Intelligent analysis system and method for optical characteristics of cloud smoke particle swarm

Abstract

The invention relates to the technical field of cloud smoke particle analysis systems, in particular to a cloud smoke particle swarm optical property intelligent analysis system and method, and the system comprises a particle physical parameter analysis module and an atmosphere simulation module which are in data connection with the particle physical parameter analysis module; the physical parameter analysis module is in data connection with the atmosphere simulation module, and the photoelectric sensor is in data connection with the physical parameter analysis module; the data fusion module is in data connection with the photoelectric sensor and the physical parameter analysis module; the data analysis module is in data connection with the data fusion module and is used for receiving the optical characteristic report output by the data fusion module; based on a deep learning algorithm, obtaining an optical characteristic database of cloud smoke particles; visual data and optical characteristic results and result precision of the cloud smoke particles are output, and the fusion not only comprises comprehensive analysis of multi-band optical data, but also considers environmental parameters and historical data, so that more comprehensive and more accurate cloud smoke particle characteristic description is provided.

Description

Technical Field

[0001] The present invention relates to the technical field of cloud smoke particle analysis systems, in particular to an intelligent analysis system and method for optical characteristics of cloud smoke particle groups. Background Art

[0002] With growing awareness of environmental protection and the increasing need for air quality monitoring, optical characterization of cloud and smoke particle populations is playing an increasingly important role in atmospheric environmental research and pollution control. Traditional cloud and smoke particle analysis methods rely primarily on physical sampling and laboratory analysis, which is not only time-consuming and labor-intensive but also difficult to achieve real-time monitoring and large-scale coverage. In recent years, advances in optical remote sensing technology have opened up new possibilities for rapid, non-contact analysis of cloud and smoke particle characteristics.

[0003] Currently, the industry has developed a variety of cloud and smoke particle analysis systems based on optical principles. These systems typically use single-band laser scattering or absorption measurement techniques to infer the concentration and size distribution of particles by analyzing the scattering or absorption characteristics of cloud and smoke particles for light of specific wavelengths. However, this single-band analysis method has significant limitations. First, it cannot fully capture the complex optical behavior of cloud and smoke particles across different spectral ranges, leading to doubts about the accuracy and reliability of the analysis results. Second, this method has difficulty distinguishing different types of cloud and smoke particles, especially in complex atmospheric environments where multiple pollutants are mixed. Single-band analysis often fails to provide sufficient identification information.

[0004] On the other hand, some research institutions are attempting to enhance the identification of cloud and smoke particle characteristics using multi-band spectral analysis. While this approach can theoretically provide more information about particle characteristics, it faces significant challenges in practical application, both in data processing and interpretation. Traditional data analysis methods, such as linear regression or simple clustering algorithms, struggle to effectively process and utilize this high-dimensional and complex spectral data. Furthermore, existing multi-band analysis systems often lack real-time processing capabilities, failing to meet the stringent timeliness requirements of environmental monitoring.

[0005] A common problem with existing technologies is the lack of comprehensive consideration of the overall optical properties of cloud and smoke particle groups. Most systems focus solely on the optical properties of individual particles, ignoring the impact of particle group effects on optical properties. This simplified approach often leads to significant errors when analyzing high-concentration or complex smoke.

[0006] Furthermore, existing cloud and smoke particle analysis systems generally lack intelligence and adaptability. They typically rely on pre-set analysis models and parameters, making them incapable of adapting to complex and changing atmospheric conditions. In real-world monitoring, changes in environmental conditions (such as temperature, humidity, and wind speed) can significantly affect the optical properties of cloud and smoke particles, and existing systems struggle to adjust their analysis strategies in real time to accommodate these changes.

[0007] Given these challenges, there is an urgent need for intelligent systems capable of comprehensively, accurately, and in real time analyzing the optical properties of cloud and smoke particle swarms. The ideal system should possess multi-band optical data acquisition capabilities, be able to effectively process and fuse large amounts of complex spectral data, account for particle swarm effects, and intelligently adapt to varying environmental conditions. Furthermore, the system should possess robust data analysis and visualization capabilities, providing intuitive and reliable support for environmental monitoring and decision-making. Summary of the Invention

[0008] This paper addresses these issues with existing technologies and proposes an innovative intelligent analysis system and method for the optical properties of cloud and smoke particle swarms. By integrating multi-band optoelectronic sensing technology, advanced data fusion algorithms, deep learning analysis methods, and intelligent adaptive mechanisms, this system achieves comprehensive, accurate, and real-time analysis of the optical properties of cloud and smoke particle swarms.

[0009] The present invention proposes an intelligent analysis system for optical characteristics of cloud and smoke particle groups, comprising:

[0010] Particle physics parameter analysis module, used for:

[0011] Receive input smoke composition data;

[0012] Based on the atmospheric diffusion model, the particle size distribution and quantity parameters of cloud smoke particles are output;

[0013] The atmospheric simulation module is connected to the particle physics parameter analysis module and is used to:

[0014] Receiving the particle size distribution and quantity parameters sent by the particle physics parameter analysis module;

[0015] Based on the particle size distribution and quantity parameters, a Monte Carlo simulation is performed to output a distribution pattern and density parameters of cloud smoke particles;

[0016] The physical parameter analysis module is connected to the atmospheric simulation module data and is used to:

[0017] Based on the particle optical properties database, output the scattering coefficient, extinction coefficient, absorption coefficient and refractive index of cloud smoke particles;

[0018] The photoelectric sensor is connected to the physical parameter analysis module and is used to:

[0019] receiving the smoke model output by the atmosphere simulation module;

[0020] Conduct multi-band simulation experiments to obtain optical data of smoke in different bands;

[0021] The data fusion module is connected to the photoelectric sensor and the physical parameter analysis module, and is used to:

[0022] receiving multi-band optical data output by the photoelectric sensor;

[0023] receiving the scattering coefficient, extinction coefficient, absorption coefficient and refractive index output by the physical parameter analysis module;

[0024] performing data fusion analysis based on the multi-band optical data and the scattering coefficient, extinction coefficient, absorption coefficient, and refractive index, and outputting a report on the optical characteristics of cloud and smoke particles;

[0025] The data analysis module is connected to the data fusion module and is used to:

[0026] receiving the optical property report output by the data fusion module;

[0027] Based on deep learning algorithms, a database of optical properties of cloud and smoke particles is obtained;

[0028] Output visualization data and optical properties of cloud and smoke particles and result accuracy.

[0029] Preferably, the particle physics parameter analysis module includes:

[0030] Input unit, used to set smoke component data input parameters and receive smoke component data input from the outside;

[0031] A data processing unit, data-connected to the input unit, configured to pre-process the smoke component data and output a pre-processing result;

[0032] a data comparison unit, data-connected to the data processing unit, configured to receive the preprocessing result output by the data processing unit and compare it with a database of physical parameters of cloud and smoke particles;

[0033] A data output unit is data-connected to the data comparison unit and is used to output the scattering coefficient, extinction coefficient, absorption coefficient and refractive index of the cloud smoke particles based on the particle optical property database and in combination with the preprocessing results.

[0034] Preferably, the atmospheric simulation module comprises:

[0035] A model building unit for outputting the distribution pattern and density parameters of cloud smoke particles;

[0036] a data output unit, data-connected to the model building unit, for outputting an optical property report of cloud smoke particles based on a particle optical property database;

[0037] A data interface unit, data-connected to the data output unit, for receiving multi-band optical data sent by the photoelectric sensor;

[0038] A physical parameter analysis interface, connected to the data interface unit for receiving the scattering coefficient, extinction coefficient, absorption coefficient and refractive index of cloud and smoke particles output by the physical parameter analysis module;

[0039] A data processing unit is connected to the data interface unit and the physical parameter analysis interface data, and is used to process the results of the comparison of the optical data in the multi-band and the particle optical performance database, and compare them with the distribution pattern and density parameters of cloud and smoke particles, the optical characteristics report of cloud and smoke particles, and the results of the physical parameter comparison output by the physical parameter analysis module, and output the comparison results.

[0040] Preferably, the photoelectric sensor comprises:

[0041] A photoelectric sensor unit for receiving light and generating an electrical signal;

[0042] a signal processing unit, electrically connected to the photoelectric sensor unit, for receiving the electrical signal and processing, storing and outputting the signal;

[0043] The communication unit is data-connected to the signal processing unit, and is used to receive the signal processed by the signal processing unit and send it to the data fusion module.

[0044] Preferably, the data fusion module includes:

[0045] A preprocessing unit, configured to receive output data from the signal processing unit;

[0046] a preprocessing result unit, data-connected to the preprocessing unit, for obtaining an optical property database of cloud and smoke particles using a deep learning algorithm based on the optical property report of the cloud and smoke particles;

[0047] The data output unit is data-connected to the pre-processing result unit and is used to output visualization data and optical characteristic results and result accuracy of cloud smoke particles.

[0048] Preferably, the data analysis module includes:

[0049] A data input unit, configured to receive visualization data and extract optical characteristic results of cloud and smoke particles from a database of optical characteristics of cloud and smoke particles;

[0050] a data processing unit, data-connected to the data input unit, for comparing the visual data and outputting a display and interpretation of the optical characteristics of the cloud and smoke particles;

[0051] The data output unit is data-connected to the data processing unit and is used to output interactive and visual optical property results of cloud and smoke particle groups.

[0052] Preferably, the photoelectric sensor comprises:

[0053] Visible light sensor, used to collect light information in the visible light range, and collect light information of the particle group in the visible light range through an optical window;

[0054] Infrared sensor, used to collect light information within the infrared range, and collect light information of the particle group within the infrared range through the infrared window;

[0055] The ultraviolet sensor is used to collect light information within the ultraviolet range and collects light information within the ultraviolet range through the ultraviolet window.

[0056] Preferably, the system further comprises a communication module, wherein the communication module comprises:

[0057] A wireless communication module, used for data transmission using wireless communication protocols, supporting connection with user devices such as smartphones or computers;

[0058] Wired communication module, used to support local device connection, using USB interface;

[0059] The cloud service module is used to provide remote data transmission and storage support through cloud communication protocols, and supports connection with user devices such as smartphones or computers.

[0060] Preferably, the deep learning algorithms in the data analysis module include algorithms based on deep convolutional neural networks, deep recurrent neural networks and autoencoders.

[0061] The intelligent analysis method of optical characteristics of cloud and smoke particle groups based on the system includes the following steps:

[0062] S1. Physical parameter analysis: pre-process the input smoke component data and physical parameters, and output the scattering coefficient, extinction coefficient, absorption coefficient and refractive index of cloud smoke particles;

[0063] S2, particle optical properties database, input the scattering coefficient, extinction coefficient, absorption coefficient and refractive index of cloud smoke particles;

[0064] S3, photoelectric sensor experiment, receives the smoke model, conducts multi-band simulation experiments, and obtains optical data of smoke in different bands;

[0065] S4. Data comparison: comparing the optical data of smoke in different wavelengths obtained from the photoelectric sensor experiment with the particle optical properties database to obtain comparison results;

[0066] S5. Based on the comparison results obtained in S4 and the distribution pattern and density parameters of the cloud smoke particles, output an optical property report of the cloud smoke particles;

[0067] S6. Data analysis, output of visualization data and optical properties of cloud smoke particles and result accuracy.

[0068] Specifically, the beneficial effects of the present invention are mainly reflected in the following aspects:

[0069] The system of this invention achieves technological breakthroughs and innovations in multiple areas. First, by employing a multi-module collaborative architecture, the system implements intelligent processing throughout the entire process, from data acquisition, preprocessing, analysis, to visual output. This holistic design not only improves system efficiency but also significantly enhances the accuracy and reliability of analytical results.

[0070] Secondly, the system uses innovative multi-source data fusion technology to effectively integrate information from various sensors and data sources. This fusion not only includes comprehensive analysis of multi-band optical data but also considers environmental parameters and historical data, providing a more comprehensive and accurate description of cloud and smoke particle characteristics. This multi-dimensional data fusion significantly improves the system's adaptability to complex smoke environments and enhances analytical accuracy.

[0071] Furthermore, this invention incorporates advanced deep learning algorithms, achieving significant breakthroughs in processing high-dimensional spectral data. By employing techniques such as deep convolutional neural networks, recurrent neural networks, and autoencoders, the system effectively extracts and utilizes the complex features in cloud smoke particle spectral data, enabling precise identification and classification of different smoke types. This intelligent data processing approach not only improves analysis accuracy but also significantly enhances the system's adaptability and generalization capabilities.

[0072] Furthermore, the system of the present invention places particular emphasis on real-time performance and scalability. By employing efficient data processing algorithms and a distributed computing architecture, the system enables rapid processing and analysis of large-scale data, meeting the needs of real-time monitoring. Furthermore, the system's modular design and cloud service integration facilitate future functional expansion and performance improvements.

[0073] Finally, this invention also achieves innovation in data visualization and result presentation. By developing an intuitive and interactive visualization interface, the system can present complex analysis results to users in a way that is easy to understand and use, greatly improving data usability and decision support capabilities.

[0074] In summary, the intelligent analysis system and method for optical properties of cloud and smoke particle swarms presented in this paper have achieved significant progress in terms of technological innovation, analytical accuracy, real-time performance, and application flexibility. This system not only addresses numerous issues existing in existing technologies but also provides a novel technical solution for atmospheric environmental monitoring and air quality assessment, possessing broad application prospects and significant practical significance. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] Figure 1 It is the overall logic block diagram of the system of the present invention.

[0076] Figure 2 This is a logic block diagram of the particle physics parameter analysis module of the present invention.

[0077] Figure 3 This is a logic block diagram of the atmospheric simulation module of the present invention.

[0078] Figure 4 This is a logic block diagram of the photoelectric sensor of the present invention.

[0079] Figure 5 This is a logic block diagram of the data fusion module of the present invention.

[0080] Figure 6 This is a logic block diagram of the data analysis module of the present invention. DETAILED DESCRIPTION

[0081] See Figure 1-6 The present invention provides a system and method for intelligently analyzing the optical properties of cloud and smoke particle swarms. The system comprises multiple functional modules that work together to achieve comprehensive analysis of the optical properties of cloud and smoke particle swarms. The present invention will be described in detail below with reference to specific embodiments.

[0082] like Figure 1 As shown, the cloud and smoke particle swarm optical characteristics intelligent analysis system of the present invention includes a particle physics parameter analysis module 1, an atmospheric simulation module 2, a physical parameter analysis module 3, a photoelectric sensor 4, a data fusion module 5, and a data analysis module 6. These modules collaborate with each other through data connections, forming a complete analysis chain.

[0083] First, the particle physics parameter analysis module 1 is used to receive input smoke composition data and output the particle size distribution and number parameters of cloud smoke particles based on the atmospheric diffusion model. In a preferred embodiment of the present invention, this module uses a modified Gaussian diffusion model to simulate the diffusion process of cloud smoke particles. Specifically, the particle physics parameter analysis module 1 calculates the particle concentration distribution using the following formula:

[0084] ,

[0085] in, For the point The particle concentration at is the emission rate, is the average wind speed, is the emission source height, and These parameters can be adjusted according to the specific smoke environment, for example, in an industrial area, May vary in the range of 10-100g / s, while Probably between 10-100m.

[0086] Atmospheric simulation module 2 is data-connected to particle physics parameter analysis module 1, receiving the particle size distribution and population parameters sent by module 1. Based on these parameters, module 2 performs a Monte Carlo simulation, outputting the distribution pattern and density parameters of cloud and smoke particles. This invention utilizes an improved Monte Carlo method, using random sampling and statistical analysis to simulate the motion trajectories of large numbers of cloud and smoke particles. The simulation preferably takes into account factors such as Brownian motion, gravitational settling, and turbulent diffusion, ensuring that the simulation results more closely resemble actual conditions.

[0087] The present invention not only considers simple Brownian motion, but also adopts a comprehensive particle motion model that is specially designed for the complex conditions in the actual atmospheric environment.

[0088] Comprehensive calculation formula for particle displacement:

[0089]

[0090] in, For particles in The total displacement in the direction, Contributed to Brownian motion, is the diffusion coefficient, is the time step, is a random number that follows a standard normal distribution, Contributes to turbulence, is the turbulent pulsation velocity, Contributes to gravity, is the acceleration due to gravity, is the particle relaxation time, is the Kronecker function, Contributes to the interaction between particles. The particle relaxation time calculation formula is:

[0091] ,

[0092] in, is the particle density, is the particle diameter, is the dynamic viscosity of air, The Cunningham correction factor is used to account for the discontinuity effects of small particles. It comprehensively considers the combined effects of Brownian motion, turbulent diffusion, gravitational settling, and interparticle interactions in the actual atmospheric environment. It significantly improves the accuracy of particle motion simulations under real atmospheric conditions. It accurately simulates the differentiated atmospheric behavior of particles of varying size ranges. It is applicable to a wide range of particle sizes, from nanometers to micrometers, avoiding the significant errors of simple Brownian motion models in the actual atmospheric environment.

[0093] The physical parameter analysis module 3 is connected to the atmospheric simulation module 2 and outputs the scattering coefficient, extinction coefficient, absorption coefficient, and refractive index of cloud and smoke particles based on the particle optical properties database. These parameters are calculated using Mie scattering theory, which calculates the interaction between cloud and smoke particles of different sizes and compositions and their light. For example, for spherical particles, the scattering efficiency can be expressed as:

[0094] ,

[0095] in, is the effective scattering efficiency of the particle group, and are the minimum and maximum values ​​of the particle size distribution, is the number of components, The radius is Among the particles The volume fraction of the components, is the particle size distribution function, For the Components, radius Particles at wavelength The scattering efficiency under For the The complex refractive index of the components.

[0096] For non-spherical particles, the shape factor is introduced Corrected scattering efficiency:

[0097] ,

[0098] in, is the shape factor, which is corrected according to the degree to which the particle shape deviates from the spherical grinding. is the scattering efficiency of the spherical atom, calculated by standard Mie theory.

[0099] The photoelectric sensor 4 is data-connected to the physical parameter analysis module 3, receiving the smoke model output by the atmospheric simulation module 2 and conducting multi-band simulation experiments to obtain optical data of smoke at different wavelengths. The present invention utilizes sensors covering three wavelengths: visible light, near-infrared, and ultraviolet (UV) to obtain comprehensive optical property data. Preferably, the wavelength range of the visible light sensor is 400-700nm, the near-infrared sensor is 700-1400nm, and the UV sensor is 200-400nm. This multi-band design enables the capture of characteristic spectral information of different types of cloud and smoke particles.

[0100] Data fusion module 5 is data-connected to photoelectric sensor 4 and physical parameter analysis module 3, receiving the multi-band optical data from photoelectric sensor 4 and the scattering coefficient, extinction coefficient, absorption coefficient, and refractive index from physical parameter analysis module 3. Based on this data, module 5 performs data fusion analysis and outputs a report on the optical properties of cloud and smoke particles. This invention utilizes an improved Kalman filter algorithm for data fusion, whose state equation and observation equation can be expressed as:

[0101] ,

[0102] ,

[0103] in, is the state vector, which contains the optical characteristic parameters of cloud smoke particles; is the state transfer matrix; is the control input matrix; is the control vector; is the process noise; is the observation vector; is the observation matrix; is the observation noise. Through iterative calculation, the optimal estimated optical characteristic parameters can be obtained.

[0104] Finally, data analysis module 6 is connected to data fusion module 5 to receive the optical properties report output by data fusion module 5. Based on a deep learning algorithm, module 6 acquires a database of optical properties of cloud and smoke particles and outputs visualization data, as well as the optical properties and accuracy of the cloud and smoke particles. This invention utilizes an improved convolutional neural network (CNN) architecture, with the following structure:

[0105] 1. Input layer: receives multi-band optical data with a size of [batch_size, channels, height, width];

[0106] 2. Convolutional layer 1: 32 3x3 convolution kernels, stride 1, ReLU activation function;

[0107] 3. Max pooling layer 1: 2x2 pooling with a stride of 2;

[0108] 4. Convolutional layer 2: 64 3x3 convolution kernels, stride 1, ReLU activation function;

[0109] 5. Max pooling layer 2: 2x2 pooling with a stride of 2;

[0110] 6. Fully connected layer 1: 128 neurons, ReLU activation function;

[0111] 7. Dropout layer: dropout rate is 0.5;

[0112] 8. Fully connected layer 2: outputs optical characteristic parameters, such as scattering coefficient, extinction coefficient, etc.

[0113] The network is trained via the back-propagation algorithm using the mean squared error (MSE) as the loss function: ,

[0114] in, is the true value, is the predicted value, is the sample size.

[0115] Through the collaborative operation of these modules, the system of the present invention can comprehensively analyze the optical properties of cloud and smoke particle swarms, providing important support for environmental monitoring and air quality assessment. The system has the following advantages: First, multi-source data fusion improves the accuracy and reliability of analysis results; second, the application of deep learning algorithms enhances the system's adaptability to complex smoke environments; and finally, the collection and analysis of multi-band optical data provides a foundation for a comprehensive understanding of smoke characteristics.

[0116] In practical applications, the system can adjust parameters according to specific environments and needs. For example, when monitoring industrial areas, the characteristic band analysis of heavy metal particles can be added; in forest fire monitoring, the optical characteristics analysis of organic aerosols can be strengthened. This flexibility enables the present invention to adapt to a variety of complex atmospheric environment monitoring scenarios. Next, the present invention will explain in detail the specific structure and function of the particle physics parameter analysis module 1. Figure 2 As shown, the module includes an input unit 11, a data processing unit 12, a data comparison unit 13 and a data output unit 14.

[0117] Input unit 11 is used to set smoke composition data input parameters and receive external smoke composition data. In a preferred embodiment of the present invention, input unit 11 utilizes a graphical user interface (GUI) to allow operators to directly input or upload smoke composition data. This data may include, but is not limited to, environmental parameters such as particulate matter concentration, gas composition ratios, temperature, and humidity. For example, for PM2.5 monitoring, the input data may include concentration values ​​within the range of 0-500 μg / m³.

[0118] The data processing unit 12 is data-connected to the input unit 11 and is used to preprocess the smoke composition data and output the preprocessing results. The preprocessing process includes steps such as data cleaning, normalization, and feature extraction. Preferably, the present invention employs a median filter algorithm_page to remove outliers and uses the Min-Max normalization method to scale the data to the [0, 1] interval. Principal component analysis (PCA) is used for feature extraction, retaining principal components that explain 95% of the variance. These preprocessing steps can effectively improve the accuracy and efficiency of subsequent analysis.

[0119] Data comparison unit 13 is connected to data processing unit 12 and is configured to receive the preprocessing results output by data processing unit 12 and compare them with a database of physical parameters of cloud and smoke particles. The system of the present invention maintains a database containing a large number of known physical properties of cloud and smoke particles. Data comparison unit 13 uses a modified k-nearest neighbor (k-NN) algorithm for rapid matching. The core concept of the algorithm is as follows:

[0120] ,

[0121] NN) algorithm for fast matching. The core idea of ​​the algorithm is as follows: is the weighted Euclidean distance, and are the input samples and the samples in the database respectively, For the The weight of the feature, is the feature dimension. By adjusting the weight , which can highlight the influence of key features. In practical applications, the present invention preferably takes k=5, that is, selects the 5 most similar samples for subsequent analysis.

[0122] The data output unit 14 is connected to the data comparison unit 13 and is used to output the scattering coefficient, extinction coefficient, absorption coefficient, and refractive index of the cloud smoke particles based on the particle optical properties database and the preprocessing results. Here, the present invention uses a modified Lorenz-Mie theory to calculate these optical parameters. For example, for spherical particles, their extinction cross section can be expressed as:

[0123] ,

[0124] in, is the wave number, and is the Mie coefficient. In practical applications, for typical cloud and smoke particles, the extinction coefficient may be The results vary within a certain range, depending on the particle concentration and composition.

[0125] In practice, for typical cloud smoke particles, the extinction coefficient may vary in the range of 0.01-1 km⁻¹, depending on the particle concentration and composition.

[0126] The structure and function of the atmospheric simulation module 2 are further discussed. The module includes a model building unit 21 , a data output unit 22 , a data interface unit 23 , a physical parameter analysis interface 24 and a data processing unit 25 .

[0127] The model building unit 21 is responsible for outputting the distribution pattern and density parameters of cloud smoke particles. The present invention adopts an improved Lagrangian particle diffusion model, which takes into account factors such as atmospheric turbulence, gravitational sedimentation, and chemical reactions. The equation of motion of the particle can be expressed as:

[0128] ,

[0129] in, is the particle position, is the average wind speed, is the turbulent velocity fluctuation, is the sedimentation velocity, is the Kronecker function. By solving this equation, we can obtain the trajectories of a large number of particles and thus construct the three-dimensional distribution pattern of cloud smoke particles.

[0130] Data output unit 22 is connected to model building unit 21 and outputs a report on the optical properties of cloud smoke particles based on the particle optical properties database. This unit utilizes the multi-scale optical property integration algorithm unique to this invention, which can simultaneously consider the microscopic optical properties of individual particles and the macroscopic optical effects of the entire smoke. For example, for a polydisperse aerosol system, its volume scattering function can be expressed as:

[0131] ,

[0132] in, is the volume scattering function, is the particle size distribution function, is the scattering cross section of a single particle.

[0133] Data interface unit 23 is used to receive multi-band optical data transmitted by photoelectric sensor 4. The system of the present invention utilizes a high-speed data transmission interface to support the reception and processing of real-time data streams. Preferably, the data transmission rate can reach 1 Gbps, ensuring the rapid transmission of large amounts of multi-band optical data.

[0134] The physical parameter analysis interface 24 is used to receive the scattering coefficient, extinction coefficient, absorption coefficient, and refractive index of cloud and smoke particles output by the physical parameter analysis module 3. This interface uses a standardized data format to facilitate data exchange and integration between different modules.

[0135] Data processing unit 25, connected to data interface unit 23 and physical parameter analysis interface 24, processes the results of comparing multi-band optical data with the particle optical properties database, compares them with the cloud and smoke particle distribution pattern and density parameters, the cloud and smoke particle optical properties report, and the physical parameter comparison results output by physical parameter analysis module 3, and outputs the final comparison results. This unit utilizes the invention's unique multi-source data fusion algorithm, which effectively integrates data from different sources and improves the accuracy and reliability of the analysis results.

[0136] Finally, the present invention will describe in detail the structure and function of the photoelectric sensor 4. The sensor comprises a photoelectric sensor unit 41, a signal processing unit 42 and a communication unit 43.

[0137] The photosensor unit 41 is used to receive light and generate electrical signals. The present invention utilizes a highly sensitive photodiode array covering three wavelengths: visible light, near-infrared light, and ultraviolet light. For example, within the visible light range, the sensor's responsivity can reach 0.5 A / W, with a quantum efficiency exceeding 80%. This high-performance sensor ensures accurate capture of weak light signals.

[0138] The signal processing unit 42 is electrically connected to the photoelectric sensor unit 41 and is used to receive, process, store, and output electrical signals. This unit utilizes a high-speed analog-to-digital converter (ADC) and a field-programmable gate array (FPGA) for real-time signal processing. The ADC has a sampling rate of up to 100 MHz and a resolution of 16 bits, ensuring high-fidelity signal acquisition. The FPGA implements complex signal filtering and feature extraction algorithms, such as wavelet and Fourier transforms, to extract key features of the optical signal.

[0139] Communication unit 43 is data-connected to signal processing unit 42, receiving the processed signals from signal processing unit 42 and transmitting them to data fusion module 5. The present invention utilizes a high-speed serial communication interface, such as USB 3.0 or Ethernet, to ensure real-time transmission of large amounts of optical data. Preferably, communication unit 43 also supports wireless transmission, such as Wi-Fi or 4G / 5G, facilitating flexible deployment in complex environments.

[0140] Through the detailed module description above, the cloud smoke particle group optical characteristics intelligent analysis system of the present invention has demonstrated its powerful data processing capabilities and flexible system architecture. Each module has been carefully designed and optimized, and can efficiently process complex smoke optical data, laying a solid foundation for subsequent intelligent analysis. Figure 3 As shown, the module includes a preprocessing unit 51, a preprocessing result unit 52 and a data output unit 53.

[0141] The pre-processing unit 51 is used to receive the data output by the signal processing unit 42. In a preferred embodiment of the present invention, the pre-processing unit 51 uses an adaptive filtering algorithm to remove noise and interference in the data. Specifically, the present invention uses an improved Wiener filter, whose transfer function can be expressed as:

[0142] ,

[0143] in, is the power spectral density of the signal, The present invention innovatively solves the problem of noise power spectrum density. The dynamic estimation problem of , adopts a two-stage estimation method:

[0144] 1. Noise estimation of signal inactive segment:

[0145] ,

[0146] in, is the initial noise power spectral density estimate, is the number of frames of the signal inactive segment, For the Spectrum of the frame.

[0147] 2. Minimum Statistics Tracking Method Dynamically Updates Noise Estimation:

[0148] ,

[0149] in, For time The noise power spectral density estimate at and To adjust the parameters, usually set to and , For time The signal spectrum at .

[0150] Signal power spectral density estimation:

[0151] ,

[0152] in, It is the lower limit coefficient, usually set to 0.1 to ensure the stability of the filter.

[0153] By combining the initial estimation of the signal's inactive segments with minimum statistic tracking, an accurate dynamic estimation of the noise power spectral density is achieved. The filter can adaptively adjust its characteristics and automatically adjust the filtering parameters as the ambient noise changes, effectively handling non-stationary noise environments and adapting to changing noise conditions in different monitoring scenarios. The system can effectively remove up to 20dB of background noise while retaining the key features of the signal, avoiding information loss caused by over-filtering and ensuring high-quality data input for subsequent analysis.

[0154] Preprocessing result unit 52 is data-connected to preprocessing unit 51 and is used to apply a deep learning algorithm to obtain a database of cloud and smoke particle optical properties based on the cloud and smoke particle optical property reports. The present invention utilizes an innovative deep learning architecture in this regard, combining a long short-term memory (LSTM) network and an attention mechanism. This structure effectively captures the time series characteristics of cloud and smoke particle optical properties while focusing on key spectral information. The core structure of the network can be represented as follows:

[0155] LSTM ,

[0156] Attention ,

[0157] Softmax ,

[0158] in, is the input optical data, is the hidden state of LSTM, is the output of the attention mechanism, The final prediction result is . During the training process, the present invention adopts a dynamic learning rate adjustment strategy, with the initial learning rate set to 0.001 and decayed by 10% every 50 epochs. This strategy effectively prevents the model from falling into local optimality and improves the training stability and convergence speed.

[0159] The data output unit 53 is data-connected to the preprocessing result unit 52 and is used to output visualization data as well as the optical properties and accuracy of cloud and smoke particles. The present invention utilizes advanced data visualization techniques in this process, including but not limited to three-dimensional scatter plots, heat maps, and parallel coordinate plots. For example, for the scattering characteristics of cloud and smoke particles, the system can generate a scattering phase diagram in a polar coordinate system, visually displaying the light intensity distribution at different scattering angles. Furthermore, the present invention incorporates uncertainty quantification technology, using the Monte Carlo dropout method to estimate the confidence interval of the model prediction, providing an accuracy assessment for each prediction result.

[0160] Next, the present invention will describe in detail the structure and function of the data analysis module 6. Figure 4 As shown, the module includes a data input unit 61, a data processing unit 62 and a data output unit 63.

[0161] Data input unit 61 is used to receive visualization data and extract optical properties of cloud and smoke particles from a database of their optical properties. In one embodiment of the present invention, data input unit 61 utilizes a distributed storage architecture to support highly concurrent data read operations. This architecture can significantly improve the efficiency of large-scale data processing, particularly when processing massive amounts of multidimensional optical data.

[0162] Data processing unit 62 is connected to data input unit 61 and is used to compare the visual data and output the results of the optical properties of cloud smoke particles for display and interpretation. The present invention introduces an innovative multimodal data fusion algorithm in this step, capable of simultaneously processing and analyzing heterogeneous data from different sensors. Specifically, this system uses a data fusion method based on a graph neural network (GNN), the core concept of which can be expressed as follows:

[0163] ,

[0164] in, Representation node In the The feature representation of the layer, For nodes The neighbor set of and are learnable parameters. This approach allows the system to effectively capture the complex relationships between different data sources, providing a more comprehensive and accurate analysis of the optical properties of cloud and smoke particles.

[0165] Data output unit 63 is data-connected to data processing unit 62 and is used to output interactive, visual results of the optical properties of cloud and smoke particle swarms. The present invention utilizes advanced web technologies, such as WebGL and D3.js, to construct an interactive data visualization platform. This platform allows users to adjust various parameters, such as particle size distribution and complex refractive index, in real time and immediately visualize the impact of these adjustments on optical properties. Furthermore, the system supports dynamic projection of multidimensional data, allowing users to freely select characteristic dimensions of interest for visualization, enabling in-depth exploration of the optical properties of cloud and smoke particles.

[0166] Finally, the present invention describes in detail the specific structure of the photoelectric sensor 4. Figure 5 As shown, the sensor includes a visible light sensor 44 , an infrared sensor 45 and an ultraviolet sensor 46 .

[0167] Visible light sensor 44 is used to collect light information within the visible light range, collecting information about the particle swarm's response to light within the visible light range through an optical window. In a preferred embodiment of the present invention, visible light sensor 44 utilizes a highly sensitive CMOS image sensor with a pixel resolution of 2048x2048 and a dynamic range exceeding 90dB. This high-performance sensor can capture minute variations in light intensity, providing reliable data for accurately analyzing the scattering characteristics of cloud and smoke particles.

[0168] The infrared sensor 45 is used to collect light information in the infrared range, and collects light information of the particle group in the infrared range through the infrared window. The present invention uses a cooled InGaAs detector with an operating wavelength range of 0.9-1.7μm and a detection rate (D*) exceeding 1x10 12 cm√Hz / W. This highly sensitive infrared sensor can effectively capture the absorption and scattering characteristics of cloud smoke particles in the infrared band, providing important data support for comprehensive analysis of smoke composition.

[0169] UV sensor 46 is used to collect light information within the ultraviolet range through a UV window. The present invention utilizes a back-illuminated, UV-enhanced CCD sensor with a quantum efficiency exceeding 50% in the 200-350nm band. This highly efficient UV sensor can accurately capture the characteristic absorption peaks of organic compounds and certain metal ions in smoke, providing critical data for qualitative and quantitative analysis of smoke composition.

[0170] By combining this multi-band photoelectric sensor, the system of the present invention can comprehensively collect the optical responses of cloud and smoke particles across different spectral ranges, providing a rich and comprehensive data foundation for subsequent intelligent analysis. This design not only improves the system's analytical accuracy but also enhances its ability to distinguish different types of smoke, making the present invention significantly advantageous in complex atmospheric environment monitoring.

[0171] The structure and function of the communication module of the present invention will be further explored. Figure 6 As shown, the module includes a wireless communication module 71, a wired communication module 72 and a cloud service module 73. This diversified communication architecture ensures stable operation and data transmission of the system in various complex environments.

[0172] Wireless communication module 71 uses wireless communication protocols for data transmission, supporting connections with user devices such as smartphones and computers. In a preferred embodiment of the present invention, wireless communication module 71 utilizes the latest Wi-Fi 6 (802.11ax) technology, achieving a theoretical transmission rate of up to 9.6 Gbps. Furthermore, the module supports Bluetooth 5.0 technology for short-range, low-power communication. This dual-mode wireless communication design significantly enhances the system's flexibility and adaptability. For example, in field smoke monitoring scenarios, operators can receive and view monitoring data in real time via their smartphones, eliminating the need for complex wired connections.

[0173] Wired communication module 72 supports local device connectivity and utilizes a USB interface. The present invention utilizes the USB 3.2 Gen 2 x 2 standard, offering data transfer rates of up to 20 Gbps. This high-speed wired connection is particularly suitable for rapidly transferring and backing up large amounts of raw data. For example, when performing high-resolution spectral analysis, the system may need to transfer large amounts of raw spectral data, making the high bandwidth advantages of a wired connection particularly important.

[0174] Cloud service module 73 provides remote data transmission and storage support via cloud communication protocols, enabling connectivity with user devices such as smartphones and computers. This module utilizes a cloud computing platform based on a microservices architecture, supporting elastic and dynamic allocation of computing and storage resources. Specifically, cloud service module 73 utilizes containerization technologies (such as Docker and Kubernetes) to manage and deploy individual microservices. This architecture not only improves system scalability and reliability but also enables efficient utilization of computing resources.

[0175] For example, when processing data from a large-scale smoke monitoring network, the system can automatically adjust the allocation of computing resources based on changes in data traffic. During peak processing times, the cloud platform can rapidly expand computing nodes to ensure timely completion of analysis tasks; during low-volume periods, excess resources can be released to reduce operating costs. Furthermore, Cloud Service Module 73 implements multiple data replicas and geographic disaster recovery, significantly improving data security and reliability.

[0176] Next, the present invention will detail the deep learning algorithms employed in Data Analysis Module 6. As previously mentioned, this module utilizes algorithms based on deep convolutional neural networks, deep recurrent neural networks, and autoencoders. These advanced deep learning techniques provide powerful support for analyzing the optical properties of cloud and smoke particle swarms.

[0177] Deep convolutional neural network (DCNN) is mainly used in this invention to process and analyze multidimensional spectral data. The core operation of DCNN can be expressed as:

[0178] ,

[0179] in, Indicates the Tier The output of the position, is the convolution kernel weight, is the input of the previous layer, is the bias term, is the activation function.

[0180] In the embodiments of the present invention, a residual network (ResNet) structure is used to effectively alleviate the gradient vanishing problem of deep networks through skip connections, thereby improving the training efficiency and performance of the model. The deep recurrent neural network (DRNN) in this system is mainly used to analyze the time series data of the optical properties of cloud smoke particles. The core calculation of the DRNN can be expressed as:

[0181] ,

[0182] ,

[0183] in, is hidden state, For input, For output, and are the weight matrix and bias vector respectively, and The present invention adopts a long short-term memory (LSTM) network, whose special gating mechanism can effectively capture long-term dependencies and is particularly suitable for analyzing the long-term variation trend of the optical properties of cloud and smoke particles.

[0184] The autoencoder is mainly used for data dimension reduction and feature extraction in this invention. Its encoding process can be expressed as:

[0185] ,

[0186] The decoding process can be expressed as:

[0187] ,

[0188] in, For input data, is the encoded representation, For the reconstructed data, and is the weight matrix, and is the bias vector, and The present invention adopts the structure of variational autoencoder (VAE) and improves the generalization and generation capabilities of the model by introducing randomness and regularization.

[0189] Finally, the present invention details a method for intelligently analyzing the optical characteristics of cloud and smoke particle groups based on the above system. The method includes the following steps:

[0190] S1. Physical Parameter Analysis: The input smoke composition data and physical parameters are preprocessed to output the scattering coefficient, extinction coefficient, absorption coefficient, and refractive index of the cloud smoke particles. In this step, the present invention uses a modified Mie scattering theory, combined with particle size distribution and complex refractive index, to accurately calculate these optical parameters. For example, for spherical particles, the scattering efficiency $Q_{sca}$ can be expressed as:

[0191] ,

[0192] in, is the size parameter, and is the Mie coefficient.

[0193] S2. Particle Optical Properties Database: Input the scattering coefficient, extinction coefficient, absorption coefficient, and refractive index of cloud smoke particles. This invention constructs a comprehensive particle optical properties database covering the optical characteristics of various common smoke types. This database includes not only experimental measurement data but also theoretical calculation results, providing a rich reference foundation for subsequent analysis.

[0194] S3. Photoelectric sensor experiment: Receive the smoke model and conduct multi-band simulation experiments to obtain optical data of the smoke at different wavelengths. In this step, the present invention uses advanced spectral analysis techniques such as Fourier transform infrared spectroscopy (FTIR) and Raman spectroscopy to obtain high-resolution spectral data.

[0195] S4. Data comparison: Compare the optical data of smoke at different wavelengths obtained by the photoelectric sensor experiment with the particle optical properties database to obtain the comparison results. The present invention uses a fast retrieval algorithm based on cosine similarity in this step, and its calculation formula is:

[0196] ,

[0197] in, and Represent the spectral data to be compared and the reference spectrum in the database respectively.

[0198] S5. Based on the comparison results obtained in S4 and combined with the cloud and smoke particle distribution pattern and density parameters, an optical property report of the cloud and smoke particles is output. In this step, the present invention utilizes multi-source data fusion technology, comprehensively considering multiple factors such as spectral data, particle distribution, and environmental parameters to generate a comprehensive and detailed optical property report.

[0199] S6. Data Analysis: Output visualization data, along with the optical properties and accuracy of cloud and smoke particles. This step not only provides traditional data charts but also develops a 3D visualization system based on virtual reality (VR) technology, allowing users to intuitively "roam" through the optical properties data of cloud and smoke particles and gain a deeper understanding of complex data relationships.

[0200] The above detailed description demonstrates the powerful analytical capabilities and broad application prospects of the intelligent cloud and smoke particle swarm optical property analysis system and method of the present invention. This system not only achieves multiple technological innovations but also demonstrates excellent performance and flexibility in practical applications, providing strong technical support for atmospheric environmental monitoring and air quality assessment.

[0201] It should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. Intelligent analysis system for optical characteristics of cloud and smoke particle groups, characterized by: include: Particle physics parameter analysis module, used for: Receive input smoke composition data; Based on the atmospheric diffusion model, the particle size distribution and quantity parameters of cloud smoke particles are output; The atmospheric simulation module is connected to the particle physics parameter analysis module and is used to: Receiving the particle size distribution and quantity parameters sent by the particle physics parameter analysis module; Based on the particle size distribution and quantity parameters, a Monte Carlo simulation is performed to output a distribution pattern and density parameters of cloud smoke particles; The physical parameter analysis module is connected to the atmospheric simulation module data and is used to: Based on the particle optical properties database, output the scattering coefficient, extinction coefficient, absorption coefficient and refractive index of cloud smoke particles; The photoelectric sensor is connected to the physical parameter analysis module and is used to: receiving the smoke model output by the atmosphere simulation module; Conduct multi-band simulation experiments to obtain optical data of smoke in different bands; The data fusion module is connected to the photoelectric sensor and the physical parameter analysis module, and is used to: receiving multi-band optical data output by the photoelectric sensor; receiving the scattering coefficient, extinction coefficient, absorption coefficient and refractive index output by the physical parameter analysis module; performing data fusion analysis based on the multi-band optical data and the scattering coefficient, extinction coefficient, absorption coefficient, and refractive index, and outputting a report on the optical characteristics of cloud and smoke particles; The data analysis module is connected to the data fusion module and is used to: receiving the optical property report output by the data fusion module; Based on deep learning algorithms, a database of optical properties of cloud and smoke particles is obtained; Output visualization data and optical properties of cloud and smoke particles and result accuracy.

2. The intelligent analysis system for optical properties of cloud and smoke particle groups according to claim 1 is characterized in that: The particle physics parameter analysis module includes: Input unit, used to set smoke component data input parameters and receive smoke component data input from the outside; A data processing unit, data-connected to the input unit, configured to pre-process the smoke component data and output a pre-processing result; a data comparison unit, data-connected to the data processing unit, configured to receive the preprocessing result output by the data processing unit and compare it with a database of physical parameters of cloud and smoke particles; The data output unit is data-connected to the data comparison unit and is used to output the scattering coefficient, extinction coefficient, absorption coefficient and refractive index of the cloud smoke particles based on the particle optical property database and in combination with the preprocessing results.

3. The intelligent analysis system for optical properties of cloud and smoke particle groups according to claim 1 is characterized in that: The atmospheric simulation module includes: A model building unit for outputting the distribution pattern and density parameters of cloud smoke particles; a data output unit, data-connected to the model building unit, for outputting an optical property report of cloud smoke particles based on a particle optical property database; A data interface unit, data-connected to the data output unit, for receiving multi-band optical data sent by the photoelectric sensor; A physical parameter analysis interface, connected to the data interface unit for receiving the scattering coefficient, extinction coefficient, absorption coefficient and refractive index of cloud and smoke particles output by the physical parameter analysis module; A data processing unit is connected to the data interface unit and the physical parameter analysis interface data, and is used to process the results of the comparison of the optical data in the multi-band and the particle optical performance database, and compare them with the distribution pattern and density parameters of cloud and smoke particles, the optical characteristics report of cloud and smoke particles, and the results of the physical parameter comparison output by the physical parameter analysis module, and output the comparison results.

4. The intelligent analysis system for optical properties of cloud and smoke particle groups according to claim 1 is characterized in that: The photoelectric sensor comprises: A photoelectric sensor unit for receiving light and generating an electrical signal; a signal processing unit, electrically connected to the photoelectric sensor unit, for receiving the electrical signal and processing, storing and outputting the signal; The communication unit is data-connected to the signal processing unit, and is used to receive the signal processed by the signal processing unit and send it to the data fusion module.

5. The intelligent analysis system for optical properties of cloud and smoke particle groups according to claim 1 is characterized in that: The data fusion module includes: A preprocessing unit, configured to receive output data from the signal processing unit; a preprocessing result unit, data-connected to the preprocessing unit, for obtaining an optical property database of cloud and smoke particles using a deep learning algorithm based on the optical property report of the cloud and smoke particles; The data output unit is data-connected to the pre-processing result unit and is used to output visualization data and optical characteristic results and result accuracy of cloud smoke particles.

6. The intelligent analysis system for optical properties of cloud and smoke particle groups according to claim 1, characterized in that: The data analysis module includes: A data input unit, configured to receive visualization data and extract optical characteristic results of cloud and smoke particles from a database of optical characteristics of cloud and smoke particles; a data processing unit, data-connected to the data input unit, for comparing the visual data and outputting display and interpretation of optical property results of cloud and smoke particles; The data output unit is data-connected to the data processing unit and is used to output interactive and visual optical property results of cloud and smoke particle groups.

7. The intelligent analysis system for optical properties of cloud and smoke particle groups according to claim 1 is characterized in that: The photoelectric sensor comprises: Visible light sensor, used to collect light information in the visible light range, and collect light information of the particle group in the visible light range through the optical window; Infrared sensor, used to collect light information within the infrared range, and collect light information of the particle group within the infrared range through the infrared window; The ultraviolet sensor is used to collect light information within the ultraviolet range and collects light information within the ultraviolet range through the ultraviolet window.

8. The intelligent analysis system for optical properties of cloud and smoke particle groups according to claim 1, characterized in that: The system further includes a communication module, which includes: A wireless communication module, used for data transmission using wireless communication protocols, supporting connection with user devices such as smartphones or computers; Wired communication module, used to support local device connection, using USB interface; The cloud service module is used to provide remote data transmission and storage support through cloud communication protocols, and supports connection with user devices such as smartphones or computers.

9. The intelligent analysis system for optical properties of cloud and smoke particle groups according to claim 1, characterized in that: The deep learning algorithms in the data analysis module include algorithms based on deep convolutional neural networks, deep recurrent neural networks, and autoencoders.

10. An intelligent analysis method for optical characteristics of cloud and smoke particle groups based on the system according to any one of claims 1 to 9, characterized in that: The following steps are involved: S1. Physical parameter analysis: pre-process the input smoke component data and physical parameters, and output the scattering coefficient, extinction coefficient, absorption coefficient and refractive index of cloud smoke particles; S2, particle optical properties database, input the scattering coefficient, extinction coefficient, absorption coefficient and refractive index of cloud smoke particles; S3, photoelectric sensor experiment, receives the smoke model, conducts multi-band simulation experiments, and obtains optical data of smoke in different bands; S4. Data comparison: comparing the optical data of smoke in different wavelengths obtained from the photoelectric sensor experiment with the particle optical properties database to obtain comparison results; S5. Based on the comparison results obtained in S4 and the distribution pattern and density parameters of the cloud smoke particles, output an optical property report of the cloud smoke particles; S6. Data analysis, output of visualization data and optical properties of cloud smoke particles and result accuracy.

Images