equipment in the prior art usually detects the concentration of harmful gases in the air at a pre-configured time resolution, aiming to ensure that the system can stably and continuously collect and record data within a certain period of time. However, in actual monitoring, the equipment may encounter some instantaneous and high-concentration gas leaks, which usually have very short duration and extremely fast release speed. If the equipment continues to detect at the preset time resolution, it may not be able to capture the dramatic fluctuations in gas concentration in a timely manner, especially for those leak events that occur in a short period of time. This situation may result in the equipment failing to detect high concentrations of dangerous gases at the moment of the leak, thus missing the early warning opportunity, increasing the risk of accidents, and posing a serious threat to personnel health.

[0005] The above information disclosed in the BACKGROUND section merely to enhance the understanding of the background of the present disclosure, and therefore it can include information that does not constitute prior art known to those of ordinary skill in the art. SUMMARY

[0006] The purpose of the present application is to provide a real-time data acquisition and analysis method in industrial intelligent manufacturing, which improves the response sensitivity and intelligent recognition ability of air quality monitoring equipment to instantaneous high-concentration harmful gas leakage, and breaks through the monitoring limitations of traditional fixed time resolution. Through the fusion of feature engineering and machine learning model, accurate judgment of gas dynamic change is realized. After identifying the state of severe fluctuation, the system dynamically adjusts the sampling frequency and enables the cache, effectively capturing the concentration peak, while relieving the storage pressure, enhancing the system stability and real-time performance, and improving the safety protection ability of the industrial environment, to solve the problems in the above background technology.

[0007] In order to achieve the above purpose, the present application provides the following technical scheme: a real-time data acquisition and analysis method in industrial intelligent manufacturing, comprising the following steps: The air quality monitoring equipment detects the harmful gas concentration in the environment according to the pre-configured time resolution, and obtains the air quality parameter information in real time; The obtained original air quality parameter data is preprocessed, and the preprocessed data is uniformly constructed into a structured data set; The feature engineering technology is applied to extract key indicators representing the severe fluctuation characteristics of harmful gas concentration in the current environment from the data set, and statistical analysis and trend modeling are performed on the extracted key indicators to quantify the fluctuation degree of the current harmful gas; The key parameter indicators processed by analysis are constructed into a feature vector and input into a pre-trained machine learning model, and the multivariate correlation between the features is deeply mined and modeled by the machine learning model to realize intelligent identification and evaluation of the dynamic change state of the harmful gas concentration in the current environment; When the machine learning model determines that the harmful gas in the current environment is in a state of severe fluctuation, the time resolution of the air quality monitoring equipment is dynamically adjusted based on the evaluation result to improve its detection sensitivity to instantaneous high-concentration leakage events; at the same time, the cache mechanism is started to temporarily store the high-frequency collected data, thereby relieving the data storage system pressure and providing sufficient data processing window period for subsequent data processing and analysis.

[0008] Preferably, the obtained original air quality parameter data is preprocessed, and the preprocessed data is constructed into a structured data set, the specific steps are as follows: Firstly, denoising is performed to filter out abnormal values caused by sensor fluctuation and communication interference factors, and to eliminate high-frequency interference and invalid information in the data; Secondly, missing values ​​are filled in using interpolation methods, historical mean methods, and model prediction methods to ensure data integrity. Next, data format standardization was implemented to unify the timestamp format, measurement unit, and numerical precision of sensor outputs, ensuring structural consistency of multi-source data. Then, the parameters are normalized to bring different physical quantities to a uniform scale, which is beneficial for feature extraction and modeling analysis. Finally, following a unified field naming convention and time alignment strategy, the processed multi-parameter data is organized into a structured table, with each row corresponding to a time node and each column corresponding to an air quality parameter, ultimately forming a high-quality dataset for subsequent intelligent analysis.

[0009] Preferably, feature engineering techniques are applied to extract key indicators characterizing the drastic fluctuations in the concentration of harmful gases in the current environment from the dataset. The extracted indicators include the number and magnitude of brief and drastic increases in gas concentration per unit time, and the speed and extent of the spread of abnormally high concentration trends from a certain monitoring point to surrounding points. The number and magnitude of brief and drastic increases in gas concentration per unit time, and the speed and extent of the spread of abnormally high concentration trends from a certain monitoring point to surrounding points are analyzed under the detection window to generate reference values ​​for the appearance of instantaneous gas concentration peaks and reference values ​​for the spread of abnormal gas concentration trends. The degree of fluctuation of the current harmful gases is quantified by the reference values ​​for the appearance of instantaneous gas concentration peaks and the reference values ​​for the spread of abnormal gas concentration trends.

[0010] Preferably, the specific steps for analyzing the number and magnitude of brief and dramatic increases in gas concentration per unit time within the detection window to generate reference values ​​for instantaneous gas concentration peaks are as follows: First, identify gas concentration spikes occurring per unit time, i.e., gas concentration changes that occur in a sudden increase-decline pattern within a very short period. For each identified spike, calculate its corresponding amplitude factor to quantify the intensity of the event. The calculation expression is as follows: In the formula, It is the peak amplitude factor, quantized first The intensity of a gas concentration spike event, It is the first The maximum concentration value in a gas concentration spike event. It is the baseline concentration before the gas concentration spike; After identifying all peak events and calculating the peak amplitude factor, in order to comprehensively measure the severity of harmful gas concentration fluctuations within the detection window, the intensity and frequency of all events are considered together. A nonlinear weighted accumulation mechanism is introduced to generate a reference value for the appearance of instantaneous gas concentration peaks, which is used to characterize the activity and severity of concentration fluctuations in the current environment. The generation formula is as follows: In the formula, This is a reference value for the appearance of instantaneous gas concentration peaks. It is the total number of spike events within the detection window. It is used to suppress the cumulative error amplification of high-frequency weak fluctuations, while enhancing the dominant effect of large-amplitude spikes.

[0011] Preferably, the specific steps for analyzing the speed and extent of the diffusion of abnormally high concentration trends from a certain monitoring point to surrounding points within the detection window to generate reference values ​​for the diffusion of abnormal gas concentration trends are as follows: First, select a monitoring point as the center, and simultaneously acquire data from the surrounding area of ​​the current monitoring point. Data from several neighboring sensors is collected, and the distance between each sensor and the center point is recorded. The spatial diffusion factor is calculated using the following expression: In the formula, It is the spatial diffusion factor. It is the first Gas concentration values ​​at neighboring monitoring points, It is the gas concentration at the central monitoring point. It is the number of neighboring monitoring points involved in the calculation. It is a non-linear adjustment index. It is the spatial weight function, and its expression is as follows: In the formula, It is the natural base. It is the distance decay factor. It is the center point and the The actual physical distance between the monitoring points; Obtaining the spatial diffusion factor Then, further The parameters are converted into standardized parameters for identifying severe fluctuations. To enhance the response to different diffusion patterns, a multidimensional nonlinear enhanced compression model is used to generate reference values ​​for the diffusion of gas concentration anomalies. The generation formula is as follows: In the formula, This is a reference value for the diffusion of abnormal gas concentration trends. It is a nonlinear enhancement factor. These are the weighting coefficients of the square root enhancement term. It involves exponentiation of the entire expression.

[0012] Preferably, the reference values ​​for the appearance of instantaneous gas concentration peaks and the reference values ​​for the diffusion of abnormal gas concentration trends, after analysis, are used as feature vectors and input into a pre-trained machine learning model. The machine learning model generates a hazardous gas concentration fluctuation coefficient, and the multivariate correlation between various features is deeply mined and modeled through the hazardous gas concentration fluctuation coefficient, so as to realize the intelligent identification and assessment of the dynamic change state of hazardous gas concentration in the current environment.

[0013] Preferably, the harmful gas concentration fluctuation coefficient generated by the pre-trained machine learning model through in-depth mining and modeling analysis of the multivariate correlation between various features is compared with a pre-set reference threshold for the harmful gas concentration fluctuation coefficient to intelligently identify and assess the dynamic changes in the current environment's harmful gas concentration. The specific steps are as follows: If the concentration fluctuation coefficient of harmful gases is greater than the preset reference threshold for the concentration fluctuation coefficient of harmful gases, it is determined that the concentration of harmful gases in the current environment is in a state of drastic fluctuation; if the concentration fluctuation coefficient of harmful gases is less than or equal to the preset reference threshold for the concentration fluctuation coefficient of harmful gases, it is determined that the concentration of harmful gases in the current environment is in a normal state.

[0014] Preferably, when the machine learning model determines that the harmful gases in the current environment are in a state of drastic fluctuation, the time-series resolution of the air quality monitoring equipment is dynamically adjusted based on the evaluation results. At the same time, a high-speed caching mechanism is activated to temporarily store the high-frequency collected data. The specific steps are as follows: After determining that the harmful gases in the current environment are in a state of drastic fluctuation, the temporal resolution of the air quality monitoring equipment will be adaptively adjusted according to the magnitude of the fluctuation, thereby improving the equipment's response capability to instantaneous high concentration peaks. The calculation expression is as follows: In the formula, It is the time series resolution dynamically adjusted under fluctuating conditions. This is the device's default timing resolution. This is a reference threshold for the coefficient of variation of harmful gas concentration. It is the sensitivity adjustment coefficient; After the device sampling frequency increases, a high-speed cache mode is automatically activated to temporarily store the large amount of air quality data streams generated at high frequencies in a circular buffer. This prevents data writing to storage from causing blockages and provides data assurance for subsequent processing. The cache window calculation expression is as follows: In the formula, This is the optimal window period for continuous cache writes. It is the cache capacity adjustment factor. It is the number of concurrent write channels in the device cache structure, used to improve the parallelism of data processing.

[0015] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention achieves high sensitivity, intelligent response, and system stability assurance for air quality monitoring equipment in the face of sudden high-concentration leaks of harmful gases. This method not only overcomes the limitations of traditional fixed-time-resolution monitoring mechanisms in capturing drastic fluctuations, but also achieves intelligent identification and accurate judgment of dynamic changes in ambient gases through deep integration of feature engineering and machine learning models. After determining that the gas is in a state of drastic fluctuation, the system can dynamically increase the sampling frequency and activate a high-speed caching mechanism, ensuring real-time capture of key concentration peaks while effectively alleviating storage pressure and improving the system's processing capacity under high-frequency data loads. The overall solution significantly enhances the monitoring system's response speed and judgment accuracy to sudden gas leaks, improves the safety protection level and operational reliability of industrial sites, and has good engineering practical value and promising prospects for promotion. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0017] Figure 1 This is a flowchart of the real-time data acquisition and analysis method in industrial intelligent manufacturing according to the present invention. Detailed Implementation

[0018] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0019] This invention provides, for example Figure 1 The real-time data acquisition and analysis method in intelligent industrial manufacturing, as shown, includes the following steps: Air quality monitoring equipment routinely detects the concentration of harmful gases in the environment according to a pre-configured time resolution and acquires air quality parameter information in real time; Air quality monitoring equipment routinely detects the concentration of harmful gases in the environment according to a pre-configured time-series resolution. This means the equipment continuously collects concentration data of harmful gases in the air at a fixed frequency based on the set time precision (i.e., time-series resolution), thus forming a continuous monitoring data sequence with time reference. During this process, the equipment not only detects the concentration of specific harmful gases (such as carbon monoxide, methane, and ammonia), but also acquires other air quality parameters in real time, such as temperature, humidity, wind speed, and air pressure, to help determine the causes and environmental characteristics of gas concentration changes. By conducting stable routine monitoring at the set time-series resolution, the system can comprehensively grasp the dynamic trends of air environment changes, providing a reliable data foundation for subsequent anomaly detection, risk assessment, and emergency response.

[0020] The purpose of routinely detecting the concentration of harmful gases in the environment according to the pre-configured time resolution is to ensure that the equipment can stably collect basic data under normal conditions without consuming excessive resources or causing excessive system load. By maintaining a reasonable fixed sampling interval, it can provide basic monitoring of changes in ambient gases and establish a reference baseline for subsequent dynamic adjustments.

[0021] The acquired raw air quality parameter data are preprocessed, and the preprocessed data is then uniformly constructed into a structured dataset. During routine testing, the equipment acquires a series of air quality parameters in real time, such as harmful gas concentration, air temperature, humidity, and terminal velocity. Since the raw data often contains noise, missing values, or outliers, it is necessary to deduce these to ensure the accuracy of subsequent analysis. For example, methods such as noise removal, missing value correction, missing value completion, and format normalization can be used to improve the validity and consistency of the data. The processed data is then uniformly stored in a dataset, becoming an important foundation for subsequent feature extraction and model analysis.

[0022] The acquired raw air quality parameter data is preprocessed and then constructed into a structured dataset. The specific steps are as follows: First, noise reduction is performed to filter out outliers caused by sensor fluctuations and communication interference, and to remove high-frequency interference and invalid information from the data. Second, missing values ​​are filled in using interpolation, historical mean, and model prediction methods to ensure data integrity. Next, data format standardization is performed to unify the timestamp format, measurement units, and numerical precision of sensor outputs, ensuring structural consistency of multi-source data. Then, various parameters are normalized to bring different physical quantities to a uniform scale, which is beneficial for feature extraction and modeling analysis. Finally, the processed multi-parameter data is organized into a structured table according to a unified field naming convention and time alignment strategy, with each row corresponding to a time node and each column corresponding to an air quality parameter, ultimately forming a high-quality dataset for subsequent intelligent analysis.

[0023] Feature engineering techniques are applied to extract key indicators from the dataset that characterize the drastic fluctuations in the concentration of harmful gases in the current environment. Statistical analysis and trend modeling are then performed on the extracted key indicators to quantify the degree of fluctuation in the concentration of harmful gases. Feature engineering techniques are applied to extract key indicators characterizing the drastic fluctuations in the concentration of harmful gases in the current environment from the dataset. The extracted indicators include the frequency and amplitude of brief and drastic increases in gas concentration per unit time, and the speed and extent of the spread of abnormally high concentration trends from a certain monitoring point to surrounding points. The frequency and amplitude of brief and drastic increases in gas concentration per unit time, and the speed and extent of the spread of abnormally high concentration trends from a certain monitoring point to surrounding points are analyzed under the detection window to generate reference values ​​for the appearance of instantaneous gas concentration peaks and reference values ​​for the spread of abnormal gas concentration trends. The degree of fluctuation of the current harmful gas is quantified by the reference values ​​for the appearance of instantaneous gas concentration peaks and the spread of abnormal gas concentration trends.

[0024] A rapid increase in the frequency and magnitude of brief, sharp spikes in gas concentration within a unit of time clearly indicates that the concentration of harmful gases in the current environment is experiencing drastic fluctuations. This phenomenon suggests that the concentration of harmful gases in the air is undergoing abnormal and unstable drastic changes, characterized by high-frequency, high-amplitude concentration anomalies, typically caused by sudden leaks, abnormal equipment emissions, or abnormal chemical reactions. When the monitoring system frequently detects sharp spikes in gas concentration within a relatively short time window, and these spikes are not only large in magnitude and short in duration, but also extremely short in interval and highly repetitive, it indicates that the pollution source is exhibiting highly unstable or intermittent, drastic release characteristics. This type of fluctuation differs from slow increases or periodic changes; instead, it manifests as strong, irregular, abnormal disturbances, reflecting a highly abnormal and unstable state of current air quality. It is highly likely to foreshadow a surge in safety risks and can therefore serve as one of the core reference indicators for judging drastic fluctuations.

[0025] The specific steps for analyzing the frequency and magnitude of brief, dramatic increases in gas concentration within a unit of time to generate reference values ​​for instantaneous gas concentration peaks within a detection window are as follows: First, identify gas concentration spikes occurring within a unit of time, which are gas concentration changes that occur in a sudden increase-decline pattern within a very short period of time. A spike event can be defined as: a gas concentration instantaneously increases above a certain fluctuation threshold at a certain sampling point, and then rapidly decreases to near its original level within several sampling points, without maintaining a high value in between. For each identified spike event, calculate its corresponding amplitude factor to quantify the intensity of the event. The calculation expression is as follows: In the formula, It is the peak amplitude factor, quantized first The intensity of a gas concentration spike event, It is the first The maximum concentration value in a gas concentration spike event. It is the baseline concentration before the gas concentration spike; The above steps amplify the impact of sudden increases on the index through squared differences, while incorporating a logarithmic function to control its growth trend and prevent extreme values ​​from amplifying the overall volatility assessment. Each peak event will be weighted through this amplitude factor, reflecting its contribution to the overall volatility index.

[0026] After identifying all peak events and calculating the peak amplitude factor, in order to comprehensively measure the severity of harmful gas concentration fluctuations within the detection window, the intensity and frequency of all events are considered together. A nonlinear weighted accumulation mechanism is introduced to generate a reference value for the appearance of instantaneous gas concentration peaks, which is used to characterize the activity and severity of concentration fluctuations in the current environment. The generation formula is as follows: In the formula, This is a reference value for the appearance of instantaneous gas concentration peaks. It is the total number of spike events within the detection window. It is used to suppress the cumulative error amplification of high-frequency weak fluctuations, while enhancing the dominant effect of large-amplitude spikes, thus avoiding misjudgment caused by relying solely on frequency.

[0027] All identified spike events within the detection window are nonlinearly fused, taking into account both their frequency of occurrence and fluctuation intensity, to construct an instantaneous gas concentration spike reference value that comprehensively reflects the degree of abnormal fluctuations in gas concentration. This step, through weighted superposition, effectively improves the sensitivity to high-frequency, drastic fluctuations while suppressing the interference of weak disturbances on the judgment results.

[0028] The larger the reference value for instantaneous gas concentration spikes generated by analyzing the number and magnitude of brief and dramatic increases in gas concentration within a unit of time under the detection window, the more frequent and dramatic the spike fluctuations in gas concentration during the current monitoring period. This indicates abnormal instability and high volatility of harmful gas concentrations in the environment, significantly deviating from the stable trend under normal operating conditions, reflecting a state of violent fluctuation in the current environment. Conversely, a smaller reference value indicates that spike events are rare or have small amplitudes, and the gas concentration changes are within an acceptable range, with stable and controllable fluctuations, indicating that the environmental gas concentration is in a state of normal fluctuation.

[0029] When an abnormally high concentration of harmful gases rapidly spreads from a monitoring point to surrounding locations, it clearly indicates that the concentration of harmful gases in the environment is experiencing drastic fluctuations. This is because, under normal circumstances, changes in gas concentration typically exhibit localized and slow-diffusion characteristics. If multiple monitoring points in different areas detect a sharp increase in concentration within a short period, it indicates that the gas is spreading rapidly in an unnatural and non-uniform manner, often caused by sudden leaks, explosive emissions, or equipment malfunctions. This drastic temporal-spatial concentration change disrupts the stable distribution pattern of environmental gases, demonstrating irregular and rapid concentration transmission, and is a typical dynamic sudden fluctuation phenomenon. Especially when the diffusion path exhibits radial or chain-like propagation, and the concentrations at multiple points simultaneously exceed the threshold, it further verifies that the fluctuation has high intensity, high frequency, and high expansiveness. This is a crucial indicator for judging whether the current environment is experiencing drastic gas concentration fluctuations, and also a high-risk signal that air quality monitoring systems must respond to and intervene in real time.

[0030] The specific steps for analyzing the speed and extent of the diffusion of abnormally high concentration trends from a certain monitoring point to surrounding points within a detection window to generate reference values ​​for the diffusion of abnormal gas concentration trends are as follows: First, select a monitoring point as the center, and simultaneously acquire data from the surrounding area of ​​the current monitoring point. Data from several neighboring sensors is collected, and the distance between each sensor and the center point is recorded. The spatial diffusion factor is calculated using the following expression: In the formula, It is the spatial diffusion factor, used to measure whether drastic fluctuations in gas concentration are spreading outwards. It is the first The gas concentration values ​​of the neighboring monitoring points indicate the concentration at the center point. Selected around Among the neighboring sensor nodes, the first Real-time gas concentration values ​​at each monitoring point It is the gas concentration at the central monitoring point. It is the number of neighboring monitoring points involved in the calculation. It is a non-linear adjustment index that adjusts the sensitivity to different degrees of gas diffusion. It is the spatial weight function, and its expression is as follows: In the formula, It is the natural base. It is the distance attenuation factor, reflecting the effect of proximity on the diffusion effect. It is the center point and the The actual physical distance between the monitoring points; The purpose of this step is to quantify the "energy" or "intensity" of the anomalous concentration spreading outward from the center, while fully reflecting the characteristics of spatial propagation.

[0031] Obtaining the spatial diffusion factor Then, further The parameters are converted into standardized parameters for identifying severe fluctuations. To enhance the response to different diffusion patterns, a multidimensional nonlinear enhanced compression model is used to generate reference values ​​for the diffusion of gas concentration anomalies. The generation formula is as follows: In the formula, This is a reference value for the diffusion of abnormal gas concentration trends. It is a nonlinear enhancement factor used to amplify the response under high-intensity diffusion; appropriately increasing it can enhance sensitivity to violent fluctuations. It is the weighting coefficient of the square root enhancement term, reflecting the response compensation to moderate-intensity diffusion, and preventing the reference value from being sensitive only to extreme values. This involves performing a power transformation on the entire expression to adjust the analytical form of the reference value for the diffusion of abnormal gas concentration trends, making it closer to the threshold response curve required by the monitoring system.

[0032] Overall, this formula will The response is decomposed into a combined response of two wave modes (strong diffusion + gentle diffusion), then the control range is compressed numerically, and finally, a power law is applied. It enables flexible mapping of output response pairs. It possesses a multi-stage nonlinear structure, allowing for hierarchical and continuous characterization of concentration diffusion events of varying intensities, making it ideal for real-time monitoring systems to accurately identify "violent fluctuation states."

[0033] The larger the reference value for the diffusion of abnormally high gas concentration trends generated after analyzing the speed and extent of the spread from a certain monitoring point to surrounding points within the detection window, the more significant the abnormal fluctuation in gas concentration indicates that it not only occurred at one monitoring point but also rapidly propagated to multiple points within a short period of time, accompanied by a significant concentration surge. This demonstrates high intensity, speed, and wide-ranging propagation characteristics, indicating the presence of a sudden and highly harmful drastic fluctuation event in the environment. Conversely, a lower reference value for the diffusion of abnormal gas concentration trends indicates that the gas concentration change is localized and slow, without forming a large-scale abnormal diffusion, and falls within the scope of normal environmental fluctuations.

[0034] The key parameters that have been analyzed and processed are constructed into feature vectors and input into a pre-trained machine learning model. The machine learning model is then used to conduct in-depth mining and modeling analysis of the multivariate correlation between the features, so as to achieve intelligent identification and assessment of the dynamic changes in the concentration of harmful gases in the current environment. The instantaneous gas concentration peak reference value and the abnormal gas concentration diffusion trend reference value are used as feature vectors and input into a pre-trained machine learning model. The machine learning model generates a hazardous gas concentration fluctuation coefficient. The hazardous gas concentration fluctuation coefficient is used to deeply mine and model the multivariate correlation between various features, so as to realize the intelligent identification and assessment of the dynamic change of hazardous gas concentration in the current environment.

[0035] "Pre-trained machine learning models" refer to models that have undergone systematic training, optimization, and validation based on a large amount of historical environmental data and known gas fluctuation events before being formally applied to air quality monitoring equipment. This enables the model to identify and judge complex gas fluctuation patterns. During the model training phase, developers collect air quality datasets covering various typical operating conditions (such as normal conditions, slight fluctuations, severe fluctuations, and sudden leaks), including concentration change curves of various harmful gases, environmental parameters (temperature, humidity, wind speed, etc.), event labels, etc., and extract representative indicators through feature engineering. Subsequently, this feature data is input into various machine learning algorithms, such as Support Vector Machines (SVM), Random Forest, Gradient Boosting Trees (GBDT), or Deep Neural Networks (DNN). The model continuously optimizes its parameters to improve prediction accuracy by learning the relationships between patterns and features in the data and the results. After training, the model is tested on an independent validation set to ensure its generalization ability and accuracy in unknown environments.

[0036] Once deployed in a real-world air quality monitoring system, a pre-trained machine learning model can receive real-time data input from sensors, such as reference values ​​for instantaneous gas concentration peaks and concentration diffusion trends. Through the model's internal computational logic, these inputs are mapped to a comprehensive output indicator—the hazardous gas concentration fluctuation coefficient. This coefficient is essentially a quantitative expression of the model's understanding of multivariate relationships, reflecting the intensity and complexity of gas concentration changes in the current environment. By analyzing the numerical trend of this coefficient, the system can intelligently determine whether the environment is stable, experiencing mild disturbances, or experiencing severe fluctuations, thereby triggering dynamic adjustment strategies, such as increasing sampling frequency or activating alarm mechanisms. Compared to traditional fixed threshold judgment methods, machine learning models possess stronger pattern recognition and adaptive capabilities, maintaining high accuracy and robustness even when facing nonlinearity, noise interference, and multivariate coupling changes, significantly improving the air quality monitoring system's ability to perceive and respond to complex conditions.

[0037] The machine learning model is not limited here, but it can achieve the goal of identifying reference values ​​for instantaneous gas concentration peaks. Reference values ​​for the diffusion of abnormal gas concentration trends A comprehensive analysis was conducted to generate the concentration fluctuation coefficient of harmful gases. Any machine learning model can be used. To achieve the technical solution of this invention, this invention provides a specific implementation method. The formula for generating the concentration fluctuation coefficient of harmful gases is as follows: In the formula, and Reference values ​​for the occurrence of instantaneous gas concentration peaks Reference values ​​for the diffusion of abnormal gas concentration trends The preset proportional coefficient, and and All are greater than 0.

[0038] Preset ratio coefficient and This refers to the coefficient of variation in the concentration of harmful gases generated. At that time, weighting parameters are used to control the degree of influence of each input indicator on the final result. Specifically, Used to adjust the reference value for the appearance of instantaneous gas concentration peaks. The strength of its role in overall volatility assessment Then adjust the reference value for the diffusion of abnormal trends in gas concentration. The weights of these two coefficients are predetermined during the model design phase based on expert experience, historical data analysis, or model training results, and are kept within a reasonable range greater than zero to ensure that the two indicators have a significant impact on the coefficients of variation of harmful gas concentrations. All of these contribute positively and effectively. The core function of the preset proportional coefficient is to reflect the relative importance of different gas fluctuation characteristics to the overall fluctuation coefficient, thereby enabling the model to have a certain degree of adjustment flexibility and adaptability, and to maintain accurate evaluation results under different operating conditions.

[0039] The larger the reference value for the instantaneous gas concentration peak generated after analyzing the number and magnitude of brief and drastic increases in gas concentration per unit time within the detection window, and the larger the reference value for the spread of abnormally high concentration trends from a certain monitoring point to surrounding points within the detection window, the larger the reference value for the spread of abnormally high concentration trends. This indicates that the hazardous gas concentration is in a state of drastic fluctuation in the current environment when the multivariate correlation between features is deeply mined and modeled by a pre-trained machine learning model. Conversely, it indicates that the hazardous gas concentration in the current environment is in a normal state.

[0040] The harmful gas concentration fluctuation coefficient generated by the pre-trained machine learning model through in-depth mining and modeling analysis of the multivariate correlations between various features is compared with a pre-set reference threshold for harmful gas concentration fluctuation coefficient. This process enables intelligent identification and assessment of the dynamic changes in harmful gas concentration in the current environment. The specific steps are as follows: If the concentration fluctuation coefficient of harmful gases is greater than the preset reference threshold for the concentration fluctuation coefficient of harmful gases, it is determined that the concentration of harmful gases in the current environment is in a state of drastic fluctuation; if the concentration fluctuation coefficient of harmful gases is less than or equal to the preset reference threshold for the concentration fluctuation coefficient of harmful gases, it is determined that the concentration of harmful gases in the current environment is in a normal state.

[0041] When the machine learning model determines that the harmful gases in the current environment are fluctuating drastically, the temporal resolution of the air quality monitoring equipment is dynamically adjusted based on the evaluation results to improve its sensitivity to instantaneous high-concentration leakage events. At the same time, a high-speed caching mechanism is activated to temporarily store the high-frequency collected data, thereby alleviating the pressure on the data storage system and providing a sufficient data processing window for subsequent data processing and analysis. When the machine learning model determines that the harmful gases in the current environment are fluctuating drastically, the system triggers a dynamic adjustment of the time-series resolution and caching mechanism based on this intelligent assessment result. This step aims to comprehensively improve the response speed, data acquisition accuracy, and system operational stability of air quality monitoring equipment during emergencies. Firstly, dynamically adjusting the time-series resolution means that the equipment increases the sampling frequency originally used for routine monitoring to a higher level, allowing it to collect more data points per unit time. This improves the sensitivity and detection capability for instantaneous high-concentration leaks, avoiding the loss of important abnormal signals due to excessively long sampling intervals, and ensuring that risk events can be detected and responded to in the first instance.

[0042] Secondly, the increased sampling frequency inevitably leads to a rapid increase in data volume within a short period. To avoid overloading the storage system, the system simultaneously activates a high-speed caching mechanism, temporarily caching high-frequency data in local high-speed storage areas (such as memory, circular buffers, etc.), effectively alleviating the pressure on the backend database or cloud interface under short-term high loads by "smoothing out peaks and filling valleys." Furthermore, the high-speed cache provides a sufficient processing window for subsequent data analysis, enabling the system to perform in-depth analysis, visualization, and trend modeling of the captured key data without affecting real-time monitoring tasks. This "assessment-control-caching" linkage mechanism constructs a highly adaptive, stable, and efficient monitoring system, particularly suitable for industrial scenarios with extremely high air safety requirements, such as chemical, metallurgical, and energy industries. This step not only enhances the system's risk perception capability under emergencies but also provides strong data support for subsequent accident tracing, cause analysis, and model optimization.

[0043] When the machine learning model determines that the levels of harmful gases in the current environment are fluctuating drastically, the temporal resolution of the air quality monitoring equipment is dynamically adjusted based on the evaluation results. At the same time, a high-speed caching mechanism is activated to temporarily store the frequently collected data. The specific steps are as follows: After determining that the harmful gases in the current environment are in a state of drastic fluctuation, the temporal resolution of the air quality monitoring equipment will be adaptively adjusted according to the magnitude of the fluctuation, thereby improving the equipment's response capability to instantaneous high concentration peaks. The calculation expression is as follows: In the formula, It is the time series resolution dynamically adjusted under fluctuating conditions. This is the device's default timing resolution. This is a reference threshold for the coefficient of variation of harmful gas concentration. It is the sensitivity adjustment coefficient, which determines the response intensity of the adjustment range. ; The above steps, through an exponential compression function, make it possible to achieve the desired result in... When the value exceeds the threshold, the sampling interval is rapidly shortened to achieve a high-frequency sampling response mechanism, thereby accurately capturing short-term high-concentration leakage events and significantly improving the system's real-time detection sensitivity and ability to resolve minute fluctuations.

[0044] After the device sampling frequency increases, a high-speed cache mode is automatically activated to temporarily store the large amount of air quality data streams generated at high frequencies in a circular buffer (or a high-performance memory buffer pool). This prevents data writing to storage from causing blockages and provides data assurance for subsequent processing. The cache window calculation expression is as follows: In the formula, It is the optimal window period for continuous cache writes, used to support continuous writes during peak data periods. It is the cache capacity adjustment factor. It is the number of concurrent write channels in the device cache structure, used to improve the parallelism of data processing.

[0045] The above steps, by establishing an adjustable caching window mechanism, effectively buffer the storage pressure caused by data surges and prevent data loss without affecting real-time monitoring tasks. Simultaneously, they provide complete, continuous, and reliable data support for subsequent analysis, behavior backtracking, and anomaly tracking of machine learning models, enhancing the system's stability and scalability.

[0046] This invention achieves high sensitivity, intelligent response, and system stability assurance for air quality monitoring equipment in the face of sudden high-concentration leaks of harmful gases. This method not only overcomes the limitations of traditional fixed-time-resolution monitoring mechanisms in capturing drastic fluctuations, but also achieves intelligent identification and accurate judgment of dynamic changes in ambient gases through deep integration of feature engineering and machine learning models. After determining that the gas is in a state of drastic fluctuation, the system can dynamically increase the sampling frequency and activate a high-speed caching mechanism, ensuring real-time capture of key concentration peaks while effectively alleviating storage pressure and improving the system's processing capacity under high-frequency data loads. The overall solution significantly enhances the monitoring system's response speed and judgment accuracy to sudden gas leaks, improves the safety protection level and operational reliability of industrial sites, and has good engineering practical value and promising prospects for promotion.

[0047] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0048] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

[0049] It should be noted that, in this document, the use of relational terms such as "first" and "second" is merely for distinguishing one entity or operation from another, and does not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0050] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0051] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0052] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0053] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0054] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0055] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0056] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A real-time data acquisition and analysis method in intelligent industrial manufacturing, characterized in that, Includes the following steps: Air quality monitoring equipment routinely detects the concentration of harmful gases in the environment according to a pre-configured time resolution and acquires air quality parameter information in real time; The acquired raw air quality parameter data are preprocessed, and the preprocessed data is then uniformly constructed into a structured dataset. Feature engineering techniques are applied to extract key indicators from the dataset that characterize the drastic fluctuations in the concentration of harmful gases in the current environment. Statistical analysis and trend modeling are then performed on the extracted key indicators to quantify the degree of fluctuation in the concentration of harmful gases. The key parameters that have been analyzed and processed are constructed into feature vectors and input into a pre-trained machine learning model. The machine learning model is then used to conduct in-depth mining and modeling analysis of the multivariate correlation between the features, so as to achieve intelligent identification and assessment of the dynamic changes in the concentration of harmful gases in the current environment. When the machine learning model determines that the harmful gases in the current environment are fluctuating drastically, the temporal resolution of the air quality monitoring equipment is dynamically adjusted based on the evaluation results to improve its sensitivity to instantaneous high-concentration leakage events. At the same time, a high-speed caching mechanism is activated to temporarily store the high-frequency collected data, thereby alleviating the pressure on the data storage system and providing a sufficient data processing window for subsequent data processing and analysis.

2. The real-time data acquisition and analysis method in intelligent industrial manufacturing according to claim 1, characterized in that, The acquired raw air quality parameter data is preprocessed, and the preprocessed data is then used to construct a structured dataset. The specific steps are as follows: First, noise reduction is performed to filter out outliers caused by sensor fluctuations and communication interference, and to remove high-frequency interference and invalid information from the data. Secondly, missing values ​​are filled in using interpolation methods, historical mean methods, and model prediction methods to ensure data integrity. Next, data format standardization was implemented to unify the timestamp format, measurement unit, and numerical precision of sensor outputs, ensuring structural consistency of multi-source data. Then, the parameters are normalized to bring different physical quantities to a uniform scale, which is beneficial for feature extraction and modeling analysis. Finally, following a unified field naming convention and time alignment strategy, the processed multi-parameter data is organized into a structured table, with each row corresponding to a time node and each column corresponding to an air quality parameter, ultimately forming a high-quality dataset for subsequent intelligent analysis.

3. The real-time data acquisition and analysis method in intelligent industrial manufacturing according to claim 1, characterized in that, Feature engineering techniques are applied to extract key indicators characterizing the drastic fluctuations in the concentration of harmful gases in the current environment from the dataset. The extracted indicators include the frequency and amplitude of brief and drastic increases in gas concentration per unit time, and the speed and extent of the spread of abnormally high concentration trends from a certain monitoring point to surrounding points. The frequency and amplitude of brief and drastic increases in gas concentration per unit time, and the speed and extent of the spread of abnormally high concentration trends from a certain monitoring point to surrounding points are analyzed under the detection window to generate reference values ​​for the appearance of instantaneous gas concentration peaks and reference values ​​for the spread of abnormal gas concentration trends. The degree of fluctuation of the current harmful gas is quantified by the reference values ​​for the appearance of instantaneous gas concentration peaks and the spread of abnormal gas concentration trends.

4. The real-time data acquisition and analysis method in intelligent industrial manufacturing according to claim 3, characterized in that, The specific steps for analyzing the frequency and magnitude of brief, dramatic increases in gas concentration within a unit of time to generate reference values ​​for instantaneous gas concentration peaks within a detection window are as follows: First, identify gas concentration spikes occurring per unit time, i.e., gas concentration changes that occur in a sudden increase-decline pattern within a very short period. For each identified spike, calculate its corresponding amplitude factor to quantify the intensity of the event. The calculation expression is as follows: In the formula, It is the peak amplitude factor, quantized first The intensity of a gas concentration spike event, It is the first The maximum concentration value in a gas concentration spike event. It is the baseline concentration before the gas concentration spike; After identifying all peak events and calculating the peak amplitude factor, in order to comprehensively measure the severity of harmful gas concentration fluctuations within the detection window, the intensity and frequency of all events are considered together. A nonlinear weighted accumulation mechanism is introduced to generate a reference value for the appearance of instantaneous gas concentration peaks, which is used to characterize the activity and severity of concentration fluctuations in the current environment. The generation formula is as follows: In the formula, This is a reference value for the appearance of instantaneous gas concentration peaks. It is the total number of spike events within the detection window. It is used to suppress the cumulative error amplification of high-frequency weak fluctuations, while enhancing the dominant effect of large-amplitude spikes.

5. The real-time data acquisition and analysis method in intelligent industrial manufacturing according to claim 3, characterized in that, The specific steps for analyzing the speed and extent of the diffusion of abnormally high concentration trends from a certain monitoring point to surrounding points within a detection window to generate reference values ​​for the diffusion of abnormal gas concentration trends are as follows: First, select a monitoring point as the center, and simultaneously acquire data from the surrounding area of ​​the current monitoring point. Data from several neighboring sensors is collected, and the distance between each sensor and the center point is recorded. The spatial diffusion factor is calculated using the following expression: In the formula, It is the spatial diffusion factor. It is the first Gas concentration values ​​at neighboring monitoring points, It is the gas concentration at the central monitoring point. It is the number of neighboring monitoring points involved in the calculation. It is a non-linear adjustment index. It is the spatial weight function, and its expression is as follows: In the formula, It is the natural base. It is the distance decay factor. It is the center point and the The actual physical distance between the monitoring points; Obtaining the spatial diffusion factor Then, further The parameters are converted into standardized parameters for identifying severe fluctuations. To enhance the response to different diffusion patterns, a multidimensional nonlinear enhanced compression model is used to generate reference values ​​for the diffusion of gas concentration anomalies. The generation formula is as follows: In the formula, This is a reference value for the diffusion of abnormal gas concentration trends. It is a nonlinear enhancement factor. These are the weighting coefficients of the square root enhancement term. It involves exponentiation of the entire expression.

6. The real-time data acquisition and analysis method in intelligent industrial manufacturing according to claim 3, characterized in that, The instantaneous gas concentration peak reference value and the abnormal gas concentration diffusion trend reference value are used as feature vectors and input into a pre-trained machine learning model. The machine learning model generates a hazardous gas concentration fluctuation coefficient. The hazardous gas concentration fluctuation coefficient is used to deeply mine and model the multivariate correlation between various features, so as to realize the intelligent identification and assessment of the dynamic change of hazardous gas concentration in the current environment.

7. The real-time data acquisition and analysis method in intelligent industrial manufacturing according to claim 6, characterized in that, The harmful gas concentration fluctuation coefficient generated by the pre-trained machine learning model through in-depth mining and modeling analysis of the multivariate correlations between various features is compared with a pre-set reference threshold for harmful gas concentration fluctuation coefficient. This process enables intelligent identification and assessment of the dynamic changes in harmful gas concentration in the current environment. The specific steps are as follows: If the concentration fluctuation coefficient of harmful gases is greater than the preset reference threshold for the concentration fluctuation coefficient of harmful gases, it is determined that the concentration of harmful gases in the current environment is in a state of drastic fluctuation; if the concentration fluctuation coefficient of harmful gases is less than or equal to the preset reference threshold for the concentration fluctuation coefficient of harmful gases, it is determined that the concentration of harmful gases in the current environment is in a normal state.

8. The real-time data acquisition and analysis method in intelligent industrial manufacturing according to claim 7, characterized in that, When the machine learning model determines that the levels of harmful gases in the current environment are fluctuating drastically, the temporal resolution of the air quality monitoring equipment is dynamically adjusted based on the evaluation results. At the same time, a high-speed caching mechanism is activated to temporarily store the frequently collected data. The specific steps are as follows: After determining that the harmful gases in the current environment are in a state of drastic fluctuation, the temporal resolution of the air quality monitoring equipment will be adaptively adjusted according to the magnitude of the fluctuation, thereby improving the equipment's response capability to instantaneous high concentration peaks. The calculation expression is as follows: In the formula, It is the time series resolution dynamically adjusted under fluctuating conditions. This is the device's default timing resolution. This is a reference threshold for the coefficient of variation of harmful gas concentration. It is the sensitivity adjustment coefficient; After the device sampling frequency increases, a high-speed cache mode is automatically activated to temporarily store the large amount of air quality data streams generated at high frequencies in a circular buffer. This prevents data writing to storage from causing blockages and provides data assurance for subsequent processing. The cache window calculation expression is as follows: In the formula, This is the optimal window period for continuous cache writes. It is the cache capacity adjustment factor. It is the number of concurrent write channels in the device cache structure, used to improve the parallelism of data processing.

Images