47results about How to "Reliable reflection" patented technology

The invention relates to a method for selecting single-phase ground fault lines of power distribution networks on the basis of convolutional neural networks. The method includes acquiring zero-sequence voltages of busbars and zero-sequence current signals of various feed lines; carrying out continuous wavelet transformation on the various zero-sequence current signals according to set decomposition scales; acquiring time-scale wavelet coefficient gray-level graphs; identifying fault feed lines by the aid of trained convolutional neural network algorithms. The method for selecting the single-phase ground fault lines of the power distribution networks on the basis of the convolutional neural networks has the advantage that the fault feed lines and the non-fault feed lines can be accurately identified by the aid of the method under various fault working conditions when single-phase ground faults occur.

The invention discloses an automatic detection method for a large transfusion production line, including the following steps: 1, making a transfusion bottle filled with liquid medicine rotate in a high-speed, then rapidly stopping the transfusion bottle, so that the liquid medicine in the bottle has a self-rotating movement under the inertia; 2, picking-up a plurality of images of the transfusion bottle in which the liquid medicine is in a self-rotating state; 3, respectively digitally processing and mutually comparing to the plurality of images, distinguishing the exogenous impurity in the liquid medicine, in order to judge whether the bottle of liquid medicine is qualified. The invention also discloses an automatic detection apparatus for a large transfusion production line, which is characterized by including a rotation device, a controller, a video camera and an image processing apparatus, wherein, the output end of the controller is connected with the control terminal of the rotation device, a gripping device for fixing up the transfusion bottle, a rotating work station and a detecting work station are provided on the rotation device, and the video camera outputs data to the image processing apparatus. The invention can automatically detect whether impurity exists in the liquid medicine, with high testing accuracy and fast testing velocity.

The invention provides an infrared oven and GC-MS based method and an infrared oven and GC-MS based device for on-line analysis of simulated combustion of cigarettes and is characterized in that the method comprises: a cigarette sample is put under a condition of inert gas or synthetic air to realize the control of the oxygen content in the cracking atmosphere by using a gas flowmeter; a focused infrared oven is used as a heating device for cracking cigarette, and under the control of a programmable temperature rise controller, the infrared oven can control the temperature change in the simulating process, realize quick temperature rise cracking of the gram-class cigarette samples and simulate temperature rise conditions of the combustion of cigarettes; the cracking products obtained at different time periods can be selectively delivered to a PTV sample injection opening by a six-way valve and condensed and enriched by liquid nitrogen, the enriched cracking products are separated by the gas chromatography GC and then delivered to the mass spectrometry for quantitative and qualitative analysis; and by simulating the conditions of the combustion of cigarettes, the cracking experiments of the gram-class cigarette samples at different temperature-rise speeds and with different oxygen contents are carried out.

The invention provides a method for detecting human body physiological parameters on basis of an infrared sequence image. By the method, non-contact measurement on heart rate and breath is realized. The invention adopts the detailed technical scheme which comprises the following steps of: 1, building an infrared dynamic image acquisition system which consists of a forward looking infrared radar (FLIR) infrared long-wave sensor, a daheng image acquisition card and a video storage working station; 2, acquiring the infrared sequence image of a tested man by using the infrared dynamic image acquisition system, and filtering irregular movement of a head of the sequence image; 3, selecting temples as interesting areas, and extracting time sequence signals from the interesting areas by using a gravity method; 4, performing empirical mode decomposition on the extracted time sequence signals; and 5, performing spectral analysis on the result of the empirical mode decomposition, and identifying breath and heart rate signals. By the method, the non-contact measurement on heart rate and breath is realized by using the infrared sequence image. Therefore, the method has better application value in physiological feature monitoring in human face feature information identification and clinical surgery.

A method for preventing low-voltage protection misoperation of a motor caused by PT disconnection belongs to an emergency protection field. Disconnection faults of a first side and a second side of PT are monitored in real time simultaneously by adopting the method of monitoring the open-delta voltage of PT, monitoring whether high-voltage fuse fuses and monitoring the voltage of the second side of PT. Once it is discovered that disconnection faults occur at the first side and the second side of PT, a latching control signal is outputted immediately. Low-voltage protection output loops of each motor on the same busbar are disconnected and latching protection is carried out. At the same time a second loop of a small busbar of the voltage is disconnected and self-locking of the latching control loop of PT disconnection is carried out. The technology scheme can reflect PT secondary disconnection faults more reliably and directly and the latching mode is safer. No adoptive equipment obtain electricity in normal operation, and the method is in favor of long-term reliable operation of equipment and can prevent low-voltage protection misoperation of motor effectively.

The invention relates to a non-contact human respiration and heart beat signal detection method based on an infrared sequence image. According to the technical scheme, the method comprises the followings steps: 1, in an appropriate environment, collecting the infrared sequence image on the side of a subject, and in the process, controlling the moving of a head of the subject; 2, using an Ostu method to realize binaryzation of the image, extracting a time series signal at a barycenter (one step interval) of the sequence image, and obtaining a respiration signal of the subject; 3, selecting a superficial temporal artery upstream region (the temple area) and an area around the nostrils as interesting regions. The mean of the area around the nostrils is calculated to obtain the respiration signal of the subject; the time series signal of the temple interesting region needs to be decomposed through an integrated empirical mode, and then is subjected to AR power spectrum analysis so as to detect the respiration and heart beat signals. According to the method, the non-contact heart beat and respiration detection is realized by using the infrared sequence image, and the method has better application value in the identification of facial physiological information and monitoring of clinical physiological characteristics during an operation.

The invention discloses a complex network node classification method based on a graph neural network so as to solve a difficult problem of complex electromechanical system coupling network community division. The method comprises the steps: applying detrending coupling correlation analysis to the calculation of the correlation between monitoring variable nodes, and carrying out the primary screening of correlation coefficients through the introduction of Gaussian noise; introducing a scale index, and carrying out secondary screening on the correlation coefficients; taking the monitoring variable as a network node, converting the correlation coefficients into edge connection weights, and constructing an undirected weighted complex network; starting from a static community detection algorithm of global modularity optimization based on module gain, taking each node in the network as a partition, calculating modularity gain of a neighbor node to the current community according to a modularity function, judging community attribution of the node according to the modularity gain, and obtaining primary partition of the network node; taking the community primarily divided by the network asa node again, carrying out a new round of iteration on a new network, obtaining an optimal community division result of the network when the modularity reaches a maximum value, and taking the result as an initial training label of the graph attention neural network; carrying out training based on real-time monitoring data through a graph attention neural network, realizing node classification of the complex network, thereby providing a reliable basis for accurate description of the complex electromechanical system coupling network.

The invention relates to an online monitoring device for a tunnel boring machine cutter. The device which is characterized by comprising a backstage controller, a cloud server, a 6LoWPAN protocol edge router and a plurality of cutter monitoring controllers adhered to a cutter ring of the cutter. The backstage controller is a large-screen microcomputer or a tablet personal computer, and the 6LoWPAN protocol edge router is connected with the backstage controller and the cloud server through an IPv4 protocol; each cutter monitoring controller comprises a 6LoWPAN protocol controller which is connected with a cutter abrasion loss measuring circuit, a temperature measuring circuit and a vibration and rotation measuring circuit respectively. The online monitoring device achieves online monitoring on abrasion, rotation speed, temperature and vibration parameter of the cutter, can accurately grasp the cutter condition at any time, prolongs service time and the optimal replacement time of the cutter and further improves work efficiency and reduces engineering cost. The invention further provides a monitoring method of the online monitoring device. The monitoring method has the advantages of reasonable technological process, strong adaptability and excellent effect.

The invention relates to a rock mass drilling and shearing elastic modulus instrument which mainly comprises a stopper rod port (1), a sleeve rod (2), an upper disc knife (3), a rubber bag (4) and a lower disc knife (5); and the whole structure of the rock mass drilling and shearing elastic modulus instrument is shaped like a long cylinder body. A hole wall rock mass shear test is carried out on different drilling positions under different positive stresses, and a destructive curve can be drawn by a rock mass shear destruction test structure under the different positive stresses, so that rock mass shearing-resistant mild parameters can be confirmed. The rock mass drilling and shearing elastic modulus instrument is simple in structure, low in manufacturing cost, easy to use and simple and easy to maintain and repair.

The invention discloses a method and system for dynamically adjusting video call code streams. The method comprises that a first client takes the timing value of a first timer as first time when a video is coded; when the first client sends video data to a second client, the first time is sent to the second client at the same time; the second client decodes and plays the received video data and takes the timing value of a second timer as second time when the video data is decoded and played; the second client obtains the difference value between the first time and the second time and sends the difference value to the first client in real time; the first client obtains a current call performance effect through comparing the difference value with a preset threshold value; and the first client dynamically adjusts the coded video code streams according to the current call performance effect. According to the method and the system, through judging the magnitude of the video image delay, the video call code streams are dynamically and self-adaptively adjusted, and therefore, abnormal phenomena such as image delay and packet loss can be reduced or avoided.

The invention relates to a pumping high-impedance protection configuration method of pumping winding delta-type wiring. The configuration method comprises: differential protection, zero-sequence differential protection, and inter-turn protection are configured to be main protection of a main electric reactor; master reactor over-current protection, overload protection and zero-sequence over-current protection are configured to be backup protection of the main electric reactor; neutral-point reactor over-current and overload protection are configured to be protection of a neutral-point reactor;and pumping zero-sequence over-current protection is configured to be inter-turn protection of a pumping winding and pumping turning re-pressing over-current protection is configured to be backup protection of a pumping winding. According to the pumping high-impedance protection configuration method, while the main reactor is protected, the inter-turn short-circuit fault and phase-to-phase shortcircuit fault of the pumping winding can be reflected sensitively and reliably. Therefore, a technical problem that the conventional protection configuration can not reflect the inter-turn short-circuit fault sensitively of pumping when the pumping winding is in a triangular wiring mode can be solved; the action characteristic of protection is improved; and the pumping high-impedance protection configuration method has the great engineering application value.

The invention discloses a detecting method of abdominal force, which comprises the following steps: setting slippery disc displacement quantity and gathering time; timing through single-chip computer; gathering displacement and pressure through temperature sensor; predisposing circuit to obtain analog signal; obtaining digital signal through module switcher; disposing through single-chip computer to produce corresponding value; obtaining pressure value to reserve in the memory every certain equal time; reducing one second when the time is over the set time; alarming when the time is less than gathering time and real slip disc displacement equals the set displacement; displaying on the screen; printing time pressure trend graph through USB interface printer; setting the diameter of contact area of slip rod at 18mm-26mm with 22mm at best; setting the height of probe sleeve not less than the transversal surface of maximum circular.

The invention discloses a software platform applying a CART algorithm to realize electric energy meter detection data management. The invention relates to the technical field of electric energy metering detection. A cloud big data management platform is adopted, and information in mass data is fully mined to realize big data analysis, mining and processing, a cloud computing technology is adoptedto process hundreds of millions of data types within a few seconds, the classification capability and calculation precision of different data are improved through a CART algorithm and an improved CARTalgorithm, the construction of a distributed hierarchical model of big data can be realized through the CART algorithm, and the data identification capability is simplified. According to the invention, a data fusion algorithm is adopted to reflect various data of the electric energy meter with diversified information representation forms, huge information quantity and complicated information relationship timely, accurately and reliably, through a BP neural network algorithm model, the error of a direct leading layer of an output layer is estimated by using an output error, and the learning algorithm precision is improved.

The invention discloses a soaking load test method of a strip-shaped foundation model. The method comprises the steps of: (1) construction of the strip-shaped foundation model; (2) construction of a load platform; (3) load loading; (4) soaking of the strip-shaped foundation model; and (5) load unloading. The method is simple in steps, reasonable in design, convenient to realize and excellent in use effect; the strip-shaped foundation model is constructed in a constructed strip-shaped foundation research field; meanwhile, the installation of a monitoring element is finished to guarantee the foundation settlement standard, the ground surface settlement standard and accurate installation of a water moisture meter; and under load loading and soaking tests, the settlement of a strip-shaped foundation, the settlement of the ground surface and the volume water content of a collapsible soil layer can be accurately detected to provide reliable basis for construction of the strip-shaped foundation to be constructed.

The invention provides a method for monitoring and evaluating a river by using unmanned aerial vehicle surveying and mapping. The method comprises the following steps: acquiring image data by using anunmanned aerial vehicle; reconstructing the image data by using a motion recovery shape algorithm to construct 3D topographic information; utilizing a Vextractor program to speculate the basic shapeof the river streamline on the 3D topographic information according to the image data, and fitting the basic shape into a river curvature index of the main river channel section; calculating an ortho-image spectral index by using an RGB ortho-image measurement program to obtain river quantitative and qualitative information; and evaluating the physical environment of the river according to the curvature index and the quantitative and qualitative information of the river. The invention provides a river monitoring and evaluating method. The unmannedaerial vehicle is taken as the infrastructure,a motion recovery shape algorithm, a Vextracto program and an RGB ortho-image measurement program are utilized to track the eutrophication process in the large river and reflect the change process ofeutrophication and vegetation overgrowth, and the method has very important application value for long-term monitoring and ecological restoration of the large river.

The invention discloses a device for determining the punching shape of a coal seam and a using method thereof. The device comprises a telescopic rod, wherein a hollow rail rod is connected to the telescopic rod, and a ranging box is nested on the rail rod; the ranging box is internally provided with a first gear slidably connected with the rail rod, a fixed rod is arranged on the first gear, one end of the fixed rod penetrates into the rail rod, and a first micro motor is fixed to the end, penetrating into the rail rod, of the fixed rod; an output shaft of the first micro motor is provided with a second gear, and the rail rod is internally provided with a strip-shaped tooth plate meshing with the second gear; a second micro motor is fixed into the ranging box, and an output shaft of the second micro motor is provided with a third gear meshing with the first gear; a limiting rod is fixed to the first gear, and the two ends of the limiting rod abut against the inner walls of the two sides of the ranging box; and a laser rangefinder is arranged in the ranging box. According to the device for determining the punching shape of the coal seam and the using method thereof, the measurementof the diameter and depth of a cavity-forming hole can be realized effectively and accurately, the form of the hole is restored, and meanwhile, the device is easy to operate, accurate and reliable andhigh in adaptability.

The invention relates to a method for controlling a lignin-phenolic resin adhesive reaction end point. The method comprises the following steps of: sampling and performing acidification and detection, wherein the acidification is to add 2 to 5mol / L acid solution into the taken sample in a stirring state until cakes are generated; the detection is to detect the cakes; and when the cakes are brown,can be lengthened by stretching after being taken out but are easy to break, and a fracture surface has brown gloss, the end point is achieved. The control method is not limited by the proportion that phenol is substituted by the lignin, is not influenced by the polymerization reaction process of the lignin, the phenol and formaldehyde, and simply, fast, conveniently, visually, accurately and reliably reflects the end point of the lignin-phenolic resin adhesive reaction, so that a reliable, economical and convenient detection method is provided for the industrialized production of a lignin-based phenolic resin adhesive.

The invention discloses a reagent for miRNA sequencing quantitative result evaluation of a cancer sample and a preparation method and application thereof. The reagent comprises at least one of RNAs ofsequences shown in SEQ ID NO.1-SEQ ID NO.8. The reagent is quantitatively added into a to-be-tested RNA sample, the amount of the added reagent is quantitatively analyzed through sequencing, and a quantitative analysis result is compared with an actual added amount, thereby evaluating the accuracy and reliability of an entire miRNA sequencing quantitative result. The reagent for the miRNA sequencing quantitative result evaluation of the cancer sample and the preparation method and application thereof have the advantages that the method is simple and used easily, and the reliability of the miRNA sequencing quantitative result can be effectively evaluated; characteristics of the reagent in all aspects are very close to a real miRNA, so that the reagent can more truly, accurately and reliably reflect actual situations of a human miRNA sequencing quantitative result, thereby making more accurate and reliable evaluation to miRNA sequencing quantitative result of the cancer sample.

The invention discloses a safety evaluation method for a tunnel anchorage. Long-term shear strength parameters of surrounding rock of the tunnel anchorage are used as a safety evaluation basis for thetunnel anchorage; the long-term shear strength parameters are measured according to the following steps: simulating engineering site surrounding rock, and preparing a tunnel anchor surrounding rock model test piece; establishing a shear strength relaxation curve of the test piece under different vertical loads through a constant-load consolidation drainage shear test; converting the shear strength relaxation curve into an lgt1 / [tau] curve, and establishing a long-term strength equation of the shear strength of the test piece along with time change under different vertical loads; and selectinga required long-term strength equation under a vertical load, and determining a long-term shear strength parameter of the test piece, namely the long-term shear strength parameter of the tunnel anchor surrounding rock. According to the invention, the long-term shear strength parameter is adopted, so that the mechanical property of the rock mass under continuous shear load can be more reliably reflected, and the deviation between a test result and engineering practice is greatly reduced.

The invention relates to a real-time online monitoring method for a protection pressing plate. The method comprises the following steps that: S1: the ID (Identity) number of a monitoring terminal andthe protection pressing plate are associatively matched, the mentoring terminal is installed just below the connecting sheet of the pressing plate, and the monitoring terminal collects whether the protection pressing plate is positioned on a proper position or not through a non-contact method; and S2: a monitoring result is wirelessly transmitted to a collector by the monitoring terminal, the collector transmits a collected monitoring terminal signal to a regulation and control master station through the network, and the regulation and control master station analyzes signals transmitted by allcollectors, and provides the signals for a human-computer interaction interface. By use of the method, heavy losses caused since a real-time monitoring alarm can not be given after a manual operationhas errors can be avoided, the management efficiency of pressing plate state information is effectively improved, and the method is suitable for the unattended requirements of a transformer substation. Working efficiency is greatly improved, and the requirement of an operation and maintenance "2+N" working mode can be met.

The invention relates to an MTO reaction kinetic model, an MTO reaction regeneration integrated model and an application thereof. The construction method of the MTO reaction kinetic model comprises the following steps: dividing a reaction system into a plurality of lumped components; determining a reaction network and a reaction kinetic equation according to a reaction mechanism among the lumped components, wherein the lumped component comprises methanol, methane, ethane, ethylene, propane, propylene, C4, C5 and above, hydrogen, carbon monoxide, carbon dioxide, and coke. The MTO reaction regeneration integrated model comprises a reactor model, a regenerator model and a catalyst circulating material balance model. The model can reflect the reaction process of the MTO reaction more fully, truly, accurately and reliably; the MTO reaction regeneration integrated model fills up the blank that the existing MTO lumped model lacks application of a regenerator and a catalyst model in actual industrial production, and provides a comprehensive model for whole-process monitoring and optimization of MTO reaction.

The embodiment of the invention provides a road section position matching method and device, a navigation method and device and a storage medium. The road section position matching method comprises the following steps: executing map block-based road section exploration on a standard precise map and a high-precision map of a target geographic area, and correspondingly obtaining respective intersection data of the standard precise map and the high-precision map; according to the intersection data of the standard-precision map and the intersection data of the high-precision map, the road section connection relation of the standard-precision map and the road section connection relation of the high-precision map are determined; performing road section exploration according to respective road section connection relations of the standard-precision map and the high-precision map, and determining respective road sections between intersections of the standard-precision map and the high-precision map; the intersections in the standard precision map are matched with the intersections in the high-precision map, and the corresponding relation of the road sections is determined; and matching the position of the road section between the intersections of the standard precise map with the position of the road section between the intersections of the high-precision map to obtain a position matching relationship between the two positions. The scheme of the embodiment of the invention is more beneficial to accurately describing the actual geographic position.

The invention relates to a method for selecting single-phase ground fault lines of power distribution networks on the basis of convolutional neural networks. The method includes acquiring zero-sequence voltages of busbars and zero-sequence current signals of various feed lines; carrying out continuous wavelet transformation on the various zero-sequence current signals according to set decomposition scales; acquiring time-scale wavelet coefficient gray-level graphs; identifying fault feed lines by the aid of trained convolutional neural network algorithms. The method for selecting the single-phase ground fault lines of the power distribution networks on the basis of the convolutional neural networks has the advantage that the fault feed lines and the non-fault feed lines can be accurately identified by the aid of the method under various fault working conditions when single-phase ground faults occur.

The invention provides a lamp for vehicle, capable of reliably leading the light from the light source to the side of the vehicle. the lamp comprises a first reflector (6) and a second reflector (7). The light from the light source (5) irradiates from the front to the side of the vehicle under the action of the first reflector (6) to form a first light-distributing pattern (P1), a opening ( 16) ia arranged on the first reflector (6). The light from the light source (5) going through the opening (16) of the first reflector is reflected to the side of the vehicle (C) by the second reflector (7) to form a second light-distributing pattern (P2). As the result, the invention can reliably lead the light from light source (5) to the side of the vehicle (C).

The invention relates to an MTO reaction kinetics model, an MTO reaction regeneration integration model and applications thereof. The construction method of the MTO reaction kinetic model comprises: dividing the reaction system into a plurality of lumped components; according to the reaction mechanism between the lumped components, determining the reaction network and the reaction kinetic equation; the lumped components include: Methanol; Methane; Ethane; Ethylene; Propane; Propylene; Carbon 4; Carbon 5 and above; Hydrogen; Carbon Monoxide; Carbon Dioxide; Coke. The MTO reaction regeneration integrated model includes a reactor model, a regenerator model and a catalyst circulation material balance model. The above model can reflect the reaction process of the MTO reaction more fully, more realistically, more accurately and more reliably; the integrated model of the MTO reaction regeneration fills the gap in the existing MTO lumped model that lacks the application of the regenerator and catalyst models in actual industrial production. Blank, providing a comprehensive model for the whole process monitoring and optimization of MTO reactions.

A method for preventing low-voltage protection misoperation of a motor caused by PT disconnection belongs to an emergency protection field. Disconnection faults of a first side and a second side of PT are monitored in real time simultaneously by adopting the method of monitoring the open-delta voltage of PT, monitoring whether high-voltage fuse fuses and monitoring the voltage of the second side of PT. Once it is discovered that disconnection faults occur at the first side and the second side of PT, a latching control signal is outputted immediately. Low-voltage protection output loops of each motor on the same busbar are disconnected and latching protection is carried out. At the same time a second loop of a small busbar of the voltage is disconnected and self-locking of the latching control loop of PT disconnection is carried out. The technology scheme can reflect PT secondary disconnection faults more reliably and directly and the latching mode is safer. No adoptive equipment obtain electricity in normal operation, and the method is in favor of long-term reliable operation of equipment and can prevent low-voltage protection misoperation of motor effectively.

The invention belongs to a control method for preparing reaction end points of (S)-(+)-epichlorohydrin by chemically splitting racemic epichlorohydrin, and belongs to the field of control of the reaction end points. The method comprises the following steps of in the presence of catalysts, enabling the dl-epichlorohydrin and glacial acetic acid to be subjected to an activating reaction under the condition that air is led, so that reaction liquid is obtained; and controlling the temperature of the reaction liquid to be 20-25 DEG C, slowly and dropwise adding water for a hydolysis reaction, after the water is dropwise added for the first time, performing a heat insulating reaction, then slowly and dropwise adding water for the second time, performing the heat insulating reaction under 20-25 DEG C, and when the hydolysis reaction is performed for 38 hours, detecting the special gravity equivalent of the reaction liquid under the condition of constant temperature of 20 DEG C, wherein detection is performed once every 31 minutes, and when the special gravity equivalent is within the range of 1.095-1.100g / cm<3>, the reaction end points are obtained. The method has the advantages of being convenient and quick in detection, low in cost, reliable in detection result and the like.

The invention discloses a fan blade icing detection method, device and equipment and a computer storage medium. The draught fan blade icing detection method comprises the steps that P images are obtained, the images are obtained by shooting draught fan blades, and P is a positive integer; respectively inputting the P images into a blade detection model, a blurred image detection model and an icing detection model which are obtained through pre-training to obtain P blade detection results, P blurred image detection results and P first icing detection results; and inputting the P blade detection results, the P blurred image detection results and the P first icing detection results into a pre-trained classification model to obtain a second icing detection result. In the embodiment of the invention, the second icing detection result can more reliably reflect whether the fan blade is iced or not, and the accuracy of icing detection of the fan blade is improved.

The invention relates to a fault mode influence and hazard analysis method, computer equipment and a storage medium. The fault mode influence and hazard analysis method comprises the steps of obtaining a first decision matrix of each risk factor of each fault mode; and obtaining a risk weight vector of each risk factor; according to an intuitionistic binary language utility weighting integration operator and the risk weight vector of each risk factor, integrating intuitionistic binary languages of n risk factors of each fault mode in the first decision matrix into a comprehensive binary language of each fault mode; and according to the comprehensive binary language of each fault mode, obtaining a risk level sequence of each fault mode. The evaluation accuracy can be effectively improved.

The invention discloses an automatic detection method for a large transfusion production line, including the following steps: 1, making a transfusion bottle filled with liquid medicine rotate in a high-speed, then rapidly stopping the transfusion bottle, so that the liquid medicine in the bottle has a self-rotating movement under the inertia; 2, picking-up a plurality of images of the transfusionbottle in which the liquid medicine is in a self-rotating state; 3, respectively digitally processing and mutually comparing to the plurality of images, distinguishing the exogenous impurity in the liquid medicine, in order to judge whether the bottle of liquid medicine is qualified. The invention also discloses an automatic detection apparatus for a large transfusion production line, which is characterized by including a rotation device, a controller, a video camera and an image processing apparatus, wherein, the output end of the controller is connected with the control terminal of the rotation device, a gripping device for fixing up the transfusion bottle, a rotating work station and a detecting work station are provided on the rotation device, and the video camera outputs data to the image processing apparatus. The invention can automatically detect whether impurity exists in the liquid medicine, with high testing accuracy and fast testing velocity.