53 results about "Linear analysis" patented technology

The invention provides a large-span beam formwork three-dimensional parameterization deepening design method, and belongs to the technical field of building construction.The method includes the steps that geometric parameters of a large-span beam three-dimensional parameterization model are obtained, and a multi-scale homogenization mechanical analysis model is established to describe nonlinear characteristics of a concrete material; an arc length control method is combined with a Newton-Raphson iterative improved algorithm to carry out large-deformation nonlinear analysis to calculate deformation control parameters, safety performance is evaluated, support structure parameters are adjusted, and a double-layer game optimization model is established to determine optimal template support system parameters. On the basis of a bending-torsion coupling analysis result, a force transmission path diagram is established through a load transmission path generation function, and an optimal force transmission path is determined by adopting a gradient optimization algorithm, so that the problems that the design of a large-span beam formwork supporting system lacks an accurate mechanical calculation basis, and structural deformation and stress distribution cannot be accurately predicted are solved; and the design of the supporting system is not reasonable.

The invention discloses a cooperative control method for stay cable force and main beam line shape of an extradosed cable-stayed bridge. The cooperative control method comprises the steps that cable force data, main beam line shape data and main beam section internal force monitoring data in the construction stage are collected; based on a preset cable force adjustment judgment criterion, whether cable force adjustment needs to be started or not is judged; when triggering conditions are met, a geometric nonlinear analysis model considering cable-beam-tower cooperative deformation is established, and a cable force-main beam line shape sensitivity influence matrix is generated; screening K stay cables as key adjustment objects by adopting an influence matrix column vector L2 norm sorting method, and performing global optimization based on a simulated annealing algorithm to obtain an optimal displacement mean square error and an optimal cable force adjustment amount; and according to the key adjustment object and the cable force adjustment amount, multi-cable synchronous grading tensioning work is carried out. The invention provides a systematic and intelligent solution for the construction control of the extradosed cable-stayed bridge, and has remarkable economic benefits and social benefits.

A method and system for analyzing a vehicle winch design includes a preprocessor determining contact locations for winch components from a mesh model. The preprocessor determines load vectors at a winch assembly and counter-vectors at a fairlead based on a relative position of a winch wire and a hook retention location based on a winch load rating. A non-linear analysis system determines plastic strains, deflections, and clearances under the winch load rating based on the contact locations, load vector and counter vector, and communicates the plastic strains, deflections, and clearances to a post processing system. The post processing system compares the plastic strains to strain limit at a post processor, compares the deflection to a deflection limit, compares clearances to a clearance limit and generating a display based on comparing the plastic strains, compares the deflection to the deflection limit and compares clearances to the clearance limit.

A first circuit receives a single-ended signal and a reference voltage. The first circuit shifts a common mode of the single-ended signal based on a level-shifting current. A preamplifier receives the single-ended signal and the reference voltage from the first circuit and amplifies the single-ended signal to obtain an amplified signal. A decision feedback equalizer (DFE) system receives the amplified signal and the reference voltage from the preamplifier. The DFE system generates a first average signal height of the amplified signal above the reference voltage and a second average signal height of the amplified signal below the reference voltage. A second circuit provides the level-shifting current to the first circuit and adjusts the level-shifting current based on a linearity parameter associated with the amplified signal.

The invention relates to the technical field of food engineering, and discloses a colorimetric principle-based fried food Maillard reaction product concentration analysis model and frying equipment, and the colorimetric principle-based fried food Maillard reaction product concentration analysis model and frying equipment are established through three steps of experimental design, data acquisition and model construction. According to the concentration analysis model for the Maillard reaction product of the fried food based on the colorimetric principle and the frying equipment, in order to accurately judge the optimal frying end point of the fried food, improve the percent of pass of the fried food and adapt to industrial production, a linear analysis model based on R, G and B values of mobile phone imaging and the absorbance of the Maillard reaction product is established; the deep-frying end point judgment value is determined in combination with sensory evaluation, the optimal deep-frying end point of the deep-fried food can be judged, data are rapidly obtained through standardized mobile phone imaging parameters and substituted into a preset model for calculation, instant monitoring and judgment of the optimal deep-frying end point can be achieved, the qualified rate of the deep-fried food is increased, and the method is suitable for industrial production.

The invention provides an intelligent monitoring method and system based on an AI suspended pouring bridge, and relates to the technical field of bridge construction.The intelligent monitoring method comprises the steps that the experimental environment temperature and experimental image data are obtained; performing linear analysis on the experimental image data to obtain an experimental elevation and an experimental axis; obtaining a first experiment downwarping trend according to the experiment elevation and a preset experiment formwork erection elevation; according to the experimental environment temperature, the experimental elevation and the experimental axis, trend prediction is carried out through the constructed down-warping trend prediction model, and a predicted down-warping trend is obtained; comparing the first experiment down-warping trend with the predicted down-warping trend to generate a hanging basket construction adjustment strategy so as to optimize and adjust hanging basket construction; and performing concrete pouring construction management through the predicted formwork erecting elevation corresponding to the optimized and adjusted second experiment downwarping trend. According to the monitoring method and system provided by the invention, the accuracy of formwork erecting elevation can be ensured, the linear effective control of the suspended pouring bridge construction is realized, and the construction precision and efficiency are improved.

The invention relates to the technical field of data processing and engineering project process management, in particular to a full-process intelligent management method and system for bridge and tunnel maintenance engineering. The method comprises the following steps: extracting a disease geometric contour based on bridge and tunnel inner wall three-dimensional point cloud data, and calculating a centroid coordinate, a volume and a surface contour area; constructing a space retrieval sphere by using a target disease centroid, screening an adjacent disease set, and calculating a local aggregation density by combining the volume and the centroid distance; calculating the average depth of the target disease, and deducing the overlapping sharing rate of the working plane; constructing a group topology consumption equivalent through linear analysis; and inputting the extreme gradient lifting model to output an optimal repair material dosage table and a material distribution operation instruction book. According to the method, the disease space distribution and aggregation degree can be accurately evaluated, the construction loss surface overlapping probability and the material sharing potential are scientifically evaluated, the material preparation accuracy is improved, waste and cost are reduced, and the whole-process intelligent management of bridge and tunnel maintenance engineering is realized.

The invention discloses a low-frequency oscillation optimization method of an electric power system containing an SDC and a DFIG based on balanced truncation and series control compensation. The low-frequency oscillation optimization method comprises the following steps: 1, establishing a linear analysis model of a wind power grid-connected electric power system containing a synchronous generator SG, the DFIG and an additional damping controller SDC; 2, truncation processing is carried out on a state matrix in the state matrix equation of the main system after balance transformation, and a main system state matrix equation after order reduction is obtained; 3, constructing a time domain matrix of the system containing the series feedback controller to obtain a system state matrix equation containing the series feedback controller; 4, solving a characteristic value and a damping ratio of the system state matrix equation containing the series feedback controller, and constructing a control parameter analytic expression of the characteristic value and the damping ratio to the SDC in a low-frequency oscillation mode; and 5, solving by using the low-frequency oscillation optimization model of the SDC to obtain the optimal adjustment amount of the control parameters of the SDC. According to the invention, the control compensator is added to optimize the characteristic value under low-frequency oscillation and the control parameters of the SDC, so that the stability of the system is optimized.

The present application relates to the field of agricultural information technology and smart agriculture, and in particular to a dominant factor screening and litchi flowering rate prediction method based on a nonlinear model, which comprises the following technical solutions: collecting multi-dimensional original data of Feizixiao litchi in the flowering induction period; performing linear regression test on the original data, and starting a nonlinear analysis process when the fitting degree of the linear model is lower than a threshold value; constructing and training an XGBoost nonlinear regression model; using a SHAP method to interpret the model, calculating the average absolute SHAP value of each feature and sorting, and screening out nonlinear dominant factors; drawing a SHAP dependence graph for the dominant factors, and automatically determining the key threshold value affecting the flowering rate through curvature analysis; inputting the dominant factor data of the target year into the model, and outputting the flowering rate prediction value; the present application can effectively capture the nonlinear relationship between the influencing factors and the flowering rate, realize objective screening of the dominant factors and automatic identification of the key threshold value, significantly improve the flowering rate prediction accuracy, and provide a scientific basis for litchi production precision management.

The present application belongs to the technical field of numerical simulation of underground water in mine, and discloses a GMS cooperated with GIS health risk prediction method for backfilling underground water in open-pit mine. The present application combines with geological data around mining area, carries out fine 3D geological modeling through GCAD, EVS, Tin and Solid module in GMS, and on the basis of analyzing hydrogeological conditions of mining area, generalizes boundary conditions and aquifer, determines source and sink, hydrogeological parameters, establishes relatively accurate 3D underground water seepage model of open-pit mine by GMS, and identifies and verifies the model combining with underground water field monitoring data; determines each parameter of solute transport model through static leaching test, dynamic adsorption test and multi-parameter free energy relationship model, establishes solute transport equation of biological, chemical and seepage mutual action by multiple regression linear analysis, outputs solute transport result of open-pit mine backfilling pollutant in different stress periods of simulation period by MT3DMS, disperses solute transport result by Scatter in GMS and Field Calculator based on Python algorithm, and visually expresses in GIS combining with health risk prediction model, so as to solve the problem that potential health risk of backfilling material to underground water pollution in the process of open-pit mine backfilling mining or backfilling treatment is difficult to predict.

The application discloses a kind of sensor fault detection method and device of structural health monitoring system, it is related to structural health monitoring technical field, comprising: obtaining the nonlinear structure monitoring data of sensor system collected by structural health monitoring system, nonlinear structure monitoring data is handled using improved geometric PNL hybrid model, obtain linear to-be-separated mixed signal, using FastICA model to process to-be-separated mixed signal, obtain multiple independent elements, determine the separation matrix of improved geometric post-nonlinear independent component analysis model according to multiple independent elements, using the processing of improved geometric post-nonlinear independent component analysis to real-time collection nonlinear structure monitoring data, determine whether sensor system exists fault and the sensor of fault occurrence.The method can still complete linearization processing to sensor mixed signal under the condition that prior knowledge is unknown, compared with simple linear ICA analysis algorithm, it is more suitable for complex nonlinear structure.

The invention provides a structural damage positioning method and system based on wavelet-PCA, and relates to the technical field of structural damage positioning. The method comprises the following steps: acquiring a structure vibration response signal through a sensor to obtain an original signal data set; de-noising the signal by adopting wavelet transform, and extracting a principal component characteristic matrix through principal component analysis; performing phase-space reconstruction on the principal component time sequence to generate a high-dimensional power system track; calculating quantization factors such as recursion rate, certainty and average diagonal length based on recursion plot analysis, and forming a recursion quantization feature matrix; generating a spatial correlation distribution diagram by analyzing the correlation between the recursion characteristics of different sensor positions; and realizing damage positioning according to the correlation abnormal region. The early weak damage can be effectively detected, the defect that a traditional linear analysis method is insensitive to nonlinear characteristics is overcome, and the method has the advantages of being high in anti-interference performance and accurate and reliable in positioning.

The application discloses a kind of high-precision digital oscilloscope nonlinear analysis and calibration method, comprising: subsection reorganization optimization process is used to extract the nonlinear error of digital oscilloscope at different working temperatures;Subsection type modeling fitting process is used to fit the nonlinear error, and obtain nonlinear error model;FPGA real-time calibration process, the error model is stored in high-precision digital oscilloscope FPGA, and is stored as calibration data table, and the output digital code of digital oscilloscope is determined by FPGA, and the corresponding error model in FPGA is found out in real time, and the nonlinear error value is obtained, and the nonlinear error value is compensated to real-time output digital code, and calibration compensation is completed;The method proposed in the application can enable high-precision digital oscilloscope to realize nonlinear digital calibration, so as to significantly improve its linearity and dynamic performance, including SFDR and ENOB.

The invention discloses a complete construction method and system for a known MDS matrix in a symmetric encryption network, a product and a storage medium, belongs to the field of information security, and is used for solving the problems that the dynamic variable diffusion layer design of a cryptographic algorithm cannot be met, and the differential resistance and the linear analysis capability are relatively weak. The method comprises the following steps: extracting a generator of an MDS matrix; on the basis of the shift number of the longest period cyclic permutation sum on the permutation group, a parameter function of the generator is constructed, and when the parameter function is identical transformation and j is equal to 0, the parameter function is identical transformation about the generator; and replacing the element multiplication operation of the MDS matrix in the n-dimensional finite field with the parameter function, n being a positive integer, and updating the mapping relation of MDS transformation in the finite field. According to the method, under the condition that the known MDS matrix structure, elements and implementation cost are not changed, the parameter number scale of the MDS matrix on the existing finite field is enlarged by O (n!) times, and powerful support is provided for dynamic design of a future password extension layer.

This invention discloses an intelligent biological reaction monitoring system and method, belonging to the field of intelligent monitoring. The system includes: a reaction stage segmentation unit for segmenting and determining multiple reaction stages; a variable relationship mining unit for determining dominant and auxiliary variables, mining linear relationships between variables, coordinating insensitive loss functions, and supervising the training of a soft monitoring module; a continuous monitoring and acquisition unit for configuring monitoring equipment groups, performing continuous monitoring, and determining reaction monitoring data; a data analysis and screening unit for preprocessing data, performing linear analysis and abnormal reaction localization, and determining valid monitoring data; a game equilibrium decision-making unit for performing game equilibrium decisions on abnormal reaction identifiers and determining reaction regulation strategies; and a reaction monitoring management unit for responding to regulation strategies, performing feedback monitoring analysis and reaction monitoring management, improving the accuracy of biological reaction monitoring, achieving real-time online monitoring, and intelligently regulating abnormal situations.

The invention discloses a terminal information security protection system and method, and relates to the technical field of information security. The encryption round number is dynamically determined through terminal hardware state data and to-be-protected data attributes, and self-adaptive matching of encryption strength and terminal resources is achieved; generating an initial key and a secret parameter according to the key material, and driving key expansion and encryption structure configuration through the secret parameter to obtain a key matched with the size of the data block; and performing multi-round encryption on the data by adopting a preset encryption strategy, and adding the meta-information to a ciphertext header for verification and decryption. The dynamic nature and the unpredictability of the secret key and the encryption structure are enhanced while light weight is guaranteed, the resistance to differential analysis, linear analysis and side channel attacks is effectively improved, and the reliability and the practicability of terminal information security protection are improved.

The invention relates to a coupling modeling method for a nonlinear loading lever system and a linear full-aircraft structure. The coupling modeling method comprises the following steps: establishing a parameterized model of the lever system, and defining a model building function, section attributes, prestress application and boundary conditions; constructing a full-aircraft substructure model, processing a full-aircraft structure by adopting a linear analysis method, and only retaining the degrees of freedom of a loading point and a fixed fulcrum; a data interface between a main structure and a substructure is defined, and through standardized data module definition, the consistency of node information and connection relation of a lever system and a full-aircraft structure is achieved. Compared with the prior art, the method has the advantages that a non-linear lever system and linear structure coupling mode is adopted, so that the calculation efficiency is improved while the calculation precision is ensured; the method realizes parametric modeling of the lever system, facilitates rapid modification and optimization of the test scheme, can effectively predict the structural response in the test process, and has the advantages of high modeling efficiency, high calculation precision, strong applicability and the like.

The present application belongs to the field of metrology and detection, and particularly relates to a measurement system reliability analysis system and method. Current measurement system analysis using computing function software has a high degree of human intervention, which introduces artificial uncertainty. The present application provides a measurement system reliability analysis system and method, which includes a stability degree data analysis module, a deviation and linearity analysis module, and a repeatability and reproducibility analysis module. The system analyzes the characteristics of the measurement system in a targeted manner and ensures compliance with standard requirements, standardizes the analysis method of the measurement system, and streamlines the process. Under the premise of ensuring analysis accuracy, the operation method is simplified. The analysis speed of the measurement system is improved, and the problems of uncontrolled calculation process caused by human factors and questionable effectiveness of analysis conclusions caused by human factors are reduced. Personnel costs are reduced, time is greatly saved, and the system is more accurate and reliable.

The invention relates to a method for predicting dynamic fracture behaviors and parameters of a quasi-brittle material based on an equivalent area, and aims to solve the technical problems that in the prior art, a dynamic fracture parameter test of quasi-brittle materials such as concrete is remarkable in size effect and complex in test process, and multiple groups of test pieces and special equipment are needed. According to the method, a technical means of taking an equivalent area Ae as a core design parameter is adopted, two groups of quantitative test pieces with a specific Ae value ratio are designed, a linear analysis model of a peak load and the equivalent area is established, the dynamic fracture toughness KIC and the tensile strength ft without a size effect are synchronously determined, and a fracture total curve with 95% reliability is constructed. The method has the technical effects that the real dynamic fracture parameters of the material can be accurately measured only by a small number of laboratory small-size test pieces, the size effect is effectively eliminated, the test process is remarkably simplified, the equipment requirement is reduced, and the reliable prediction of the fracture behaviors of engineering structures with different geometric sizes is realized.

The invention discloses a site differentiation earthquake response analysis method and system based on earthquake motion types, and belongs to the technical field of building earthquake-proof engineering, and the method comprises the steps: dividing a target site into three types of sites according to the obtained drilling information of the target site, and constructing a site model corresponding to each type of site; respectively calculating dynamic response results of each site model under the near-field pulse type seismic motion and the near-field non-pulse type seismic motion by adopting a time domain nonlinear analysis method; generating two groups of surface response spectrum amplification coefficients according to the dynamic response result of each site model, and comparing peak values with periods; and according to a comparison result of the peak value and the period, combining each site model to obtain a differential earthquake response analysis result. According to the method, the significant amplification effect of pulse type seismic oscillation on a soft soil site and the significant amplification effect of non-pulse type seismic oscillation on a medium-hard soil site are effectively revealed, and the problem that the difference of seismic oscillation types is not fully considered in existing analysis is solved.

The application discloses a data security communication method of a 6G terahertz communication system, and relates to the technical field of data processing, wherein a key generation strategy combining continuous randomization and multi-dimensional target optimization is used to generate all round keys of a symmetric encryption algorithm, so that each round key generated is kept to have maximum difference with previous keys in mathematical characteristics, linear correlation between the keys is reduced, the ability to resist differential analysis and linear analysis is enhanced, all round keys of the symmetric encryption algorithm are combined to form a total key, and fragmentation processing is performed on the total key; the fragmented key obtained through the fragmentation processing is encrypted and transmitted by using the public key of a receiver in the 6G terahertz communication system; even if part of the fragments are intercepted in the transmission process, an attacker cannot restore the complete key; and the total key can be obtained only after a sufficient number of fragments are collected and decrypted by combining the private key, so that the security of key distribution is improved, and the method is suitable for the complex environment of 6G terahertz communication.

The invention provides a basic design method for a high-rotating-speed centrifugal compressor based on finite elements. Comprising the following steps: 1, collecting technical data (including rotating speed, vibration load and basic template drawing) of a unit, and constructing a finite element three-dimensional model; 2, inputting the section size, the material attribute and the support condition of the component; 3, defining an equipment time-history function and a three-axis self-weight load, and analyzing vibration by adopting steady-state / transient response; and 4, executing linear analysis, verifying a result, and controlling an eccentric distance, a top surface vibration speed peak value and a root-mean-square value. According to the method, through three-dimensional finite element dynamic simulation, the resonance risk is accurately avoided, the structural economy is optimized, and standardized closed-loop design is achieved.

The invention belongs to the technical field of informatization agriculture, and particularly relates to a climate-soil synergistic effect-based apple quality multi-index prediction model construction method, which comprises the following steps of: data collection and preprocessing: collecting climate indexes, soil chemical indexes and quality index data corresponding to apple samples, carrying out standardization processing on all continuous variables, and carrying out calculation on the standard variables; dividing a training set and a test set according to a preset proportion; feature engineering: high-order features are constructed based on agronomy knowledge, an engineering feature pool is formed, and the high-order features comprise an interaction feature, a seasonal aggregation feature, a soil-climate balance feature and a ratio feature; according to the method, a systematic basic statistical analysis-high-order feature engineering-two-stage modeling research framework is constructed, the limitation of traditional single-factor linear analysis is broken through, the nonlinear synergistic effect of climate-soil factors is effectively captured, and high-quality feature input and a scientific modeling thought are provided for apple quality prediction.

The invention relates to the technical field of cryptographic analysis, and discloses an automatic linear path analysis method and system for an ARX type cryptographic algorithm, and the method comprises the following steps: constructing a round function, configuring a linear analysis target, and selecting a search strategy; performing linear mask initialization on the round function, and establishing an association relationship between inter-round mask variables; generating local linear mask propagation expressions based on the linear approximation characteristic of each basic operation unit, combining the local linear mask propagation expressions into a global linear approximation model, and calculating a correlation expression of a linear path; executing a search strategy on the global linear approximation model; converting the global linear approximation model and the search strategy into constraint description and solving the constraint description; according to the method, the modeling complexity of linear analysis is reduced through visual modeling and selection of the path search strategy, and automation of the linear path search process is achieved.

A method for visually indicating whether any of the linear analysis assumptions have been violated, a computer, and a non-temporary computer-readable medium are provided. [Solution] The method receives data indicating whether each of the multiple linear metrics generated by the linear analysis of the model in the simulation violated any of the linear analysis assumptions. Each status icon displayed on the GUI is visually associated with one corresponding linear metric among the multiple linear metrics and one corresponding model element of the computer implementation model. Each model element is displayed on the GUI in a manner that represents the hierarchical model structure of the model elements. Each status icon visually indicates whether one of the multiple linear metrics for one corresponding model element among the multiple model elements generated by the linear analysis of the model in the simulation violated any of the linear analysis assumptions.

The invention discloses a carburized gear internal oxidation rating method and system based on deep learning, and belongs to the technical field of carburized gear quality detection.The carburized gear internal oxidation rating method comprises the steps that a scanning electron microscope image of a carburized gear is obtained and preprocessed, and after noise is reduced through anisotropic diffusion filtering, an internal oxidation structure area is segmented through morphological operation; textural features are extracted based on gray-level co-occurrence matrix and local binary pattern fusion, and feature vectors are screened and optimized through Pearson's correlation coefficients; inputting the feature vectors into a support vector machine classification model subjected to particle swarm optimization to obtain an internal oxidation degree preliminary classification result; and in combination with an internal oxidation comprehensive evaluation system, a final quantitative rating result is output through probability threshold judgment, deviation matrix linear analysis correction and service condition and material characteristic matching. According to the carburizing gear internal oxidation grading method, automation, standardization and precision of carburizing gear internal oxidation grading are achieved, and scientific support is provided for gear quality detection and service safety guarantee.

The application discloses a lightweight encryption method and device based on an ARX module. The method comprises the following steps: storing a master key in a linear feedback shift register (LFSR); taking four beats of LFSR precession as one round and generating a sub-key; generating 32 sub-keys in sequence to form a sub-key sequence; obtaining plaintext to be encrypted, and grouping the plaintext to be encrypted; performing 32 rounds of encryption on each group of plaintext based on the sub-key sequence; merging the encrypted ciphertext of each group of plaintext to obtain the overall encrypted ciphertext corresponding to the plaintext to be encrypted; and obtaining the overall encrypted ciphertext by a decryption party, grouping the overall encrypted ciphertext, and taking each group to include 128 bits; and performing a decryption operation on each group of ciphertext based on the sub-key sequence. The application can meet the high efficiency in the resource-limited scene, and can resist the security threats caused by differential analysis and linear analysis.

The invention relates to the technical field of water environment monitoring, in particular to a water environment ecological pollution detection big data analysis system which comprises the steps of setting an influence area, dividing a key collection water area based on an intersection point of an influence area range and a water environment river bank boundary, and collecting real-time monitoring data in the key collection water area. Dividing historical data into visual data and observation data, processing the visual data and the observation data to obtain a subjective analysis curve and an observation curve, performing linear analysis on the curves to generate a performance characteristic curve of the subjective analysis curve, and dividing real-time monitoring data into candidate curve segments; calculating a similarity value of the fragment and the performance characteristic curve to determine an anomaly detection parameter and a main detection area; according to the method, the abnormal nodes in the observation data can be accurately identified, the data processing efficiency and accuracy are improved, the abnormal detection parameters and the main detection area can be rapidly and reliably positioned, and the preciseness of the abnormal judgment result is further guaranteed through secondary sampling verification of the main detection area.

A method for carrying out a non-linear analysis of at least two gears in the frequency domain provides for generating a gear mathematical model, define a motion equation of such model and solve such equation with a method of the coupled harmonic balancing, the implementation of method comprising the solution of such equation simultaneously considering a static condition and a dynamic condition of the interaction between the gears.

The invention discloses an insulator pollution level discrimination method and system based on spectrum and machine learning, belongs to the field of intelligent state detection of high-voltage power transmission equipment, and aims to solve the problems that traditional detection depends on manpower, the real-time performance is poor, and the precision is interfered. The method comprises the following steps: preprocessing an insulator test piece, simulating salt spray pollution, synchronously acquiring discharge arc spectrum and electrical parameters on a high-voltage discharge platform, performing baseline correction and filtering preprocessing, extracting spectral line characteristics through Voigt fitting, and obtaining characteristic vectors such as electron temperature and density according to a double-spectral-line method formula and a Stark broadening simplification formula; and constructing an IPSO-BP model containing a triple optimization mechanism, and inputting the feature vector to complete training so as to discriminate the pollution level. The system comprises a spectrum acquisition unit, a linear analysis and parameter extraction unit and an intelligent diagnosis unit, supports remote online monitoring, realizes non-contact accurate discrimination, averagely reduces MSE by 74.3%, can be integrated in a multi-scene monitoring system, and provides technical support for safe operation and maintenance of a power grid.