A dust concentration detection method and a detection system
By establishing a multivariate nonlinear correction model and decoupling algorithm, combined with a periodic calibration mechanism, the problem of humidity-agglomeration coupling interference was solved, achieving high accuracy and stability in dust concentration detection, adapting to complex environments, and meeting the high standards required in industrial and environmental protection fields.
Patent Information
- Application Number
- CN202610417158.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-04-01
- Publication Date
- 2026-07-14
Smart Images
Figure CN122388967A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of dust concentration detection technology, specifically to a method and system for detecting dust concentration. Background Technology
[0002] Dust concentration detection is a core component of industrial safety production, environmental protection, and occupational health protection, and is widely used in mining, chemical production, construction, and other fields. Currently, the mainstream detection method is light scattering, but the results are easily affected by ambient humidity and dust agglomeration. Changes in humidity alter the surface characteristics of dust particles, causing them to adhere and agglomerate, which directly interferes with the intensity of the scattered signal, leading to deviations in concentration detection. As the industry's requirements for the accuracy of detection data continue to increase, effectively avoiding humidity-agglomeration coupling interference has become a key demand for improving the reliability of dust concentration detection.
[0003] Existing dust concentration detection technologies have many shortcomings in addressing the aforementioned problems. Most technologies only consider humidity or agglomeration as a single factor, ignoring their synergistic coupling effect and failing to comprehensively capture the changing patterns of the scattered signal. Traditional calibration models often employ linear designs, making it difficult to adapt to the complex nonlinear relationships between humidity, agglomeration, and the scattered signal, resulting in limited calibration accuracy. Existing decoupling algorithms lack specificity, failing to effectively separate the coupling interference between humidity and agglomeration, and lack a systematic parameter training and verification process, leading to insufficient algorithm parameter optimization and further amplifying detection errors. In addition, existing systems generally lack a periodic calibration mechanism, and as usage time increases, the model is prone to drift, failing to adapt to environmental changes in a timely manner, resulting in poor long-term detection stability and making it difficult to meet the requirements for high-precision and continuously reliable detection.
[0004] In summary, existing dust concentration detection technologies suffer from low accuracy and poor stability due to their failure to effectively address core issues such as humidity-agglomeration coupling interference, insufficient model adaptability, lack of parameter optimization, and lack of regular calibration. These issues make them unsuitable for meeting the high standards required in industrial production and environmental protection. Therefore, developing a technical solution that can accurately extract the quantitative relationship between humidity, agglomeration, and scattering signals, achieve interference separation through a nonlinear correction model and a dedicated decoupling algorithm, and ensure long-term detection reliability through regular calibration mechanisms, has become a pressing technical challenge in the field of dust concentration detection. This solution possesses significant practical application value and industry implications. Summary of the Invention
[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a method and system for detecting dust concentration. By setting multiple sets of humidity conditions to collect data, a multivariate nonlinear correction model of environmental humidity, agglomeration degree parameters and scattering signal intensity is established. The dust concentration is calculated using decoupling and concentration inversion algorithms, and the algorithm parameters are trained and optimized. A periodic calibration mechanism is set up to collect new data periodically or under certain conditions to achieve iterative updates of the model.
[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In one aspect, a method for detecting dust concentration, the method comprising:
[0007] S100, Data Acquisition: Set multiple sets of different humidity conditions, introduce dust under each set of humidity conditions and control the dust concentration to be stable. After the dust agglomerates to a stable state, synchronously collect and record the corresponding environmental humidity, agglomeration degree parameters and scattering signal intensity data to form the original dataset.
[0008] S200. Establishing a correlation and calibration model: Receive the original dataset, organize and analyze it, and extract the quantitative relationship between environmental humidity, aggregation degree parameters and scattered signal intensity through multivariate correlation analysis technology; establish a multivariate nonlinear correction model with environmental humidity and aggregation degree parameters as independent variables and scattered signal intensity as dependent variable.
[0009] S300, Decoupling and Concentration Calculation: Based on the established multivariate nonlinear correction model, the reference scattering signal value is calculated using the humidity-agglomeration coupling effect decoupling algorithm; with the reference scattering signal value as input, the final dust concentration value is calculated and output using the concentration inversion algorithm; the two algorithms are executed sequentially, forming a signal decoupling and concentration correction link;
[0010] S400, Training and Validation: Based on the signal decoupling and concentration correction link, the parameters in the humidity-agglomeration coupling effect decoupling algorithm and the concentration inversion algorithm are trained and optimized; by dividing the training set and the validation set, the parameters are fine-tuned using the training set to reduce the calculation deviation, and the error is checked using the validation set until the output dust concentration value reaches the preset requirement.
[0011] S500, System Deployment and Calibration: Based on the trained and optimized algorithm parameters, a periodic calibration mechanism is set up to collect environmental humidity, aggregation degree parameters and scattering signal data according to a set cycle or trigger conditions as new sample data; and the multivariate nonlinear correction model is periodically iterated and updated based on the new sample data.
[0012] Further, the process of simultaneously collecting and recording corresponding environmental humidity, agglomeration degree parameters, and scattering signal intensity data to form the original dataset is as follows: In a closed detection environment, multiple sets of different humidity conditions are set through a humidity adjustment device to maintain a stable humidity environment; the humidity adjustment device includes a humidification unit, a dehumidification unit, and a real-time humidity monitoring and feedback unit, which is a device capable of adjusting and maintaining preset humidity conditions in a closed detection environment; dust is introduced into the closed detection environment in a stable humidity state, and the dust concentration in the environment is monitored in real time through a dust concentration monitoring device, adjusting the dust introduction and airflow state in the environment to keep the dust concentration stable; the dust concentration monitoring device is a device capable of monitoring the dust concentration in the closed detection environment in real time and outputting corresponding concentration monitoring data; the agglomeration state of dust in the environment is monitored, and the agglomeration morphology and distribution changes of dust particles are observed. When the agglomeration morphology and distribution of dust particles no longer change, the dust agglomeration reaches a stable state; after the dust agglomeration reaches a stable state, the humidity value, dust agglomeration degree parameters, and scattering signal intensity data of the environment are collected simultaneously. After the collection is completed, the data is marked and stored accordingly to form the original dataset.
[0013] Furthermore, the process of extracting the quantitative relationship between ambient humidity, aggregation degree, and scattered signal intensity using multivariate correlation analysis technology is as follows: The received raw dataset is cleaned, outlier data is removed, missing data is filled, and standardization is performed to obtain a preprocessed dataset; the ambient humidity, aggregation degree parameters, and scattered signal intensity data in the preprocessed dataset are analyzed to determine the correlation trend among the three; based on the correlation trend, the correlation patterns affecting the scattered signal intensity under the synergistic effect of ambient humidity and aggregation degree are explored; the explored correlation patterns are filtered to obtain correlation factors; based on the filtered correlation factors, the correspondence between ambient humidity, aggregation degree parameters, and scattered signal intensity is analyzed to form a quantitative relationship representation.
[0014] Furthermore, the multivariate nonlinear correction model includes a model input layer, a nonlinear mapping layer, and a model output layer. The model input layer receives two independent variables: an environmental humidity parameter and a clustering degree parameter, and uses them as input variables for model adaptation and transmission. The nonlinear mapping layer performs nonlinear transformation processing on the input environmental humidity parameter and clustering degree parameter to characterize the nonlinear correlation between them and the scattered signal intensity. The model output layer receives the processing result from the nonlinear mapping layer and outputs the dependent variable data corresponding to the scattered signal intensity.
[0015] Furthermore, the calculation formula for the humidity-agglomeration coupling effect decoupling algorithm is as follows: ,in, The reference scattering signal value; This is the data on the intensity of the scattered signal; This refers to the ambient humidity value. This is a parameter representing the degree of dust agglomeration. This is the influence coefficient of the quadratic term of humidity; This represents the influence coefficient of the quadratic term of the agglomeration; The humidity-aggregation interaction coupling coefficient; The baseline correction coefficient is used to process the collected scattering signal, removing the effects of ambient humidity alone, agglomeration degree alone, and their combined effect, to obtain the baseline scattering signal value after eliminating coupling interference, providing a reliable signal basis for the accurate calculation of dust concentration.
[0016] Furthermore, the calculation formula for the concentration inversion algorithm is as follows: ,in, This represents the final output dust concentration value. The scattering signal-concentration conversion coefficient; The reference scattering signal value; It is a non-linear correction exponent; The concentration baseline correction value is used as the input. The reference scattering signal after eliminating coupling interference is used as the input. Through nonlinear correlation transformation and baseline deviation compensation, the scattering signal is quantitatively mapped to the dust concentration, and the output can reflect the dust content of the detected environment.
[0017] Furthermore, the signal decoupling and concentration correction link includes a decoupling algorithm execution unit and a concentration inversion execution unit, which are connected in series to form a complete link. The decoupling algorithm execution unit receives the quantization relationship extracted by the multivariate nonlinear correction model and the collected environmental humidity, agglomeration degree parameters, and scattering signal intensity data, and outputs a reference scattering signal value after calculation. The reference scattering signal value serves as the input data for the concentration inversion execution unit, which performs calculations based on the input data and outputs the final dust concentration value.
[0018] Furthermore, the specific steps of dividing the dataset into training and validation sets, using the training set to fine-tune parameters to reduce calculation deviations, and using the validation set for error verification until the output dust concentration value meets the preset requirements are as follows: The original dataset is grouped to obtain independent training and validation sets; the training set data is input into the signal decoupling and concentration correction link; based on the link output results, the parameters of the humidity-agglomeration coupling effect decoupling algorithm and the concentration inversion algorithm are adjusted and optimized to reduce the link output deviation; the validation set data is input into the parameter-adjusted signal decoupling and concentration correction link, and error verification is performed to determine whether the dust concentration value output by the link meets the preset requirements; the preset requirements are whether the relative error between the dust concentration value and the actual value does not exceed 5%, and whether the absolute error does not exceed 0.1 mg / m³; when the output result does not meet the preset requirements, the training and verification steps are repeated until the dust concentration value output by the signal decoupling and concentration correction link meets the preset requirements.
[0019] Furthermore, the periodic calibration mechanism includes a trigger unit, a data acquisition unit, and a sample storage unit. The trigger unit monitors the operating status and determines whether a preset periodic node or preset trigger condition is met. When the periodic node or trigger condition is met, the data acquisition unit initiates the data acquisition process, collecting data corresponding to environmental humidity, aggregation degree parameters, and scattering signals. The sample storage unit classifies and organizes all collected data, archives and stores the organized data, forming new sample data. The preset periodic node is a calibration trigger node set at fixed time intervals, specifically a calibration periodic node is set every 72 hours. The preset trigger condition is a calibration trigger condition set according to the equipment operating status or environmental parameter fluctuations, specifically when the equipment runs continuously for 120 hours, the equipment is turned on and off a cumulative total of 50 times, or the environmental humidity fluctuates by more than 15% in a single instance.
[0020] On the other hand, a dust concentration detection system includes:
[0021] Data acquisition module: Used to simultaneously collect and store ambient humidity, dust agglomeration parameters and scattering signal intensity data under stable humidity and dust concentration conditions, forming a raw dataset;
[0022] Model building module: used to preprocess the original dataset, extract quantitative relationships through multivariate correlation analysis, and establish a multivariate nonlinear correction model with humidity and agglomeration degree as independent variables and scattering signal intensity as dependent variable;
[0023] Decoupling and Calculation Module: Includes a decoupling algorithm unit and a concentration inversion algorithm unit. It sequentially performs humidity-agglomeration coupling effect decoupling calculation to obtain the reference scattering signal value, and then outputs the final dust concentration value through the concentration inversion algorithm.
[0024] Training and validation module: used to divide the training set and validation set, optimize the decoupling and inversion algorithm parameters, and ensure that the concentration output meets the preset requirements through error verification;
[0025] Periodic calibration module: It is equipped with triggering, data acquisition and sample storage units, which collect new data according to preset cycles or conditions and iteratively update the multivariate nonlinear correction model.
[0026] Compared with existing technologies, this method and system for detecting dust concentration has the following advantages:
[0027] I. This invention simultaneously collects environmental humidity, dust agglomeration degree parameters, and scattering signal intensity data under different humidity conditions. Combined with multivariate correlation analysis technology, it deeply explores the inherent correlation among these three factors, constructing a multivariate nonlinear correction model with humidity and agglomeration degree as independent variables and scattering signal intensity as the dependent variable. Through targeted humidity-agglomeration coupling effect decoupling algorithms and concentration inversion algorithms, a complete signal decoupling and concentration correction link is built, achieving accurate separation and effective correction of coupling interference. This invention overcomes the limitations of traditional technologies that only consider a single influencing factor and rely on linear models. It can comprehensively capture the complex influence of the synergistic effect of humidity and agglomeration on the detection results, significantly improving the accuracy of dust concentration detection and ensuring that the detection results truly reflect the actual dust concentration level under different humidity environments and dust agglomeration states.
[0028] Second, this invention systematically optimizes the parameters of the humidity-agglomeration coupling effect decoupling algorithm and the concentration inversion algorithm by dividing them into independent training and validation sets. Through an iterative error verification process, it ensures a high degree of fit between the algorithm parameters and the actual detection scenario, reducing computational deviations from the underlying logic. A periodic calibration mechanism including triggering, data acquisition, and sample storage units is set up, enabling automatic acquisition of new sample data according to preset cycles or triggering conditions, achieving periodic iterative updates of the multivariate nonlinear correction model. This effectively solves the problems of insufficient parameter optimization and easy model drift in existing detection technologies, improves the long-term operational stability of the detection system, adapts to the impact of environmental changes and equipment wear and tear, and broadens the applicable scenarios of the technology.
[0029] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description
[0030] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.
[0031] Figure 1 This is a flowchart illustrating the steps of a method for detecting dust concentration.
[0032] Figure 2 This is a schematic diagram illustrating signal decoupling and dust concentration calculation data transmission in a dust concentration detection method.
[0033] Figure 3 This is a flowchart of a dust concentration detection system. Detailed Implementation
[0034] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.
[0035] Example 1:
[0036] In a closed testing environment, multiple sets of different humidity conditions are set and maintained stably using a humidity control device. These humidity conditions comprehensively cover the humidity range that wood dust may encounter in various stages of enterprise production, storage, and transportation, with a gradient distribution from low to high humidity to ensure that the impact of different humidity environments on the characteristics of wood dust can be captured. Wood dust is introduced into this closed environment, and changes in the concentration of wood dust in the environment are monitored in real time using a high-precision dust concentration monitoring device. At the same time, the valve opening of the wood dust inlet pipe is dynamically adjusted to control the inlet volume, and the operating power and wind direction of the airflow circulation equipment in the environment are adjusted to keep the wood dust concentration stable within the set range, avoiding abnormal agglomeration caused by concentration fluctuations and ensuring the validity of subsequent data collection. The agglomeration state of wood dust is continuously monitored using high-definition imaging equipment and particle analysis sensors. The aggregation morphology, particle size, and distribution uniformity of dust particles are carefully observed. When these indicators no longer change significantly for a continuous period of time, the wood dust agglomeration is determined to have reached a stable state. At this point, the humidity level, wood dust agglomeration degree, and scattering signal intensity data of the current environment are collected using synchronous data acquisition equipment. The acquisition process strictly adheres to the principle of synchronization to ensure that each set of data corresponds completely at each time point. After the acquisition is completed, the data are labeled according to the correlation rules of humidity-agglomeration degree-scattering signal and stored in a dedicated database using encrypted storage, forming a complete and clearly correlated original dataset. This provides comprehensive and reliable basic data support for subsequent model construction and algorithm calculation.
[0037] The process begins by receiving the raw dataset and initiating a data preprocessing workflow. Abnormal data caused by equipment malfunctions or external interference is removed, and missing data from the data acquisition process is filled in. Standardization converts data on environmental humidity, wood dust agglomeration parameters, and scattering signal intensity of different magnitudes and units into values under a unified standard, resulting in a preprocessed dataset. This effectively improves data quality and eliminates the impact of data bias on subsequent analysis. In-depth analysis is then performed on the three core data points in the preprocessed dataset. By plotting scatter plots and calculating correlation coefficients, the correlation trends between environmental humidity and scattering signal intensity, the wood dust agglomeration parameter and scattering signal intensity, and environmental humidity and wood dust agglomeration parameter are determined, clarifying the general direction of their interaction. Based on these correlation trends, multivariate correlation analysis is used to explore the synergistic effect of environmental humidity and wood dust agglomeration on scattering signal intensity. The analysis focuses on the weight of agglomeration changes on scattering signal intensity within different humidity ranges, as well as the fluctuation patterns of scattering signal intensity caused by humidity changes at the same agglomeration level. The discovered correlation patterns were screened, eliminating secondary and incidental factors and retaining the core correlation factors that significantly affect the intensity of the scattered signal. Based on the screened correlation factors, a one-to-one correspondence between environmental humidity, wood dust agglomeration degree parameters, and scattered signal intensity was established through mathematical modeling, forming a precise quantitative representation that clearly shows the degree and mechanism of each parameter's influence on the scattered signal intensity. Subsequently, a multivariate nonlinear correction model was established, consisting of a model input layer, a nonlinear mapping layer, and a model output layer. The model input layer receives two independent variables, environmental humidity and wood dust agglomeration degree parameters, and converts them into a model-recognizable format through interface adaptation technology before transmission, ensuring that the parameters enter the subsequent processing stages without loss or delay. The nonlinear mapping layer processes the two input parameters, accurately representing the complex nonlinear correlation characteristics between them and the scattered signal intensity, and can capture the nonlinear fluctuations in scattered signal intensity caused by small changes in the parameters. The model output layer receives the processing results from the nonlinear mapping layer, and after data integration and format conversion, outputs the dependent variable data corresponding to the scattered signal intensity, providing direct and accurate model support for subsequent decoupling operations.
[0038] Based on the established multivariate nonlinear correction model, the humidity-agglomeration coupling effect decoupling algorithm is used to process the collected scattering signal intensity data. The formula is as follows: ,in, The reference scattering signal value; This is the data on the intensity of the scattered signal; This refers to the ambient humidity value. This is a parameter representing the degree of dust agglomeration. This is the influence coefficient of the quadratic term of humidity; This represents the influence coefficient of the quadratic term of the agglomeration; The humidity-aggregation interaction coupling coefficient; This serves as the base correction coefficient. The algorithm addresses the coupling interference characteristics of environmental humidity and the degree of wood dust agglomeration on the scattering signal. Through separation techniques, it removes the superimposed influence of these two factors, restoring a baseline scattering signal value that accurately reflects the actual concentration of wood dust, effectively eliminating detection errors caused by environmental humidity and agglomeration effects. After obtaining the baseline scattering signal value, it is used as input data to the concentration inversion algorithm, as shown in the formula: ,in, This represents the final output dust concentration value. The scattering signal-concentration conversion coefficient; The reference scattering signal value; It is a non-linear correction exponent; This algorithm, based on the intrinsic correlation between the scattered signal and dust concentration, transforms the abstract baseline scattered signal value into an intuitive and directly applicable wood dust concentration value through signal conversion and numerical calculation. The humidity-agglomeration coupling effect decoupling algorithm and the concentration inversion algorithm are executed sequentially, forming a complete signal decoupling and concentration correction link. In this link, the decoupling algorithm execution unit first receives the quantization relationship extracted from the multivariate nonlinear correction model, along with collected environmental humidity, wood dust agglomeration parameters, and scattered signal intensity data. After calculation, it outputs the baseline scattered signal value. This baseline scattered signal value is then transmitted to the concentration inversion execution unit, which performs calculations according to preset operational logic and parameters, ultimately outputting an accurate wood dust concentration value. This ensures that the conversion process from the original signal to the final concentration value is logically coherent and computationally accurate, minimizing the impact of various interference factors on the detection results.
[0039] The original dataset is randomly divided into training and validation sets according to a preset ratio to ensure consistency in data distribution and feature representation, avoiding distortion of training or validation results due to uneven data distribution. The training set data is then fully input into the signal decoupling and concentration correction link. The link performs decoupling operations and concentration inversion according to a predetermined process, outputting the corresponding wood dust concentration calculation result. The calculation result is compared with the actual concentration value in the training set, and the coefficients in the humidity-agglomeration coupling effect decoupling algorithm and the conversion coefficients and correction exponents in the concentration inversion algorithm are fine-tuned and optimized. After each parameter adjustment, the training set data is re-inputted into the link for computation, continuously reducing the deviation between the link output result and the actual value until the deviation is within a reasonable range. After optimizing the parameters of the training set, the validation set data is input into the parameter-adjusted signal decoupling and concentration correction link, and a complete error verification process is executed. The computational performance and accuracy of the link are evaluated to determine whether the output wood dust concentration value meets the preset accuracy requirements and error standards. If the output does not meet the preset requirements, return to the training phase, readjust the parameters and train again. Then, use the validation set to verify again, and continue the loop until the wood dust concentration value output by the signal decoupling and concentration correction link meets the preset requirements. This ensures that the detection method has good generalization ability and stable detection accuracy, and can accurately output wood dust concentration data in practical applications.
[0040] Based on the trained and optimized algorithm parameters, the detection system is deployed and installed. A comprehensive periodic calibration mechanism is established, comprising three core modules: a trigger unit, a data acquisition unit, and a sample storage unit. These modules work collaboratively to ensure the orderly conduct of the calibration process. The trigger unit monitors the system's operating status in real time, including running time, number of tests, operating temperature of core components, voltage stability, and other indicators. It also monitors whether preset periodic nodes are met, such as weekly or monthly fixed cycles, or preset trigger conditions, such as test result deviations exceeding thresholds or after equipment maintenance. When a periodic node or trigger condition is met, the trigger unit immediately sends a start signal to the data acquisition unit. Upon receiving the signal, the data acquisition unit initiates the data acquisition process, collecting the current environmental humidity, wood dust agglomeration parameters, and scattering signal intensity data according to the same standards and procedures as the initial data acquisition. This ensures that the newly acquired data is consistent with the original dataset in format and dimensions, possessing comparability and usability. The sample storage unit receives the data transmitted by the data acquisition unit, classifies and organizes it, removes invalid and duplicate data, and stores it according to established archiving rules, forming new sample data. New sample data is periodically input into the multivariate nonlinear correction model, initiating an iterative update process. The model adjusts its internal parameters and correlation rules based on the effective information in the new sample data, optimizing the nonlinear mapping relationship. This allows the model to adapt to dynamic changes in environmental conditions, subtle degradation of equipment performance, and minor alterations in the characteristics of wood dust. Through periodic iteration and updates, the multivariate nonlinear correction model maintains good adaptability and computational accuracy, thereby ensuring the long-term stable operation of the entire detection system and continuously outputting accurate and reliable wood dust concentration detection results. Figure 1 As shown, this provides scientific and accurate data support for enterprises to formulate safety protection measures and avoid fire and explosion risks.
[0041] Example 2:
[0042] Within the enclosed industrial workshop testing area, multiple sets of different humidity conditions are set using an industrial-grade high-precision humidity control device. These conditions cover the humidity range that may occur in the workshop under different seasons, weather conditions, and ventilation conditions, with gradient settings from dry to high-humidity environments. After each humidity condition is set, the environment is maintained in a stable state for a period of time to ensure that humidity fluctuations are controlled within a minimal range. This provides a comprehensive and realistic environmental simulation for subsequent analysis of the interference of humidity on metal dust detection. Metal dust is introduced into the testing area with stable humidity, and the concentration of metal dust in the area is monitored in real time using an online dust concentration monitoring device. This device has high-frequency acquisition and real-time feedback functions, enabling it to quickly capture concentration changes. Based on the concentration data fed back by the device, the opening of the feed valve below the metal dust storage tank is adjusted to control the introduction speed. At the same time, the ventilation volume and airflow guidance device of the ventilation duct in the testing area are adjusted to ensure that the metal dust is evenly distributed in the testing area and the concentration is maintained within the preset stable range. This avoids abnormal agglomeration caused by excessively high or low concentrations or uneven distribution, ensuring that the collected agglomeration parameters can accurately reflect the inherent characteristics of metal dust under the corresponding humidity and concentration conditions. The aggregation morphology, particle size, and distribution density of metal dust particles are continuously observed using a laser particle size analyzer and image recognition system. A continuous monitoring period is set. When the aggregation morphology of the metal dust particles shows no significant change, the particle size distribution tends to stabilize, and the distribution density is uniform without significant fluctuations within this period, the metal dust aggregation is considered to have reached a stable state. At this point, a synchronous data acquisition system is activated to simultaneously collect environmental humidity values, metal dust aggregation degree parameters, and scattering signal intensity data of the detection area. The acquisition process is timestamped to ensure that each set of data is completely synchronized in time. After acquisition, all data are tagged according to the classification rules of detection batch-humidity level-aggregation degree level and properly stored in an industrial database using distributed storage technology. This forms a complete, closely correlated, and clearly traceable original dataset, providing a high-quality data foundation for subsequent model building, algorithm optimization, and system verification, ensuring the scientific rigor and reliability of the entire detection process.
[0043] The process begins by receiving the raw dataset and initiating a data preprocessing procedure. This process identifies and removes abnormal data caused by sensor malfunctions, electromagnetic interference, operational errors, etc., and fills in missing data due to signal interruptions during data acquisition. Standardization is then applied to convert environmental humidity, metal dust agglomeration parameters, and scattered signal intensity data into standardized data of uniform magnitude and units, eliminating magnitude differences and unit interference between different data dimensions. This preprocessed dataset significantly improves data usability and consistency, laying a solid foundation for subsequent analysis. In-depth analysis of the three core data points in the preprocessed dataset clarifies the correlation trends between environmental humidity and scattered signal intensity, metal dust agglomeration parameters and scattered signal intensity, and environmental humidity and metal dust agglomeration parameters, providing a clear direction for further exploration of deeper correlations. Based on these correlation trends, the study investigates the synergistic effect of environmental humidity and metal dust agglomeration on scattered signal intensity. The focus is on analyzing how humidity changes indirectly affect scattered signal intensity by influencing agglomeration, and the cumulative effect of humidity and agglomeration on scattered signal intensity when they act together. The discovered correlation patterns were screened, and random and secondary correlations were eliminated through significance testing, retaining the core correlation factors that significantly affect the intensity of the scattered signal. This focused on key influencing factors, improving the relevance and accuracy of the quantitative relationships. Based on the screened correlation factors, mathematical modeling tools were used to establish the quantitative correspondence between environmental humidity, metal dust agglomeration degree parameters, and scattered signal intensity, constructing mathematical expressions to form a precise quantitative relationship representation, clearly defining the influence weight and mechanism of each parameter on the intensity of the scattered signal. Next, a multivariate nonlinear correction model was constructed, consisting of a model input layer, a nonlinear mapping layer, and a model output layer. The model input layer is equipped with a dedicated interface module that receives the two independent variables, environmental humidity and metal dust agglomeration degree parameters. Through data format conversion, signal amplification, and noise filtering, the parameters are adapted and transmitted to the nonlinear mapping layer, ensuring distortion-free and delay-free parameter transmission. The nonlinear mapping layer uses a neural network algorithm to perform nonlinear transformation processing on the two input parameters. Through the collaborative operation of multiple layers of neurons, it accurately represents the complex nonlinear correlation characteristics between these two parameters and the intensity of the scattered signal, effectively capturing the nonlinear response of the scattered signal intensity caused by small changes in the parameters. The model output layer receives the processing results from the nonlinear mapping layer. After data integration and error correction, it outputs the dependent variable data corresponding to the intensity of the scattered signal, providing accurate and stable model support for subsequent decoupling operations and concentration calculations.
[0044] Based on the established multivariate nonlinear correction model, a humidity-agglomeration coupling effect decoupling algorithm is used to deeply process the collected scattering signal intensity data. Metal dust possesses unique physical characteristics; its agglomeration state is significantly affected by humidity, and the interference of humidity and agglomeration degree on the scattering signal exhibits a strong coupling effect. This decoupling algorithm addresses this characteristic by using techniques such as separation matrix operations and interference factor cancellation to effectively separate the superimposed interference of humidity and agglomeration degree on the scattering signal, stripping away irrelevant interference components, and restoring a benchmark scattering signal value that truly reflects the metal dust concentration. This fundamentally eliminates detection bias caused by environmental factors, improving the authenticity and reliability of the signal. After obtaining the benchmark scattering signal value, it is used as core input data to the concentration inversion algorithm. Based on the physical correlation between scattering signal intensity and metal dust concentration, this algorithm, through a series of operations including signal amplitude conversion, nonlinear correction, and baseline calibration, transforms the decoupled abstract benchmark scattering signal value into a direct and usable metal dust concentration value, meeting the practical needs of industrial workshops for quantitative monitoring of metal dust concentration. The humidity-agglomeration coupling effect decoupling algorithm and the concentration inversion algorithm are executed sequentially in series, forming a complete signal decoupling and concentration correction link. In this link, the decoupling algorithm execution unit first receives the quantization relationship extracted from the multivariate nonlinear correction model, along with collected environmental humidity, metal dust agglomeration parameters, and scattering signal intensity data. It then performs calculations according to the preset algorithm logic and outputs a reference scattering signal value. This reference scattering signal value is transmitted in real-time to the concentration inversion execution unit via a data transmission channel. The concentration inversion execution unit calls preset calculation parameters and logic to perform precise calculations on the input data, ultimately outputting an accurate metal dust concentration value, such as... Figure 2 As shown, the entire process forms a closed loop, with each link closely connected and data flowing in real time. This ensures that the entire process, from raw signal acquisition to final concentration value output, is precise and controllable, maximizing the accuracy and reliability of metal dust concentration detection and providing scientific and accurate data references for the safety management of industrial workshops.
[0045] The original dataset was scientifically grouped using stratified sampling to create independent training and validation sets. This ensured high consistency between the two datasets in key features such as humidity distribution, agglomeration distribution, and concentration distribution, preventing data sampling bias from distorting training or validation results and guaranteeing the objectivity and fairness of the training and validation process. The training set data was completely and systematically input into the signal decoupling and concentration correction link. The link performed operations according to a predetermined process of decoupling and concentration inversion, outputting the corresponding metal dust concentration calculation results. The calculation results were compared one by one with the actual metal dust concentration values labeled in the training set to analyze the magnitude, distribution characteristics, and causes of errors. Based on the error analysis results, parameters such as the influence coefficient of the quadratic term of humidity, the influence coefficient of the quadratic term of agglomeration, the humidity-agglomeration interaction coupling coefficient, and the basic correction coefficient in the humidity-agglomeration coupling effect decoupling algorithm, as well as parameters such as the scattering signal-concentration conversion coefficient, the nonlinear correction exponent, and the concentration baseline correction value in the concentration inversion algorithm, were finely adjusted and optimized. After each parameter adjustment, the training set data is re-input into the link for computation, and the error is analyzed again. Continuous iterative optimization is performed to reduce the deviation between the link's output and the true value until the error indicators reach the preset training standards. After optimizing the training set parameters, the validation set data is input into the signal decoupling and concentration correction link with adjusted parameters, initiating a comprehensive error verification process. Using the same error evaluation standards as the training phase, various error indicators between the validation set output and the true concentration value are calculated. Simultaneously, the detection performance of the link under different humidity levels, agglomeration degrees, and concentration ranges is analyzed to determine whether the metal dust concentration value output by the link meets the preset accuracy requirements and error standards of the industrial workshop. If the output result does not meet the preset requirements, the process returns to the training phase, the algorithm parameters are readjusted, and training is performed again. Then, the validation set is used for verification again. Through multiple iterations of training and verification, the algorithm parameters are continuously optimized, and the link performance is improved until the metal dust concentration value output by the signal decoupling and concentration correction link meets the preset requirements. This ensures that the detection method has good stability, accuracy, and generalization ability, can adapt to the complex and changing environmental conditions of the industrial workshop, and continuously outputs reliable detection results.
[0046] Based on the trained and optimized algorithm parameters, the detection system was installed, debugged, and deployed in the industrial workshop detection area to ensure coordinated operation of all system modules and smooth data transmission. A comprehensive periodic calibration mechanism was also established, comprising a trigger unit, a data acquisition unit, and a sample storage unit. Each unit performs its specific function and works collaboratively to ensure efficient and orderly calibration. The trigger unit monitors the system's operating status in real time, including key indicators such as core processor operating temperature, sensor stability, data transmission latency, and power supply voltage fluctuations. It also tracks system runtime and the number of tests, monitoring whether preset periodic nodes are reached, such as once every two weeks or once every quarter, or whether preset triggering conditions are met, such as multiple consecutive deviations in test results, replacement of core components after equipment maintenance, or significant changes in the workshop environment. When the trigger unit determines that a periodic node or triggering condition is met, it immediately sends a start command to the data acquisition unit. Upon receiving the instruction, the data acquisition unit quickly initiates the data acquisition process, strictly adhering to the initial data acquisition standards and specifications. It collects environmental humidity values, metal dust agglomeration parameters, and scattering signal intensity data within the current industrial workshop detection area, ensuring that the newly acquired data maintains consistency with the original dataset in terms of acquisition accuracy, data format, and parameter dimensions, possessing good compatibility and comparability. The sample storage unit receives the new data transmitted by the data acquisition unit, classifies, filters, and organizes it, removing invalid, duplicate, and abnormal data. It then encrypts and stores the data according to the archiving rules of calibration date-environmental conditions-data type, forming new sample data with a standardized structure and complete content. Periodically, the new sample data is batch-inputted into the multivariate nonlinear correction model, initiating the model iteration and update program. The model learns from the effective information in the new sample data, automatically adjusting its internal parameter configuration, association rules, and nonlinear mapping relationships, optimizing its computational logic and adaptability. This allows the model to adapt promptly to dynamic changes in industrial workshop environmental conditions, natural degradation of equipment performance, and subtle changes in metal dust characteristics. Through periodic iteration and updates, the multivariate nonlinear correction model maintains high computational accuracy and good environmental adaptability, thereby ensuring the long-term stable and reliable operation of the entire detection system. Figure 3 As shown, it continuously outputs accurate and real-time metal dust concentration monitoring data for industrial workshops, providing solid data support for workshops to formulate safety production plans, optimize the operating parameters of ventilation and dust removal systems, and prevent safety accidents such as metal dust explosions, thus helping enterprises achieve refined and intelligent safety production management.
[0047] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.
Claims
1. A method for detecting dust concentration, characterized in that, The method includes: S100, Data Acquisition: Set multiple sets of different humidity conditions, introduce dust under each set of humidity conditions and control the dust concentration to be stable. After the dust agglomerates to a stable state, synchronously collect and record the corresponding environmental humidity, agglomeration degree parameters and scattering signal intensity data to form the original dataset. S200. Establishing a correlation and calibration model: Receive the original dataset, organize and analyze it, and extract the quantitative relationship between environmental humidity, aggregation degree parameters and scattered signal intensity through multivariate correlation analysis technology; establish a multivariate nonlinear correction model with environmental humidity and aggregation degree parameters as independent variables and scattered signal intensity as dependent variable. S300, Decoupling and Concentration Calculation: Based on the established multivariate nonlinear correction model, the reference scattering signal value is calculated using the humidity-agglomeration coupling effect decoupling algorithm; with the reference scattering signal value as input, the final dust concentration value is calculated and output using the concentration inversion algorithm; the two algorithms are executed sequentially, forming a signal decoupling and concentration correction link; S400, Training and Validation: Based on the signal decoupling and concentration correction link, the parameters in the humidity-agglomeration coupling effect decoupling algorithm and the concentration inversion algorithm are trained and optimized; by dividing the training set and the validation set, the parameters are fine-tuned using the training set to reduce the calculation deviation, and the error is checked using the validation set until the output dust concentration value reaches the preset requirement. S500, System Deployment and Calibration: Based on the trained and optimized algorithm parameters, a periodic calibration mechanism is set up to collect environmental humidity, aggregation degree parameters and scattering signal data according to a set cycle or trigger conditions as new sample data; and the multivariate nonlinear correction model is periodically iterated and updated based on the new sample data.
2. The method for detecting dust concentration according to claim 1, characterized in that, In step S100, the process of synchronously collecting and recording corresponding environmental humidity, agglomeration degree parameters, and scattering signal intensity data to form the original dataset is as follows: In a closed detection environment, multiple sets of different humidity conditions are set through a humidity adjustment device to maintain a stable humidity environment; dust is introduced into the closed detection environment in a stable humidity state, and the dust concentration in the environment is monitored in real time through a dust concentration monitoring device. The dust introduction and airflow state in the environment are adjusted to keep the dust concentration stable; the agglomeration state of the dust in the environment is monitored, and the agglomeration morphology and distribution changes of dust particles are observed. When the agglomeration morphology and distribution of dust particles no longer change, the dust agglomeration reaches a stable state; after the dust agglomeration reaches a stable state, the humidity value, dust agglomeration degree parameters, and scattering signal intensity data of the environment are collected synchronously. After the collection is completed, the data is marked and stored accordingly to form the original dataset.
3. The method for detecting dust concentration according to claim 1, characterized in that, In step S200, the process of extracting the quantitative relationship between environmental humidity, aggregation degree parameters and scattering signal intensity through multivariate correlation analysis technology is as follows: the received raw dataset is cleaned, abnormal data is removed, missing data is filled in and standardized to obtain a preprocessed dataset; The environmental humidity, aggregation degree parameters, and scattering signal intensity data in the preprocessed dataset were analyzed to determine the correlation trend among the three. Based on the correlation trend, we explored the correlation patterns that affect the intensity of scattered signals under the synergistic effect of environmental humidity and aggregation degree; we screened the discovered correlation patterns to obtain correlation factors; based on the screened correlation factors, we sorted out the correspondence between environmental humidity, aggregation degree parameters and scattered signal intensity to form a quantitative relationship characterization.
4. The method for detecting dust concentration according to claim 1, characterized in that, In step S200, the multivariate nonlinear correction model includes a model input layer, a nonlinear mapping layer, and a model output layer; the model input layer is used to receive two independent variables, environmental humidity parameter and aggregation degree parameter, and use them as input variables of the model for adaptation and transmission; The nonlinear mapping layer is used to perform nonlinear transformation on the input environmental humidity parameter and aggregation degree parameter, and to characterize the nonlinear correlation between the two and the intensity of the scattered signal. The model output layer receives the processing results from the nonlinear mapping layer and outputs the dependent variable data corresponding to the intensity of the scattered signal.
5. The method for detecting dust concentration according to claim 1, characterized in that, In step S300, the calculation formula for the humidity-agglomeration coupling effect decoupling algorithm is as follows: ,in, The reference scattering signal value; This is the data on the intensity of the scattered signal; This refers to the ambient humidity value. This is a parameter representing the degree of dust agglomeration. This is the influence coefficient of the quadratic term of humidity; This represents the influence coefficient of the quadratic term of the agglomeration; The humidity-aggregation interaction coupling coefficient; The base correction factor.
6. The method for detecting dust concentration according to claim 1, characterized in that, In step S300, the calculation formula of the concentration inversion algorithm is as follows: ,in, This represents the final output dust concentration value. The scattering signal-concentration conversion coefficient; The reference scattering signal value; It is a non-linear correction exponent; This is the baseline correction value for concentration.
7. The method for detecting dust concentration according to claim 1, characterized in that, In step S300, the signal decoupling and concentration correction link includes a decoupling algorithm execution unit and a concentration inversion execution unit, which are connected in series to form a complete link. The decoupling algorithm execution unit receives the quantization relationship extracted by the multivariate nonlinear correction model and the collected environmental humidity, agglomeration degree parameters, and scattering signal intensity data, and outputs a reference scattering signal value after calculation. The reference scattering signal value is used as the input data of the concentration inversion execution unit, which performs calculations based on the input data and outputs the final dust concentration value.
8. The method for detecting dust concentration according to claim 7, characterized in that, In step S400, the specific steps of dividing the training set and validation set, using the training set to fine-tune parameters to reduce the calculation deviation, and using the validation set to perform error verification until the output dust concentration value reaches the preset requirement are as follows: the original dataset is grouped to obtain mutually independent training set and validation set; the input signal of the training set data is decoupled from the concentration correction link; and the parameters of the humidity-agglomeration coupling effect decoupling algorithm and the concentration inversion algorithm are adjusted and optimized according to the link output results to reduce the link output deviation. After adjusting the input parameters of the validation set data, the signal decoupling and concentration correction link is processed, and error verification is performed to determine whether the dust concentration value output by the link meets the preset requirements. If the output result does not meet the preset requirements, the training and verification steps are repeated until the dust concentration value output by the signal decoupling and concentration correction link meets the preset requirements.
9. The method for detecting dust concentration according to claim 1, characterized in that, In step S500, the periodic calibration mechanism includes a triggering unit, a data acquisition unit, and a sample storage unit; The triggering unit monitors the operating status and determines whether the preset period node or preset triggering condition is met. When the period node or triggering condition is met, the data acquisition unit starts the data acquisition process and acquires data corresponding to the ambient humidity, aggregation degree parameters and scattering signals. The sample storage unit classifies and organizes all the collected data, archives and stores the organized data, and forms new sample data.
10. A dust concentration detection system, applicable to the dust concentration detection method according to any one of claims 1-9, characterized in that, The system includes: Data acquisition module: Used to simultaneously collect and store ambient humidity, dust agglomeration parameters and scattering signal intensity data under stable humidity and dust concentration conditions, forming a raw dataset; Model building module: used to preprocess the original dataset, extract quantitative relationships through multivariate correlation analysis, and establish a multivariate nonlinear correction model with humidity and agglomeration degree as independent variables and scattering signal intensity as dependent variable; Decoupling and Calculation Module: Includes a decoupling algorithm unit and a concentration inversion algorithm unit. It sequentially performs humidity-agglomeration coupling effect decoupling calculation to obtain the reference scattering signal value, and then outputs the final dust concentration value through the concentration inversion algorithm. Training and validation module: used to divide the training set and validation set, optimize the decoupling and inversion algorithm parameters, and ensure that the concentration output meets the preset requirements through error verification; Periodic calibration module: It is equipped with triggering, data acquisition and sample storage units, which collect new data according to preset cycles or conditions and iteratively update the multivariate nonlinear correction model.