CN113740778A - Fault diagnosis device and method for 500kV high-power transformer for power plant - Google Patents

Abstract

The invention discloses a device and a method for diagnosing faults of a 500kV high-power transformer for a power plant, and relates to the technical field of transformer fault diagnosis. The invention has the advantages of simple instrument structure and operation, high detection accuracy, simple structure, small operation and maintenance workload of operation and maintenance personnel, great reduction of maintenance difficulty and long service life.

Description

Fault diagnosis device and method for 500kV high-power transformer for power plant

Technical Field

The invention relates to the technical field of transformer fault diagnosis, in particular to a 500kV high-power transformer fault diagnosis device for a power plant.

Background

The transformer is one of the most important devices in a power transmission and transformation system, the transformer is used as an important power transformation device of a power system and is responsible for voltage transformation and electric energy transmission tasks, the running state of the transformer directly influences the reliability of power supply and the normal running of the whole system, fault monitoring and diagnosis of the transformer are always scientific research projects which are emphasized by power system departments at home and abroad, and particularly, the monitoring of the running of the transformer and judgment of latent faults become important means for ensuring the safe running of the transformer by monitoring the content of dissolved characteristic gas in oil.

(1) The method is characterized in that a gas sensor with the characteristics of a designed transformer is used for measuring in a gas phase and is not directly detected in a transformer oil tank, so that gas is required to be separated from an oil phase before detection so as to perform subsequent data analysis and processing, the gas-oil separation is one of the most basic and important components in a system, common methods comprise a mechanical oscillation method, a vacuum method and the like, and the mechanical oscillation method and the vacuum method are complex in instrument construction and operation.

(2) Because a chromatographic column is not adopted to separate gas, the gas in the gas chamber is mixed gas, the development level of the gas sensor does not meet the requirements of people, and the stability and the selectivity of the gas sensor are insufficient, interference and cross sensitivity exist among gas components when the mixed gas is detected, so that the accuracy of gas concentration detection of the gas chamber is low.

(3) Generally, a gas sensor is sensitive to other gases besides a measured gas, namely, is sensitive to cross, so that measurement of mixed gas is prone to cross interference and measurement results are inaccurate.

Disclosure of Invention

Technical problem to be solved

Aiming at the defects of the prior art, the invention provides a device and a method for diagnosing faults of a 500kV high-power transformer for a power plant, and solves the problems of great maintenance difficulty and short service life caused by complicated instrument framework and operation, low detection accuracy, complicated structure and large operation and maintenance workload of operation and maintenance personnel.

(II) technical scheme

In order to achieve the purpose, the invention is realized by the following technical scheme: a500 kV high-power transformer fault diagnosis device for a power plant comprises a permeable membrane tube, wherein an oil-gas separation membrane is arranged in the permeable membrane tube, one side of the permeable membrane tube is connected with an oil outlet pipe orifice of a transformer oil tank, an isolation valve is arranged between the oil outlet pipe orifice and the permeable membrane tube, the other side of the permeable membrane tube is connected with an air chamber for storing gas, and the isolation valve is usually in an open position so that the permeable membrane is usually in a gas permeable state; gas permeation in oil is a diffusion process, gas molecules are diffused from the oil to one side of a gas chamber due to gas pressure difference on two sides of a membrane, and the gas pressure on the two sides of the membrane tends to be balanced after a certain time at a certain temperature to achieve dynamic balance, so that oil-gas separation is automatically realized; the gas chamber is internally provided with an intelligent sensor array consisting of an artificial neural network system and a gas sensor array, the artificial neural network system can realize complex nonlinear mapping, after the artificial neural network system is used for processing the information of the gas sensor array, the problems of serious nonlinearity and the like caused by cross sensitivity can be better solved, the artificial neural network system has good fault-tolerant performance on sensor drift or noise, and the detection precision of gas can be improved.

Preferably, the transformer box further comprises a display and control device, the display and control device is installed on a door plate on one side of the transformer box body, the display and control device comprises an LED display, a loudspeaker and a control panel, the bottoms of the LED display and the loudspeaker are fixed to the top of the control panel, an LED display screen is installed on the upper side of the front face of the LED display, three display lamps are installed on the lower side of the front face of the LED display, a keyboard is arranged on the control panel, a cover plate is connected to the control panel, the bottom of the cover plate is hinged to the bottom of the control panel through a pin shaft, a clamping block is fixed to the top of the cover plate, a clamping groove corresponding to the clamping block in position is formed in the bottom of the LED display, and the clamping block is matched with the clamping groove; the LED display screen and the display lamp are electrically connected with the LED display, the input ends of the LED display and the loudspeaker are electrically connected with the output end of the control panel, and the keyboard is electrically connected with the control panel.

Preferably, the inlet end of the isolation valve is connected with the oil outlet pipe end of the transformer oil tank in a sealing mode through a first flange, the outlet end of the isolation valve is connected with the inlet end of the permeable membrane pipe through a second sealing flange, and the outlet end of the permeable membrane pipe is connected with the air chamber in a sealing mode.

Furthermore, first sealing rubber rings are arranged in the first flange seal and the second sealing flange, a first connecting lug is arranged on the periphery of the outlet end of the membrane permeating pipe, a second connecting lug matched with the first connecting lug is arranged at the inlet end of the air chamber, a second sealing rubber ring is arranged between the first connecting lug and the second connecting lug, and the first connecting lug and the second connecting lug are connected through fastening bolts.

Furthermore, the oil level sensor is arranged in the transformer oil tank, and the oil level sensor and the gas sensor array are electrically connected with the control panel.

Furthermore, the oil-gas separation membrane is made of PTEF polymeric membrane materials, the oil-gas separation membrane is not permeable to oil and is only permeable to gas, the polytetrafluoroethylene membrane has better gas permeability, and in order to guarantee effective detection of gas content in oil of the device, soluble polytetrafluoroethylene can be selected as the breathable membrane.

Furthermore, the artificial neural network system adopts 8255A programmable general parallel interface chip to form a singlechip, LED and keyboard interface circuit, and the control panel adopts a singlechip 80C196KB as an information processor.

Furthermore, the artificial neural network system comprises an input layer node number unit, an output layer node number unit, and an implicit layer number and in-layer contact number unit.

Furthermore, the number of input layer nodes of the input layer node number unit is 5, the number of output layer nodes of the output layer node number unit is also 5, and the selection of the number of hidden layers and the number of hidden layer units of the in-layer contact number unit adopts a formula

And (4) calculating, wherein m is the number of output neurons, n is the number of input neurons, a is a constant between 0 and 10, and 5 nodes of the hidden layer are taken.

A method for a device for diagnosing faults of a 500kV high-power transformer for a power plant comprises the following steps:

s1, starting;

s2, receiving parameters;

s3, reading the neural network structure information;

s4, reading the data information of the neural network;

s5, initializing a weight and a threshold;

s6, setting an initial variable;

s7, sending a sample into the neuron;

s8, calculating the output value of each layer;

s9, calculating the error of the output layer;

s10, counting the mean square error of all samples;

s11, reversely modifying the weight and the threshold;

s12, judging whether all samples are calculated completely, if not, returning to the step S7 to send the samples into the neurons again; if the completion, enter the next step;

s13, judging whether the average value is smaller than the expected value, if not, returning to the step S6 to reset the initial variables; and if the value is less than the preset value, ending the process.

(III) advantageous effects

The invention provides a device and a method for diagnosing faults of a 500kV high-power transformer for a power plant, which have the following beneficial effects:

(1) the oil-gas separation unit comprises an oil-gas separation membrane which is impermeable to oil and only permeates gas, a gas chamber for storing gas, an isolation valve and relevant equipment, wherein the isolation valve is usually in an open position, so that the gas permeable membrane is usually in a gas permeable state, gas permeation in oil is a diffusion process, gas molecules are diffused from oil to one side of the gas chamber due to the gas pressure difference on two sides of the membrane, and the gas pressure on two sides of the membrane tends to be balanced after a certain time at a certain temperature to achieve dynamic balance, so that oil-gas separation is automatically realized.

(2) The invention adopts the gas detection of the artificial neural network system, can obtain the content of various components of the mixed gas on the premise of not separating the gas, the artificial neural network system can realize complex nonlinear mapping, and after the artificial neural network system is used for the information processing of the gas sensor array, the serious nonlinear problem caused by cross sensitivity is better solved, and the artificial neural network system has good fault-tolerant performance on the drift or noise of the sensor, and can greatly improve the detection precision of the gas.

(3) The method diagnoses the fault of the transformer by analyzing the dissolved gas in the transformer oil, and is very valuable for timely and accurately detecting the early latent fault of the transformer, so that the safety and the stability of a power system are ensured.

(4) According to the invention, on the basis of the analysis result of the content of the dissolved gas in the oil, the artificial neural network system is combined to identify and diagnose the fault of the transformer by the components and the concentration of the mixed gas, and the high-performance singlechip 80C196KB is used as an information processor, so that the design of an interface circuit is simplified, the operation speed and the data throughput capacity are increased, and the system is easy to realize, low in cost and easy to operate.

(5) The invention has the advantages of simple instrument structure and operation, high detection accuracy, simple structure, small operation and maintenance workload of operation and maintenance personnel, great reduction of maintenance difficulty and long service life.

Drawings

FIG. 1 is a schematic structural diagram of a display control device of the present invention installed on a transformer tank;

FIG. 2 is a schematic structural view of the display control device according to the present invention after the cover plate is opened to expose the keyboard;

FIG. 3 is a schematic view of a connection structure of a transformer tank and a transformer tank according to the present invention;

FIG. 4 is a schematic structural diagram of oil-gas separation by a polymer membrane method according to the present invention;

FIG. 5 is a schematic diagram of an optimized implementation structure of the oil-gas separation device;

FIG. 6 is a schematic structural view of a sealing rubber ring of the sealing flange of the present invention;

FIG. 7 is a schematic view of the construction of the sealing rubber ring between the connecting lugs of the present invention;

FIG. 8 is a block diagram of an artificial neural network system according to the present invention;

FIG. 9 is a block diagram of the structure of an intelligent sensor array according to the present invention;

FIG. 10 is a block diagram of a control panel control system of the present invention;

FIG. 11 is a block diagram of a process for an artificial neural network system of the present invention.

In the figure: a membrane permeable tube 1; an oil-gas separation membrane 101; a first connecting lug 102; a transformer oil tank 2; an oil outlet pipe opening 201; an isolation valve 3; a first flange 301; a second sealing flange 302; a first sealing rubber ring 303; an air chamber 4; a second connecting lug 401; a second sealing rubber ring 402; a fastening bolt 403; an artificial neural network system 5; an input layer node count unit 501; an output layer node number unit 502; a hidden layer number and in-layer contact number unit 503; a gas sensor array 6; an intelligent sensor array 7; a display control device 8; an LED display 81; an LED display screen 811; a display lamp 812; a speaker 82; a control panel 83; a keyboard 831; a cover plate 832; a fuel level sensor 9; a transformer tank 10.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments.

In the present invention, unless otherwise expressly stated or limited, the terms "disposed," "mounted," "connected," "secured," and the like are to be construed broadly and can, for example, be fixedly connected, detachably connected, or integrally formed; they may be mechanically coupled, directly coupled, or indirectly coupled through intervening agents, both internally and/or in any other manner known to those skilled in the art. The specific meanings of the above terms in the present invention can be understood by those skilled in the art according to specific situations. Furthermore, the terms "first", "second" and "first" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature.

Referring to fig. 1-11, the present invention provides a technical solution: 1. the utility model provides a diagnostic device of 500kV high-power transformer trouble for power plant, includes permeable membrane pipe 1, its characterized in that: be provided with oil-gas separation membrane 101 in penetrating membrane pipe 1, penetrate one side of membrane pipe 1 and be connected with the mouth of pipe 201 that produces oil of transformer tank 2, and the mouth of pipe 201 that produces oil and penetrate and install isolating valve 3 between the membrane pipe 1, the opposite side of penetrating membrane pipe 1 is connected with the air chamber 4 that stores up gas, is provided with the intelligent sensor array 7 that artifical neural network system 5 and gas sensor array 6 are constituteed in the air chamber 4 to realize improving gaseous detection precision.

As a preferred technical scheme of the invention: the transformer box body 10 is characterized by further comprising a display and control device 8, the display and control device 8 is installed on a door panel on one side of the transformer box body 10, the display and control device 8 comprises an LED display 81, a loudspeaker 82 and a control panel 83, the bottoms of the LED display 81 and the loudspeaker 82 are fixed to the top of the control panel 83, an LED display screen 811 is installed on the upper side of the front face of the LED display 81, three display lamps 812 are installed on the lower side of the front face of the LED display 81, a keyboard 831 is arranged on the control panel 83, a cover plate 832 is connected to the control panel 83, the bottom of the cover plate 832 is hinged to the bottom of the control panel 83 through a pin shaft, a fixture block is fixed to the top of the cover plate 832, a clamping groove corresponding to the position of the fixture block is formed in the bottom of the LED display 81, and the fixture block is matched with the clamping groove; the LED display 811 and the display lamp 812 are electrically connected to the LED display 81, the input terminals of the LED display 81 and the speaker 82 are electrically connected to the output terminal of the control panel 83, and the keyboard 831 is electrically connected to the control panel 83.

As a preferred technical scheme of the invention: the inlet end of the isolation valve 3 is connected with the oil outlet pipe port 201 end of the transformer oil tank 2 in a sealing mode through a first flange 301, the outlet end of the isolation valve 3 is connected with the inlet end of the permeable membrane pipe 1 through a second sealing flange 302, and the outlet end of the permeable membrane pipe 1 is connected with the air chamber 4 in a sealing mode.

As a preferred technical scheme of the invention: the first sealing rubber ring 303 is arranged in each of the first flange 301 seal and the second sealing flange 302, the first connecting lug 102 is arranged on the periphery of the outlet end of the membrane permeable pipe 1, the second connecting lug 401 matched with the first connecting lug 102 is arranged on the inlet end of the air chamber 4, the second sealing rubber ring 402 is arranged between the first connecting lug 102 and the second connecting lug 401, and the first connecting lug 102 and the second connecting lug 401 are connected through a fastening bolt 403.

As a preferred technical scheme of the invention: the oil level sensor 9 is arranged in the transformer oil tank 2, the oil level sensor 9 and the gas sensor array 6 are electrically connected with a control panel 83, and the control panel 83 adopts a single chip microcomputer 80C196KB as an information processor.

As a preferred technical scheme of the invention: the oil-gas separation membrane 101 is made of a PTEF polymer membrane material.

As a preferred technical scheme of the invention: the artificial neural network system 5 includes an input layer node number unit 501 and an output layer node number unit 502, and an implicit layer number and in-layer contact number unit 503.

As a preferred technical scheme of the invention:

1) determination of the number of input layer nodes:

the input layer acts as a buffer memory for applying the data source to the network, the number of nodes of which depends on the dimension of the data source, i.e. the dimension of the input feature vector; after the data source is determined, determining nodes required by an input layer; the concentration of 5 characteristic gases is required to be measured by adopting a three-ratio method, the dimension of a data source can be set to be 5, namely the number of nodes of an input layer is 5;

2) determination of the number of output layer nodes:

the neural network is adopted to accurately measure the concentration of each characteristic gas, and the output of the neural network is the concentration value of each characteristic gas; the number of output layer nodes is also 5;

3) determination of the number of hidden layers and contacts in the layers:

the selection of the number of the hidden layer units adopts a formula

Calculation, where m is the number of output neurons, n is the number of input neurons, a is 0Constant between 10, the nodes of the hidden layer take 5.

The neural network can be realized by assembly language or high-level language such as C, C + +, JAVA, the program diagram is shown in fig. 2, the learning stage of the neural network is completed on a PC, after training is completed, parameters such as fixed weight, threshold, algorithm and network structure of the neural network are obtained, and the parameters are burned into an external ROM for detection.

2. A method for a device for diagnosing faults of a 500kV high-power transformer for a power plant comprises the following steps:

s1, starting;

s2, receiving parameters;

s3, reading the neural network structure information;

s4, reading the data information of the neural network;

s5, initializing a weight and a threshold;

s6, setting an initial variable;

s7, sending a sample into the neuron;

s8, calculating the output value of each layer;

s9, calculating the error of the output layer;

s10, counting the mean square error of all samples;

s11, reversely modifying the weight and the threshold;

s12, judging whether all samples are calculated completely, if not, returning to the step S7 to send the samples into the neurons again; if the completion, enter the next step;

s13, judging whether the average value is smaller than the expected value, if not, returning to the step S6 to reset the initial variables; and if the value is less than the preset value, ending the process.

3. Data acquisition and signal processing:

3.1 selection of the single chip microcomputer:

the MCS-9616 bit single chip microcomputer is an embedded microcontroller, because of its high-performance register-register structure, the CPU operation is directly oriented to all data registers in the memory, and the bottleneck effect that some single chip microcomputers only use accumulators for operation is eliminated, so the operation speed and the data throughput capacity are greatly improved, and the method can be well applied to real-time control, and the design adopts Intel 80C196 KB.

3.2 hardware design:

the artificial neural network system 5 adopts 8255A programmable general parallel interface chip to form a singlechip, LED and keyboard interface circuit.

1) Principle of LED display:

the LED display is an output device in a singlechip application system, eight sections of light emitting diodes are used for displaying numbers, each section is a long and thin filament capable of emitting light, the wavelength of the emitted light is about 6000A and is orange red, one end of each section is generally connected to the ground (common cathode connection method), the other end of each section is connected to a current limiting resistor and then connected to a signal end, when the signal is high level, the section is lightened, otherwise, the common anode connection method is not lightened, the connected ports are high level, only when the other end of the connected ports is low level, the connected ports are lightened, otherwise, the connected ports are not lightened.

The lighting display has two methods of static display and dynamic display, the static display is that when the display displays a certain character, the corresponding light-emitting diode is constantly conducted or cut, each bit of the display mode needs to be controlled by an 8-bit output port, but because the resource is limited, only one 8-bit output port can be used for controlling the LED, so a plurality of nixie tubes are displayed, the visual effect is that a plurality of nixie tubes are statically displayed, scanning is used for dynamic display, the realization method is that another nixie tube is displayed on the basis of the static display of a single nixie tube, and circulation is carried out, the time delay is properly shortened, namely another nixie tube is displayed immediately after one is displayed, so that the naked eye can not distinguish, a plurality of simultaneously displayed effects are obtained, each nixie tube is sequentially lighted, the control is very important, by utilizing the visual persistence principle, the display is conducted more than 16 times per second, the digital tube cannot be distinguished by naked eyes from being temporarily not bright, and is considered to be always bright (the digital tube is in flickering at a certain frequency), but the delay (conduction frequency) is not smaller and better, the digital tube needs a certain time to reach a certain brightness, so that the characteristic of the digital tube cannot be avoided, if the delay control is not good, the flickering occurs, or the brightness is not enough, and the satisfactory effect can be achieved by preferably delaying for 0.005 second.

2) The keyboard working principle is as follows:

the row-column keyboard, also called matrix keyboard, uses I/O port group to form row and column structure, the press key is set on the intersection point of row and column, for example, 2X 2 row and column structure can be used to form 4-key keyboard, 4X 4 row and column structure can be used to form 16-key keyboard, therefore, under the condition of more press keys, it can save I/O port line, the total number of press keys of said design is 12, 0-9 are number keys, and one key is used to represent decimal point for modifying information of neural network, and another press key is used to represent modification end.

3) Designing an interface circuit:

in the display interface circuit formed by using 8255A parallel extended I/O port, the LED adopts a common cathode, the segment selection code is provided by a PB port of 8255A, the bit selection code is provided by a PA port of 8255A, an address latch chip 74LS373 latches 80C196KB input data from a P3 port, meanwhile, P3, 0, P3 and 1 ports are used as control signal input ends of 8255A, the LED adopts dynamic display software decoding, the keyboard adopts a line-by-line scanning query working mode, and the LED is driven by an open-collector output eight-bit driver 8718 produced by Beijing integrated circuit design center.

4) Expansion of external memory:

because the neural network is realized by software, higher requirements are put on the capacity of the memory, and the size of the required memory capacity can be calculated according to the following formula, wherein the size of the memory capacity is the number of bytes per item multiplied by 2 x (interconnection number + active value number), and the number of bytes per item is 4; number of interconnected neurons x (number of cryptic neurons +1) + number of cryptic neurons x (number of input neurons + 1); the active value number is the number of network neurons multiplied by the number of samples, and it can be seen that the size of the required memory capacity is related to the number of the selected samples, because 80C196KB has no ROM inside, a program memory needs to be expanded outside, 8K of E2PROM 2864A is selected as the external ROM for storing the parameters such as the weight, the threshold value, the network algorithm and the structure of the trained neural network on the PC, and the E2PROM is an electrically erasable programmable read-only memory, so the E2PROM is selected because when the E2PROM is used as the program memory, the codes in the E2PROM can be rewritten online, and is more flexible than the EPROM, and when the parameters in the program memory are modified later, the parameters can be modified online directly through the keyboard, the E2PROM selects Intel 2864A, and the memory capacity is 8K.

5) Resetting and protecting the system:

the single chip microcomputer samples five kinds of external gases once every pair, so that the state of the light emitting diode connected with the HSO, 0 is changed once, the display system works normally, the reset circuit plays an important role in resisting interference of the microcomputer, and when the state of the light emitting diode is not changed or the LED display is abnormal, the manual reset circuit can be adopted to reset the system and operate again to determine faults.

4. Designing software:

1) design of main program

The design of the main program includes initialization for 8255, timers, etc.

(1) Initialization of the timer T1 in 80C196KB implements the timer 1s, the timing function is a/D conversion every 1s, timer T1 is a real time clock in the system, its clock signal comes from the internal clock circuit, it works constantly once the system is reset, 1 is added to the counter every 8 state periods (2 μ s at 12MHz crystal), the count value of T1 is placed in the special register area, the addresses 0AH (lower 8 bits) and 0BH (upper 8 bits), denoted timer (lo) and timer (hi), respectively, it can only be read by word, the port address is 0AH, the only way to stop the count of T1 and restore to "0" is to reset the system, when the 16 bit counter is full, the timer is triggered to overflow interrupt, IOS1, 5 is set, the interrupt vector of timer T1 is 2000H, the overflow time is 87, 3ms, a total of 11 interrupts are required to generate a timing of 1 s.

(2)8255, the input ports are A and B, the input ports are C, the input address of 8255 is 8000H, and the address of mode control word is 8003H.

(3) Initialization of the RAM area selects 40H, 42H, 44H, 46H, 48H to store the concentration values of the five processed gases (C2H 2, C2H4, C2H6, CH4 and H2 in sequence), and 60H is used as the first address of the display buffer area.

(4) And the address allocation of the external E2PROM 2864A, namely the address range of the E2PROM is 2000H-3 FFFH according to a hardware wiring diagram.

2) Design of A/D conversion subroutine:

the A/D conversion subprogram has the function of sampling five paths of analog quantity signals and sending the signals into the single chip microcomputer, the program is designed to sample 1 time every 1 second, and whether the A/D conversion is finished or not is judged by inquiring whether the 3 rd bit of the conversion result low byte register is 0 or not.

3) Design of display subroutine:

and 4 nixie tubes are adopted to display the fault types, the relationship between the display content and the fault types is that F000 is normal, F001 is the first fault type, and by analogy, 3 coding values obtained in the data processing subprogram are integrated to judge the fault types, and the fault numbers can be obtained through the corresponding relationship between different combinations of the coding values and the fault numbers.

4) Debugging of the program:

the design adopts Wave6000 integrated development environment of Nanjing Weifu, which can debug MCS-51, MCS-96, 8086 and the like, and an assembly compiler and a connector of TASKING company, and Wave6000 integrated development environment.

The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art should be considered to be within the technical scope of the present invention, and the technical solutions and the inventive concepts thereof according to the present invention should be equivalent or changed within the scope of the present invention.

Claims (10)

1. The utility model provides a diagnostic device of power plant with high-power transformer trouble of 500kV, includes permeable membrane pipe (1), its characterized in that: be provided with oil-gas separation membrane (101) in permeable membrane pipe (1), one side of permeable membrane pipe (1) is connected with the mouth of pipe (201) that produces oil of transformer tank (2), and produces oil and install isolating valve (3) between mouth of pipe (201) and permeable membrane pipe (1), and the opposite side of permeable membrane pipe (1) is connected with stores up gaseous air chamber (4), is provided with intelligent sensor array (7) that artifical neural network system (5) and gas sensor array (6) are constituteed in air chamber (4) to realize improving gaseous detection precision.

2. The device for diagnosing the faults of the 500kV high-power transformer for the power plant as claimed in claim 1, wherein: the transformer cabinet is characterized by further comprising a display and control device (8), the display and control device (8) is installed on a door panel on one side of the transformer cabinet body (10), the display and control device (8) comprises an LED display (81), a loudspeaker (82) and a control panel (83), the bottoms of the LED display (81) and the loudspeaker (82) are fixed to the top of the control panel (83), an LED display screen (811) is installed on the upper side of the front face of the LED display (81), three display lamps (812) are installed on the lower side of the front face of the LED display (81), the keyboard (831) is arranged on the control panel (83), the control panel (83) is connected with a cover plate (832), the bottom of the cover plate (832) is hinged with the bottom of the control panel (83) through a pin shaft, a clamping block is fixed at the top of the cover plate (832), the bottom of the LED display (81) is provided with a clamping groove corresponding to the position of the clamping block, and the clamping block is matched with the clamping groove; the LED display screen (811) and the display lamp (812) are electrically connected with the LED display (81), the input ends of the LED display (81) and the loudspeaker (82) are electrically connected with the output end of the control panel (83), and the keyboard (831) is electrically connected with the control panel (83).

3. The device for diagnosing the faults of the 500kV high-power transformer for the power plant as claimed in claim 2, wherein: the inlet end of the isolating valve (3) is connected with the oil outlet pipe opening (201) end of the transformer oil tank (2) in a sealing mode through a first flange (301), the outlet end of the isolating valve (3) is connected with the inlet end of the permeable membrane pipe (1) through a second sealing flange (302), and the outlet end of the permeable membrane pipe (1) is connected with the air chamber (4) in a sealing mode.

4. The device for diagnosing the faults of the 500kV high-power transformer for the power plant as claimed in claim 3, wherein: the membrane permeable pipe is characterized in that first sealing rubber rings (303) are arranged in the first flange (301) seal and the second sealing flange (302), first connecting lugs (102) are arranged on the periphery of the outlet end of the membrane permeable pipe (1), second connecting lugs (401) matched with the first connecting lugs (102) are arranged at the inlet end of the air chamber (4), second sealing rubber rings (402) are arranged between the first connecting lugs (102) and the second connecting lugs (401), and the first connecting lugs (102) and the second connecting lugs (401) are connected through fastening bolts (403).

5. The device for diagnosing the faults of the 500kV high-power transformer for the power plant as claimed in claim 4, wherein: the oil level sensor is characterized by further comprising an oil level sensor (9) arranged in the transformer oil tank (2), wherein the oil level sensor (9) and the gas sensor array (6) are electrically connected with the control panel (83).

6. The device for diagnosing the faults of the 500kV high-power transformer for the power plant as claimed in claim 5, wherein: the oil-gas separation membrane (101) is made of a PTEF high-molecular membrane material.

7. The device for diagnosing the faults of the 500kV high-power transformer for the power plant as claimed in claim 6, wherein: the artificial neural network system (5) adopts 8255A programmable general parallel interface chips to form a singlechip, an LED and keyboard (831) interface circuit, and the control panel (83) adopts a singlechip 80C196KB as an information processor.

8. The device for diagnosing the faults of the 500kV high-power transformer for the power plant as claimed in claim 7, wherein: the artificial neural network system (5) comprises an input layer node number unit (501), an output layer node number unit (502) and an implicit layer number and in-layer contact number unit (503).

9. The device for diagnosing the faults of the 500kV high-power transformer for the power plant according to claim 8, characterized in that: the number of input layer nodes of the input layer node number unit (501) is 5, and the number of output layer nodes of the output layer node number unit (502) is also 55, the hidden layer number of the hidden layer number and the number of contacts in the layer (503) are selected by adopting a formula

And (4) calculating, wherein m is the number of output neurons, n is the number of input neurons, a is a constant between 0 and 10, and 5 nodes of the hidden layer are taken.

10. The method for diagnosing the fault of the 500kV high-power transformer used in the power plant according to any one of claims 1 to 9, wherein: the method comprises the following steps:

s1, starting;

s2, receiving parameters;

s3, reading the neural network structure information;

s4, reading the data information of the neural network;

s5, initializing a weight and a threshold;

s6, setting an initial variable;

s7, sending a sample into the neuron;

s8, calculating the output value of each layer;

s9, calculating the error of the output layer;

s10, counting the mean square error of all samples;

s11, reversely modifying the weight and the threshold;

s12, judging whether all samples are calculated completely, if not, returning to the step S7 to send the samples into the neurons again; if the completion, enter the next step;

s13, judging whether the average value is smaller than the expected value, if not, returning to the step S6 to reset the initial variables; and if the value is less than the preset value, ending the process.

Images