93results about How to "Reduce the scope of investigation" patented technology

The invention relates to an abnormal power consumption detection method based on a neural network. The method diagnoses and analyzes an operation condition of equipment based on an established abnormal power consumption detection model, judges whether metering equipment is in a normal operation condition and realizes an aid decision-making function. The method concretely comprises the following steps of (1) data acquisition: data are mainly from electric energy metering data, operation condition data and event recording data in an electric energy meter and an acquisition terminal; (2) data cleaning: the used data can enter the model after being subjected to data cleaning and screening; (3) data classification: after data cleaning completes, the data are calibrated, one column of numbers for representing data classification is added at the end of the data for classification, and the data subjected to data calibration are integrated into training data; (4) a modeling process: an algorithm model is constructed in a manner of supervised learning; (5) model implementation; and (6) result analysis: the final accuracy rate of abnormal power consumption found by the model maintains at a high level.

The invention relates to video monitoring and discloses an intelligent person capture system and a person monitoring management method. In the invention, the whole system comprises a front-end human face snapshot subsystem, a center analyzing server and a data management subsystem, wherein the human face snapshot subsystem, the center analyzing server and the data management subsystem are respectively connected with a switchboard; and the switchboard is utilized for forming a local area network for carrying out mutual communication. Because a front-end snapshot and rear-end analysis mode is adopted in the whole system, the system is not just in charge of acquisition of images, realizes intelligent person capture, can support retrieval afterwards, thereby being greatly convenient for the rapid search and gradual check of a public security department on criminal suspects.

The invention relates to the field of industrial control system information safety, in particular to an industrial control information safety monitoring system adopting a black list and a white list for analysis, which comprises the following steps of: (1) collecting an initial sample library; (2) blacklist rule formation and alarm generation; (3) white list rule formation and alarm generation; (4) intelligently generating a communication white list; (5) protocol rule formation and alarm generation; and (6) performing visual alarm and report display. The beneficial effects of the invention arethat the system carries out the credible monitoring of assets in the industrial control network, carries out the credible monitoring of communication in the industrial control network, carries out the deep analysis of a plurality of industrial protocols, carries out the visual display of a topological structure of the industrial control network, and carries out the report statistics of a threat form in the industrial control network; the method is beneficial to realization of abnormal monitoring in the network, is beneficial to realization of quick positioning and troubleshooting of problem assets and problem communication behaviors, is beneficial to avoiding human operation errors, improves the accuracy of a white list, reduces missing reports and false reports, and constructs a monitoring system easy to operate and capable of quickly positioning problems.

The invention discloses a fake plate vehicle detection method based on vehicle hot spot area probability aggregation. The fake plate vehicle detection method comprises the steps that S1, checkpoint data is extracted; S2, the checkpoint data is cleaned; S3, flow directions between the checkpoints are calculated; S4, a list of abnormally-acting vehicles are calculated: according to a checkpoint pair of every vehicle extracted by S2, checkpoint pair probabilities are calculated, and are arranged in a descending order according to times having probabilities lower than a threshold value, and then first N rows are selected to be used as the list of the abnormally-acting vehicles; S5, vehicle hot spot area checkpoint points are calculated; S6, the probabilities of the vehicles moving between two hot spot area checkpoint points are calculated; S7, the checkpoints are aggregated together according to the probabilities calculated by S6; S8, probabilities of fake plate vehicles Z are calculated according to a vehicle checkpoint aggregation condition. Random checkpoint license number identification errors are overcome, and by using a characteristic of relatively fixed vehicle moving areas, the vehicles having large suspicion of faking the license plates are selected, the suspicion range is greatly reduced, and the practicability is good.

The invention is applicable to the technical field of power supply,and provides a distribution wire fault online monitoring method and terminal equipment. The method comprises the steps that distribution wire work attribute information sent by a plurality of acquisition devices is received,wherein the distribution wire work attribute information comprises voltage,current and position information of a plurality of collection points of a distribution wire; then the voltage of a target collection point is compared with a preset voltage standard threshold,the current is compared with a preset current standard threshold,and whether the distribution wire of the target collection point has a fault or not is judged,wherein the target collection point is any one of the collection points; finally,the position information of the target collection point is sent to a maintenance terminal if it is judged that the distribution wire of the target collection point has the fault. The troubleshooting range can be reduced,the fault position of the distribution wire can be quickly located,and the maintenance processing efficiency is improved.

The invention relates to the technical field of water resource monitoring, in particular to a water environment pollution analysis system based on big data, comprising: an acquisition unit for acquiring river basic data and pollution source data; the division unit that is used for carrying out river reach division on the river; the modeling unit that is used for establishing a pollutant tracing relationship and a monitoring section water quality response relationship; the analysis unit that is used for performing pollution source structure analysis on each water outlet of each control unit according to the internal and external source contribution analysis and the pollution source type contribution analysis, and analyzing the contribution rate of the pollution source to the water quality of the monitoring section; and the output unit that is used for outputting an analysis result. According to the invention, the monitoring section is taken as the core, the pollution source causing themonitoring section can be analyzed, the pollution source with relatively large contribution degree to the river section water quality in the river is locked, and the technical problem that when an existing method is used for coping with sudden pollution sources, pollution source checking and tracing work cannot be effectively carried out in time is solved.

The invention relates to a gridding monitoring, diffusion early warning and tracing method for unorganized VOCs of an industrial enterprise. The method comprises the following steps: establishing a VOCs unorganized emission source list of the industrial enterprise; vOCs gridding monitoring stations are screened and arranged on the basis of a VOCs unorganized emission source list and local perennial meteorological conditions through a comprehensive rating method of functional region division and gridding point distribution; creating a VOCs management and control platform, displaying the grid VOCs emission level, time change trend and characteristics of the whole plant, and analyzing the total VOCs emission amount, high concentration value and occurrence time of the pollution map; through the optimized gas diffusion model and in combination with VOCs monitoring data of the whole plant, diffusion early warning analysis of VOCs pollution in the whole enterprise plant area is realized. And traceability analysis of VOCs pollution is realized through a source analysis technology. According to the invention, an integrated fine supervision mode of source combing, process monitoring and system management and control early warning of unorganized VOCs pollution of industrial enterprises is realized.

The invention relates to a prefabricated pump station. The prefabricated pump station comprises a base, a cylinder arranged on the base, a submersible sewage pump arranged on the bottom of the interior of the cylinder, and a control cabinet used for controlling the submersible sewage pump to operate, wherein a water inlet pipe is arranged on one side of the cylinder, a water outlet pipe is arranged on the other side of the cylinder, a crush type grating is arranged in the cylinder, and the water inlet pipe is connected with the crush type grating through a grating connector; a water outlet of the submersible sewage pump is connected with the water outlet pipe through a pressure pipeline, the control cabinet is connected with the crush type grating and the submersible sewage pump, a sluice valve, a water quality detecting device and an anti-blocking probe are arranged on the water inlet pipe, the water quality detecting device transmits a signal to the sluice valve to control the sluice valve to be open and closed, the anti-blocking probe transmits a signal to the control cabinet when blocking of the water inlet is detected, and a hazardous gas detecting device is arranged inside the cylinder and transmits a signal to the control cabinet. Due to the arrangement of a gas alarm device, people can be warned of poison gas generated by sediment on the bottom of the pump station, and using is safe.

The invention discloses a fault positioning circuit and positioning method for a high voltage loop interlocking. The circuit comprises a potential detecting circuit and a main controller, wherein thepotential detecting circuit sends a potential signal at the front end and the rear end of an interlocking switch of the detected high voltage component to the main controller, and the main controllerdetermines whether the high voltage component corresponding to the interlocking switch interlocks normally and / or an interlocking fault position according to the potential signal at the front end andthe rear end of the interlocking switch. The fault positioning circuit and positioning method for the high voltage loop interlocking has the advantages that the fault position can be quickly positioned after an interlocking fault occurs by determining whether the high voltage component is faulty and the position of the fault relative to the high voltage component according to the potential signalat the front end and the rear end of the interlocking switch of the high voltage component, so that the maintenance can be facilitated, the time and cost of manual troubleshooting are saved, and the troubleshooting efficiency can be greatly improved.

The invention discloses a method and a device for identifying users with abnormal electricity consumption. The method comprises the following steps: acquiring historical power consumption data, and identifying an abnormal value in the historical power consumption data; filtering abnormal values in the historical power consumption data; searching a user electricity consumption characteristic variable in the electricity consumption historical data; and utilizing a preset modeling algorithm to construct an abnormal power consumption user identification model, utilizing the abnormal power consumption user identification modeld to identify abnormal users in the historical power consumption data, wherein input data of the abnormal power consumption user identification model are data corresponding to the user power consumption characteristic variables, and output data are abnormal user identifiers. According to the invention, the technical problems of low detection efficiency and small coverage area of the inspection of the abnormal power utilization users in the related art are solved.

A two-time screening method for cars with false license plates based on probability distribution comprises the following steps that firstly, traffic flow recording data of bayonets is obtained and subjected to data cleaning, and traffic flow recording data of the bayonets is obtained; secondly, the traffic flow recording data of the bayonets is sorted, and traffic through bayonet vector pairs are extracted; thirdly, the spatial probability distribution of vehicle flows between the bayonets is calculated; fourthly, the bayonet records are compared with a data base, and a primary screened false license plate set is obtained; fifthly, based on the spatial probability distribution of the vehicle flows in the third step, license plates meeting the spatial probability distribution are obtained in the primary screened false license plate set; sixthly, according to the spatial probability distribution of the vehicle flows, each character recognition error probability is determined; seventhly, according to the probability that the license plates meet the spatial probability distribution and the character recognition error probabilities, the probability that the license plates are false is comprehensively judged. By means of the two-time screening method, the bayonet recognition precision deficiency can be overcome to a certain extent, the false license plate checking range can be effectively narrowed, and the hitting precision can be improved.

The invention provides an operator data-based public security case time-space analysis method and apparatus. The method comprises the steps of obtaining occurrence information of serial cases, whereinthe occurrence information comprises occurrence time, occurrence places, occurrence time error ranges and occurrence place error ranges of the cases; according to a sequence of the occurrence time ofthe serial cases, sequentially connecting the occurrence places into a case action track on a map; and screening out operator clients occurring on the case action track in the occurrence time, and classifying the operator clients as suspects, thereby obtaining a suspect list. The method is based on operator data, so that the coverage range is wide and the blind area is small; the operator clientssatisfying the case action track are screened and classified as the suspects, so that the suspect check range is effectively narrowed, the case solving time is shortened, and the case solving efficiency and quality of public security organs are improved.

The invention discloses an abnormal user identification method based on a GBDT algorithm and factor fusion. The abnormal user identification method comprises the following steps: step 1, data preprocessing: collecting user power consumption data and transformer area power supply data and preprocessing the user power consumption data and transformer area power supply data; step 2, feature extraction: screening out features related to electricity stealing from historical electricity consumption and electricity consumption transaction related features by adopting a support vector machine algorithm according to the relationship between each feature and the electricity stealing behavior; step 3, model construction: using the features related to electricity stealing screened in the step 2 as input data of an abnormal user identification model, and using GBDT as a modeling algorithm to construct the abnormal user identification model; and calculating a suspected abnormal electricity user through the abnormal user identification model, and calculating the electricity stealing probability P of the abnormal user according to a suspected electricity stealing probability formula. According tothe invention, accurate identification of electricity consumption behavior abnormity and calculation of the electricity stealing probability of the user can be effectively realized, so that the troubleshooting range is reduced, the troubleshooting key point is realized, the workload of troubleshooting personnel is reduced, and the submission accuracy is improved.

The invention provides a screening method of a non-ribosomal protein-RNA composite near-nature structure, and belongs to the field of protein-RNA molecular docking composite structure prediction. Firstly, various possible combination modes of protein-RNA are obtained through conformation searching; then rationality of the combination modes is assessed, static and van and der waals interaction energy among the protein-RNA molecules and amino acid nucleotide pairing preference on composite interfaces are comprehensively taken into consideration, and the weight of each item is obtained in the mode that a linear regression method is adopted and accordingly fitting of a ligand root-mean-square deviation and weighted combination values of energy items of a docking structure is carried out; finally, according to the order that the values are sequenced from smaller ones to larger ones, a near-nature structure is judged. The screening method has a good effect on screening of the non-ribosomal protein-RNA molecular docking near-nature structure, the successful rate is high, and the method can be used for the field of protein-RNA composite structure prediction and provides an important basis for molecule improvement and design.

The invention discloses an electromagnetic spectrum monitoring system and method based on a wireless ad hoc network. The electromagnetic spectrum monitoring system comprises a plurality of monitoringnodes used for monitoring electromagnetic spectrums in a monitoring area and a control center performing data interaction with the monitoring nodes. And the plurality of monitoring nodes forms a multi-hop network to carry out data interaction with the control center. By adopting low-cost wireless ad hoc network transmission, transmission of monitoring data to a superior control center is achieved,and the problem that traditional monitoring equipment is high in dependence on infrastructure is solved; The electromagnetic spectrum environment of the monitoring area can be effectively grasped, the monitoring system can give an alarm in real time when abnormal signals occur, the occurrence area can be rapidly positioned, the troubleshooting range is reduced, and convenience is brought to dailymonitoring of supervisors.

The invention discloses a zero-electric-quantity user troubleshooting method based on a radar map comprehensive evaluation method, and relates to a zero-electric-quantity user troubleshooting method.An original general survey mode often consumes a large amount of manpower and material resources, and sampling survey often neglects many abnormal users. The method comprises the following steps: firstly, acquiring multi-source data information of a user; extracting zero-electric-quantity user troubleshooting key factors including user basic attributes, user electricity consumption characteristics, station area characteristics to which users belong and important abnormal events of the users; carrying out data preprocessing, determining the weight of each index; and finally, selecting a radar map comprehensive evaluation method to sort the zero-electric-quantity users, obtaining the troubleshooting score and sequence of the zero-electric-quantity users, drawing a comprehensive evaluation radar map of each user, and analyzing the overall score and the single index score of each user. According to the technical scheme, the troubleshooting priorities of the zero-electric-quantity users areranked, accurate positioning of the zero-electric-quantity abnormal users is achieved, the troubleshooting range is greatly reduced, and manpower and material resources are saved.

The invention discloses a root cause positioning algorithm based on call chain data. Anomaly detection is performed on the call chain data of a time window, if an anomaly occurs, the call chain data is traversed from the request header of each abnormal request, exceptions are filtered from each service instance requested to be accessed through 3delta exception filtering, and are scored one by one,and the scores, the dependency relationship among the service instances and the information of each service instance requested to be covered by each request are input into a custom square matrix andare subjected to iterative computation to obtain a service instance exception score; the requested coverage condition of the service instance of the normal request, the requested coverage condition ofthe service instance and the dependency relationship between the service instances are input into a custom square matrix and are subjected to iterative computation to obtain a normal score of the service instance; and a final score of each service instance is calculated by adopting a spectrum method in combination with the abnormal score and the normal score of the service instance, and a suspicious service instance list is returned according to the final score. A fault root cause list with a ranking sequence is output, so that the troubleshooting range of operation and maintenance personnelis reduced.

The invention discloses an LED lamp detection equipment. The LED lamp detection equipment comprises a box, wherein a plurality of accommodating plates are arranged inside the box horizontally; a lampbase is arranged at the upper end of each accommodating plate; the side wall of the box is provided with a water tank; a water injecting tube and a water outlet tube are symmetrically inserted into the box; the water injecting tube communicates to the water tank and the box; valves are arranged on the water injecting tube and the water outlet tube; an L-shaped tube is fixedly inserted into the upper end of the box; one end of the L-shaped tube far away from a through hole is fixedly connected with a mounting shell; and detection mechanisms are arranged in the L-shaped tube and the mounting shell. Through adoption of the LED lamp detection equipment, the checking time for qualified products can be saved; the checking range is narrowed greatly; a large amount of time and energy are saved; and high detection sensitivity, accurate detection result and higher efficiency are achieved.

The invention discloses a pressure vessel weld joint detection tool and detection method, and the tool comprises a pedestal, wherein the top of the pedestal is fixedly connected with a supporting rod,the top end of the supporting rod is fixedly connected with a supporting frame, the pedestal is provided with a driving mechanism, and the left side of the top of the pedestal is provided with an airexhaust assembly; and an adjusting assembly is arranged on the right side of the top of the base. According to the pressure vessel weld joint detection tool and the detection method, through mutual cooperation of the adjusting assembly and the air exhaust assembly, a worker can be helped to judge the approximate leakage position in the detection process, the troubleshooting range after leakage detection is reduced, the labor intensity of the worker is reduced, and the practicability is improved; the supporting plate can be drawn forwards and backwards; when the supporting plate moves the foremost part, the containing base moves to the position in front of the base, the tank body is conveniently contained, use is more convenient, the distance between the air exhaust assembly and the adjusting assembly can be adjusted, and the application range is wider.

The invention discloses a pedestrian re-identification method and device based on space-time analysis and depth features. In pedestrian re-identification application, global search for pedestrian images in an actual large-scale video monitoring scene has complexity and irrationality, and in order to further improve the identification accuracy and the identification speed, the invention provides amethod combining spatio-temporal information analysis and depth feature extraction. Firstly, the moving speed of a pedestrian is obtained through analysis to accord with gamma distribution, and then the space-time information of the pedestrian is further analyzed through the distribution to obtain the space-time prior probability of the pedestrian; training a convolutional neural network on the large-scale data set in combination with an actually acquired image, and extracting depth features to calculate a visual space-time probability; and finally, combining the two probabilities to judge whether the two images are the same pedestrian or not. The pedestrian re-identification efficiency can be effectively improved from massive monitoring or collected data in an actual large-scale video monitoring application scene, the high pedestrian re-identification precision can be kept, and the efficient and accurate pedestrian re-identification effect is achieved.

The invention relates to a method for checking the causes of road noise structural sound under a uniform speed condition, and provides an idea and a method for checking the causes of road noise structural sound in a vehicle. The scope of checking can be reduced by fast diagnosis of the frequency characteristics of road noise structural sound occurrence, thereby improving the work efficiency compared with checking one by one based on experience. Through a series of testing and analysis work, the transmission path of abnormal vibration can be identified more accurately; the key elements of the transmission path can be locked; and economic and effective rationalization improvement suggestions are given according to the actual situation.

The invention discloses a video monitoring system for a pedestrian crosswalk. The system comprises a real-time image collection unit which comprises two first network cameras and two second network cameras and outputs photographed image information; a display recording unit which comprises a video display and a video storage server, wherein the video display and the video storage server are in communication connection with the real-time image collection unit; an authentication server unit which comprises a first server and a second server which are in communication connection, wherein the first server receives the image information output by the first network cameras, identifies human face information and carries out deduplication, and the second server pre-stores a human face information library of wanted persons of a public security system, judges the human face information processed by the first server and outputs an alarm instruction; and an alarm unit which receives the alarm instruction output by the second server and sends alarm sounds. According to the system, the existing video monitoring system for the pedestrian crosswalk is improved, advanced alarm and early actions can be realized through application of intelligent alarm, and the working efficiency of the system is improved.

The invention discloses a positioning method and device for signal leakage antenna nodes. The method comprises the steps of identifying a signal leakage cell through utilization of measurement report(MR) data; judging a signal leakage area in the signal leakage cell according to bidirectional two-dimensional data of the signal leakage cell and surrounding cells of the signal leakage cell; obtaining basic information of the signal leakage area; calculating signal leakage volume of each antenna node in the signal leakage area based on the basic information; and positioning the signal leakage antenna nodes according to the signal leakage volume of each antenna node. According to the embodiment of the invention, the signal leakage area in the signal leakage cell is judged, the signal leakagevolume of each antenna node in the signal leakage area is calculated, and the signal leakage antenna nodes are positioned according to the signal leakage volume of each antenna node. Operation is simple and convenient. Indoor distribution signal leakage processing efficiency is improved.

The invention discloses a vehicle real-time image tracking method based on GPS trajectory data, comprising the following steps: estimating positions of vehicles at a future time point by using historical GPS trajectory data, pre-selecting monitoring points where vehicles may possibly be captured according to the estimated position information to narrow the scope of investigation and prepare for capture in advance so as to enhance the capture effect, detecting the images of the monitoring points by a deep convolutional neural network model with high real-time, filtering the vehicle-free monitoring points, and storing the images taken by the monitoring points with vehicles is, thereby realizing the no-delay tracking of images.

The invention relates to a method and device for automatic detection of boundary line breakage and an intelligent mower. The method includes the following steps that a detection signal sent by a boundary line is received; if no detection signal is received, whether a cutting tool disc of the intelligent mower works or not is judged, and the intelligent mower is controlled to stop; if it is determined that the cutting tool disc of the intelligent mower works when no detection signal is received, whether the boundary line is broken or not is judged; if it is determined that the boundary line isbroken, a prompting message about the boundary line breakage is sent to a client. According to the technical scheme, when the detection signal sent by the boundary line is not received, whether the boundary line is cut off by the cutting tool disc of the intelligent mower or not is judged, meanwhile the intelligent mower is controlled to stop, so that the intelligent mower stops near the broken boundary line, and a user is prompted to only check around the intelligent mower, the check range of the broken boundary line is largely reduced, user's workload in checking a breaking point of the boundary line is reduced, and efficiency of locating the breakpoint of the boundary line is greatly improved.

The invention discloses an automatic fault management method, device and system. The method comprises the following steps of: firstly, identifying a system fault according to received fault prompt information, and triggering a fault work order of a corresponding dimension; generating a plurality of troubleshooting tasks in the corresponding dimension according to the fault work order; executing the plurality of troubleshooting tasks in parallel to locate a fault point; and finally, searching a recovery plan matched with the fault point in a preset recovery plan matching relationship, executingthe recovery plan to repair the system fault. According to the method, full-process automatic fault processing is realized, the influence of man-made subjective factors is avoided, and fault identification, positioning and repair processes are more accurate and efficient; and a parallel fault troubleshooting mode is adopted, so that the troubleshooting time can be further shortened, the troubleshooting comprehensiveness is improved, and the fault management efficiency is improved.

The invention discloses a characterization method and device for a low-efficiency water injection belt of an ultra-high water-cut period oil reservoir. The method comprises the steps: obtaining production dynamic data of an oil-water well, and screening out a well group of which the probability of containing a low-efficiency water injection belt reaches a first preset value according to the production dynamic data; calculating the stage water storage rate of each small layer in the well group with the probability of containing the low-efficiency water injection belt reaching the first preset value, and screening out a horizon of which the existence probability of the low-efficiency water injection belt reaches a second preset value according to the stage water storage rates; carrying out single-layer analysis and calculation on each horizon with the existence probability of the low-efficiency water injection belt reaching the second preset value, and determining the horizon where the low-efficiency water injection belt is located; and based on a grid system, performing analysis and calculation on grids in the horizon where the low-efficiency water injection belt is located, and determining low-efficiency water injection grids and the development level of the low-efficiency water injection grids. The method can determine the horizons, the development position and the developmentlevel of the low-efficiency water injection belt, and improves the representation effect of the low-efficiency water injection belt.

The invention discloses an atmospheric environment parameter monitoring method. The method comprises the steps of grid division, point distribution, monitoring processing and pollution source tracing.A lifting platform is matched with a driving mechanism for use, wherein the height of a monitoring probe can be adjusted according to practical requirements so as to monitor the atmosphere at different heights. The driving mechanism is matched with a protection module for use; when the wind speed is higher than the triggering threshold value; start driving mechanism, the lifting platform descendsto the lowest end; the height of the monitoring station is automatically reduced; the monitoring station is prevented from turning over due to overhigh wind speed; improving stability of a device, two adjacent monitoring stations are associated; establishing associated groups, presetting a phase difference threshold, by comparing the difference monitored by the associated groups in real time withthe difference threshold, when the pollution source occurs in the monitoring area, the abnormal associated groups are recorded, and the plurality of abnormal groups are connected, so that the pollution source area can be positioned and obtained, an administrator can lock the pollution source at the first time and quickly find the pollution source, and the working efficiency is improved.

The invention discloses an electric power system power distribution protection device, an application and a method. The electric power system power distribution protection device comprises a power distribution cabinet body, a travel switch, three electromagnetic relays, a signal lamp and a controlled switch, wherein the three electromagnetic relays act on the controlled switch through the signal lamp, electromagnetic coil circuits of two of the electromagnetic relays are mutually controlled and respectively act in normal power supply and temporary maintenance, the other electromagnetic relay performs delay timing, assists standby application and is also controlled by the other two electromagnetic relays, and the three electromagnetic relays form three control circuits to act on the controlled switch. The electric power system power distribution protection device can perform operations such as maintenance, transformation, switching and periodic inspection on power distribution equipment in a power distribution cabinet or box without interrupting power supply; when the power distribution cabinet or box is illegally opened by a law breaker, power supply for the system is interrupted if a special key is still not inserted within a set time, and the power supply is recovered only when the special key is inserted.

The invention discloses a control method and device for preventing stealthy discharge of pollution sources, wherein the method includes: collecting working condition data of production equipment and pollutant treatment equipment, and collecting environmental protection data; according to the working condition data of production equipment, The working condition data of the pollutant treatment equipment and the data of the sewage concentration pool determine the theoretical total amount of pollutants discharged per unit time; determine the actual total amount of pollutants discharged per unit time according to the discharge outlet data; judge the theoretical total amount of pollutants discharged per unit time and Whether the difference between the actual total amount of discharged pollutants per unit time exceeds the preset range, and when the difference exceeds the preset range, it is determined that there is a sneak discharge behavior. This method can judge whether polluting enterprises have closed down pollutant treatment equipment in violation of regulations, and whether there is illegal discharge behavior, which provides a reliable basis for local environmental law enforcement, helps administrative law enforcement personnel reduce the scope of investigation, and reduces the work intensity of personnel.