A dry-type transformer characteristic decomposition gas detection device and method

Abstract

The present invention discloses a device and method for detecting characteristic decomposition gases of dry-type transformers. The device comprises: a formaldehyde data detection module that detects the formaldehyde concentration in the environment and obtains formaldehyde concentration data; a carbon monoxide data detection module that detects the carbon monoxide concentration in the environment and obtains carbon monoxide concentration data; a main control module that determines the output probability of the dry-type transformer characteristic decomposition gas content exceeding the standard based on the formaldehyde concentration data and the carbon monoxide concentration data, and determines the environmental state based on the probability; wherein the environmental state includes: a state of exceeding the standard content and a state of normal content. The device of the present invention can effectively monitor multiple characteristic decomposition gases generated during the operation of the dry-type transformer, and avoids cross-interference between multiple gases through a sensor array and a neural network optimization algorithm. The device can also evaluate the characteristic decomposition gas content of the gas surrounding the dry-type transformer and provide effective early warnings to prevent electrical accidents in the dry-type transformer.

Description

Technical Field

[0001] The present invention relates to the technical field of gas detection, and more particularly to a dry-type transformer characteristic decomposition gas detection device and method. Background Art

[0002] Currently, domestic dry-type transformers typically use vacuum-cast epoxy resin insulation. When overheating or prolonged full-load operation occurs, the epoxy resin inside the dry-type transformer decomposes, generating associated gases. This odor is particularly prominent in enclosed box-type substations. Furthermore, varying degrees of odor can be emitted during temperature rise and load tests, especially when the capacity is significantly inflated.

[0003] Under heating conditions, the characteristic decomposition products of epoxy resin are CO2, CH2O, H2O, and CO, in order of concentration. By detecting CH2O and CO gases around dry-type transformers during operation or testing, it is possible to effectively detect overheating faults in dry-type transformers and assess product quality. Currently, there are no portable detection devices for both CH2O and CO gas concentrations. Existing portable CH2O or CO analyzers mostly detect a single gas. However, due to gas cross-interference, the signal output of a sensor may include cross-interference from other gases, resulting in poor detection accuracy and unsuitable for low-concentration, multi-gas detection in the electrical field. Summary of the Invention

[0004] The present invention proposes a dry-type transformer characteristic decomposition gas detection device to solve the problem of how to efficiently and accurately detect the dry-type transformer characteristic decomposition gas.

[0005] In order to solve the above problems, according to one aspect of the present invention, a dry-type transformer characteristic decomposition gas detection device is provided, the device comprising: a formaldehyde data detection module, a carbon monoxide data detection sensor and a main control module; wherein,

[0006] The formaldehyde data detection module is connected to the main control module and is used to detect the formaldehyde concentration in the environment and obtain formaldehyde concentration data;

[0007] The carbon monoxide data detection module is connected to the main control module and is used to detect the carbon monoxide concentration in the environment and obtain carbon monoxide concentration data;

[0008] The main control module is used to determine the output probability of the dry-type transformer characteristic decomposition gas content exceeding the standard based on the formaldehyde concentration data and the carbon monoxide concentration data, and determine the environmental state based on the probability; wherein the environmental state includes: a characteristic decomposition gas content exceeding the standard state and a characteristic decomposition gas content normal state.

[0009] Preferably, the device further comprises:

[0010] The AD conversion module is connected to the input ends of the formaldehyde data detection module, the carbon monoxide data detection module and the main control module, respectively, and is used to convert the formaldehyde concentration data and the carbon monoxide concentration data into digital signals, and send the formaldehyde concentration data and the carbon monoxide concentration data converted into digital signals to the main control module.

[0011] Preferably, the formaldehyde data detection module includes: a first formaldehyde sensor, a second formaldehyde sensor and a third formaldehyde sensor; wherein the first formaldehyde sensor is a semiconductor type formaldehyde sensor; the second formaldehyde sensor is equipped with an automotive-grade fuel cell solid electrolyte; the third formaldehyde sensor is a fuel cell type electrochemical sensor, in which formaldehyde and oxygen can undergo corresponding redox reactions on the working electrode and the counter electrode and release charges to form current.

[0012] Preferably, the main control module determines the output probability of the dry-type transformer characteristic decomposition gas content exceeding the standard according to the formaldehyde concentration data and the carbon monoxide concentration data, including:

[0013] The central processing unit is configured to determine the output probability of the dry-type transformer characteristic decomposition gas content exceeding the standard by the following method, including:

[0014]

[0015] Where P(x) is the output probability of the dry-type transformer characteristic decomposition gas content exceeding the standard; x i is the gas concentration data detected by the i-th sensor, which includes a formaldehyde sensor and a carbon monoxide sensor; ω i is the weight parameter corresponding to the gas concentration data detected by the i-th sensor, and b is the preset coefficient; the weight parameter and preset coefficient are obtained by neural network learning based on historical formaldehyde concentration data, carbon monoxide concentration data and environmental status data using a neural network processor.

[0016] Preferably, the neural network processor includes: a convolution unit, a batch normalization unit, an activation unit and a pooling operation unit; the neural network processor is a fixed-point model trained based on a mainstream training framework according to specific restriction rules, and supports separate configuration of convolutional neural network parameters for each layer, including the number of input and output channels, input and output row width and column height. The neural network processor supports two convolution kernels, 1x1 and 3x3, and the maximum supported neural network parameter size is 5.5MiB to 5.9MiB when working in real time.

[0017] Preferably, the device further comprises:

[0018] an alarm, connected to the main control module, for issuing an alarm when the environmental state is determined to be a state where the formaldehyde content exceeds the standard based on the probability;

[0019] Display module, used to display formaldehyde concentration data, carbon monoxide concentration data, output probability of dry-type transformer characteristic decomposition gas content exceeding the standard, environmental status and alarm working status;

[0020] A power supply module is used to provide power support for the dry-type transformer characteristic decomposition gas detection device.

[0021] Preferably, the alarm is a buzzer;

[0022] The display module is a display including a liquid crystal display screen, the liquid crystal screen is an LED resistive touch screen, and the LED resistive touch screen can dynamically display formaldehyde concentration data and carbon monoxide concentration data.

[0023] According to another aspect of the present invention, a dry-type transformer characteristic decomposition gas detection method is provided, the method comprising:

[0024] The formaldehyde data detection module detects the formaldehyde concentration in the environment and obtains formaldehyde concentration data;

[0025] The carbon monoxide data detection module detects the carbon monoxide concentration in the environment and obtains carbon monoxide concentration data;

[0026] The main control module determines the output probability of the dry-type transformer characteristic decomposition gas content exceeding the standard based on the formaldehyde concentration data and the carbon monoxide concentration data, and determines the environmental state based on the probability; wherein the environmental state includes: a characteristic decomposition gas content exceeding the standard state and a characteristic decomposition gas content normal state.

[0027] Preferably, the method further comprises:

[0028] The AD conversion module converts the formaldehyde concentration data and the carbon monoxide concentration data into digital signals, and sends the formaldehyde concentration data and the carbon monoxide concentration data converted into digital signals to the main control module.

[0029] Preferably, the formaldehyde data detection module includes: a first formaldehyde sensor, a second formaldehyde sensor and a third formaldehyde sensor; wherein the first formaldehyde sensor is a semiconductor type formaldehyde sensor; the second formaldehyde sensor is equipped with an automotive-grade fuel cell solid electrolyte; the third formaldehyde sensor is a fuel cell type electrochemical sensor, in which formaldehyde and oxygen can undergo corresponding redox reactions on the working electrode and the counter electrode and release charges to form current.

[0030] Preferably, the main control module determines the output probability of the dry-type transformer characteristic decomposition gas content exceeding the standard according to the formaldehyde concentration data and the carbon monoxide concentration data, including:

[0031] The central processing unit determines the output probability of the dry-type transformer characteristic decomposition gas content exceeding the standard by the following method, including:

[0032]

[0033] Where P(x) is the output probability of the dry-type transformer characteristic decomposition gas content exceeding the standard; x i is the gas concentration data detected by the i-th sensor, which includes a formaldehyde sensor and a carbon monoxide sensor; ω i is the weight parameter corresponding to the gas concentration data detected by the i-th sensor, and b is the preset coefficient; the weight parameter and preset coefficient are obtained by neural network learning based on historical formaldehyde concentration data, carbon monoxide concentration data and environmental status data using a neural network processor.

[0034] Preferably, the method further comprises:

[0035] The alarm device issues an alarm when it is determined based on the probability that the environmental state is a state where the characteristic decomposition gas content exceeds the standard;

[0036] The display module displays formaldehyde concentration data, carbon monoxide concentration data, output probability of dry-type transformer characteristic decomposition gas content exceeding the standard, environmental status and alarm working status;

[0037] The power supply module provides power support for the dry-type transformer characteristic decomposition gas detection device.

[0038] The present invention provides a device and method for detecting characteristic decomposition gases in dry-type transformers. The device comprises: a formaldehyde data detection module that detects the formaldehyde concentration in the environment and obtains formaldehyde concentration data; a carbon monoxide data detection module that detects the carbon monoxide concentration in the environment and obtains carbon monoxide concentration data; a main control module that determines the output probability of the dry-type transformer characteristic decomposition gas content exceeding the standard based on the formaldehyde concentration data and the carbon monoxide concentration data, and determines the environmental state based on the probability; wherein the environmental state includes: a state where the characteristic decomposition gas content exceeds the standard and a state where the characteristic decomposition gas content is normal. The device of the present invention can effectively monitor multiple characteristic decomposition gases generated during dry-type transformer operation, avoiding cross-interference between multiple gases through a sensor array and a neural network optimization algorithm. It can also assess the characteristic decomposition gas content in the gas surrounding the dry-type transformer and issue effective warnings to prevent electrical accidents in the dry-type transformer. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] A more complete understanding of exemplary embodiments of the present invention may be obtained by referring to the following drawings:

[0040] Figure 1 2 is a schematic structural diagram of a dry-type transformer characteristic decomposition gas detection device 100 according to an embodiment of the present invention;

[0041] Figure 2 This is an exemplary diagram of a dry-type transformer characteristic decomposition gas detection device according to an embodiment of the present invention;

[0042] Figure 3 Schematic diagram of detection results of three formaldehyde sensor devices according to an embodiment of the present invention;

[0043] Figure 4 4 is a flow chart of a dry-type transformer characteristic decomposition gas detection method 400 according to an embodiment of the present invention. DETAILED DESCRIPTION

[0044] Exemplary embodiments of the present invention will now be described with reference to the accompanying drawings. However, the present invention may be embodied in many different forms and is not limited to the embodiments described herein. These embodiments are provided to provide a thorough and complete disclosure of the present invention and to fully convey the scope of the present invention to those skilled in the art. The terminology used in the exemplary embodiments shown in the accompanying drawings is not intended to limit the present invention. In the accompanying drawings, identical elements are denoted by the same reference numerals.

[0045] Unless otherwise specified, the terms used herein (including technical terms) have the meanings commonly understood by those skilled in the art. In addition, it is understood that terms defined in commonly used dictionaries should be understood to have the same meanings as those in the context of the relevant fields, and should not be understood as idealized or overly formal meanings.

[0046] To address the existing deficiencies, the present invention proposes a dry-type transformer characteristic decomposition gas detection device and method. By adding a gas sensor matrix to calculate multiple signal quantities to eliminate cross-interference, the cross-detection of the concentrations of gases such as CH2O and CO around the dry-type transformer is utilized to evaluate the gas content of the dry-type transformer gas characteristic decomposition, thereby determining whether the dry-type transformer has an overheating fault or quality problem, and preventing the occurrence of dry-type transformer electrical accidents.

[0047] Figure 1 FIG. 1 is a schematic structural diagram of a dry-type transformer characteristic decomposition gas detection device 100 according to an embodiment of the present invention. Figure 1As shown, the dry-type transformer characteristic decomposition gas detection device provided in an embodiment of the present invention can effectively monitor multiple characteristic decomposition gases generated during dry-type transformer operation. Through a sensor array and a neural network optimization algorithm, cross-interference between multiple gases is avoided. The device can assess the characteristic decomposition gas content in the gas surrounding the dry-type transformer and provide effective early warnings to prevent electrical accidents in the dry-type transformer. The dry-type transformer characteristic decomposition gas detection device 100 provided in an embodiment of the present invention includes: a formaldehyde data detection module 101, a carbon monoxide data detection sensor 102, and a main control module 103.

[0048] Preferably, the formaldehyde data detection module 101 is connected to the main control module, and is used to detect the formaldehyde concentration in the environment and obtain formaldehyde concentration data.

[0049] Preferably, the formaldehyde data detection module includes: a first formaldehyde sensor, a second formaldehyde sensor and a third formaldehyde sensor; wherein the first formaldehyde sensor is a semiconductor type formaldehyde sensor; the second formaldehyde sensor is equipped with an automotive-grade fuel cell solid electrolyte; the third formaldehyde sensor is a fuel cell type electrochemical sensor, in which formaldehyde and oxygen can undergo corresponding redox reactions on the working electrode and the counter electrode and release charges to form current.

[0050] Preferably, the carbon monoxide data detection module 102 is connected to the main control module, and is used to detect the carbon monoxide concentration in the environment and obtain carbon monoxide concentration data.

[0051] Preferably, the device further comprises:

[0052] The AD conversion module is connected to the input ends of the formaldehyde data detection module, the carbon monoxide data detection module and the main control module, respectively, and is used to convert the formaldehyde concentration data and the carbon monoxide concentration data into digital signals, and send the formaldehyde concentration data and the carbon monoxide concentration data converted into digital signals to the main control module.

[0053] In an embodiment of the present invention, the formaldehyde data detection module may include multiple formaldehyde sensors, and the data detected by each formaldehyde sensor is used as formaldehyde concentration data. The carbon monoxide (CO) sensing module includes at least one carbon monoxide sensor (ie, CO sensor).

[0054] Combine Figure 2As shown, in one embodiment of the present invention, the formaldehyde data detection module includes: a first formaldehyde sensor, a second formaldehyde sensor, and a third formaldehyde sensor. The first formaldehyde sensor is exposed to the mixed gas in the environment to detect the formaldehyde concentration in real time and generate first data; the second formaldehyde sensor includes a solid electrolyte and is exposed to the mixed gas in the environment to detect the formaldehyde concentration in real time and generate second data; the third formaldehyde sensor is exposed to the mixed gas in the environment to detect the formaldehyde concentration in real time and generate fourth data. In this case, the formaldehyde concentration data includes: the first data, the second data, and the fourth data. The CO sensor is exposed to the mixed gas in the environment to detect the carbon monoxide concentration in real time and generate third data. In this case, the carbon monoxide concentration data includes the third data.

[0055] The AD conversion module is connected to the first formaldehyde sensor, the second formaldehyde sensor, the CO sensor and the third formaldehyde sensor, and converts the first data, the second data, the third data and the fourth data into digital signals and transmits them to the main control circuit.

[0056] Preferably, the main control module 103 is used to determine the output probability of the dry-type transformer characteristic decomposition gas content exceeding the standard based on the formaldehyde concentration data and the carbon monoxide concentration data, and determine the environmental state based on the probability; wherein the environmental state includes: a characteristic decomposition gas content exceeding the standard state and a characteristic decomposition gas content normal state.

[0057] Preferably, the main control module determines the output probability of the dry-type transformer characteristic decomposition gas content exceeding the standard according to the formaldehyde concentration data and the carbon monoxide concentration data, including:

[0058] The central processing unit is configured to determine the output probability of the dry-type transformer characteristic decomposition gas content exceeding the standard by the following method, including:

[0059]

[0060] Where P(x) is the output probability of the dry-type transformer characteristic decomposition gas content exceeding the standard; x i is the gas concentration data detected by the i-th sensor, which includes a formaldehyde sensor and a carbon monoxide sensor; ω i is the weight parameter corresponding to the gas concentration data detected by the i-th sensor, and b is the preset coefficient; the weight parameter and preset coefficient are obtained by neural network learning based on historical formaldehyde concentration data, carbon monoxide concentration data and environmental status data using a neural network processor.

[0061] Preferably, the neural network processor includes: a convolution unit, a batch normalization unit, an activation unit and a pooling operation unit; the neural network processor is a fixed-point model trained based on a mainstream training framework according to specific restriction rules, and supports separate configuration of convolutional neural network parameters for each layer, including the number of input and output channels, input and output row width and column height. The neural network processor supports two convolution kernels, 1x1 and 3x3, and the maximum supported neural network parameter size is 5.5MiB to 5.9MiB when working in real time.

[0062] In one embodiment of the present invention, the main control module includes: a central processing unit and a neural network processor.

[0063] The central processing unit is connected to the AD conversion module, and the central processing unit outputs a probability based on the digital signal, wherein the expression of the output probability is:

[0064]

[0065] Where P(x) is the output probability of the dry-type transformer characteristic decomposition gas content exceeding the standard; x i is the formaldehyde concentration data detected by the i-th gas sensor; ω i is the weight parameter corresponding to the gas concentration data detected by the i-th gas sensor, and b is the preset coefficient.

[0066] A neural network processor is used to perform neural network learning based on historical formaldehyde concentration data, carbon monoxide concentration data and environmental status data to obtain the weight parameters and preset coefficients.

[0067] When the output probability of the characteristic decomposition gas content of the dry-type transformer exceeding the standard is greater than a preset threshold, the environmental state can be determined as a state of characteristic decomposition gas content exceeding the standard based on the probability, and it is necessary to confirm the operating status of the dry-type transformer and check whether there is an overheating fault or quality problem; otherwise, the environmental state is a state of normal characteristic decomposition gas content, and the dry-type transformer is normal.

[0068] Combine Figure 2 As shown, in the present invention, the dry-type transformer characteristic decomposition gas detection device includes three formaldehyde sensors, a carbon monoxide sensor, an AD conversion module and a main control module. The first formaldehyde sensor is a semiconductor type formaldehyde sensor; the second formaldehyde sensor has a built-in automotive-grade fuel cell solid electrolyte; the third formaldehyde sensor is a fuel cell type electrochemical sensor, and formaldehyde and oxygen undergo corresponding redox reactions on the working electrode and the counter electrode and release charges to form current.

[0069] Among them, the first formaldehyde sensor is used to detect the formaldehyde concentration and generate the first data; the second formaldehyde sensor, which includes a solid electrolyte, is used to detect the formaldehyde concentration in real time and generate the second data; the CO sensor is used to detect the carbon monoxide concentration in real time and generate the third data; the third formaldehyde sensor is used to detect the formaldehyde concentration in real time and generate the fourth data; the AD conversion module is connected to the first formaldehyde sensor, the second formaldehyde sensor, the CO sensor and the third formaldehyde sensor, and converts the first data, the second data, the third data and the fourth data into digital signals; the main control module includes a central processing unit and a neural network processor, and the central processing unit outputs an output probability based on the digital signal. At this time, the expression of the output probability is:

[0070]

[0071] Among them, x1, x2, x3, x4 are the first data, second data, third data and fourth data respectively, w1, w2, w3, w4 are weight coefficients, and b is a preset coefficient.

[0072] The central processing unit is a dual-core 64-bit processor with a fast Fourier transform accelerator. The neural network processor includes a convolution unit, a batch normalization unit, an activation unit, and a pooling operation unit.

[0073] The converted data is fed into a trained model on a neural network processor. The model returns the probability that the dry-type transformer's characteristic decomposition gas content in the environment exceeds the specified level. If the probability is greater than 0.5, the device will trigger an alarm. The neural network processor collects large amounts of data on both mixed gases requiring and not requiring alarms, processes the data, and then uses it to train the model.

[0074] In an embodiment of the present invention, a rechargeable, portable dry-type transformer characteristic decomposition gas detection device comprises four sensors, an AD conversion module, and a K210 main control circuit (i.e., the main control module). The K210 main control module includes a K210 chip, and the AD conversion module includes an 8-bit bipolar input chip. The four sensors include three formaldehyde gas sensors of different models and one carbon monoxide gas sensor. These four gas sensors collect a dataset of mixed gases from the environment. The AD module then converts the analog signals collected by the four gas sensors into digital signals, which are then sent to the K210 main control module for processing and analysis. The K210 main control module analyzes the gas data to determine whether the dry-type transformer characteristic decomposition gas content exceeds the standard. If so, it controls a buzzer alarm. Furthermore, the processed data is sent to an intelligent LCD screen for display, which dynamically displays the real-time data collected by the four gas sensors. The bipolar input chip features four analog inputs, A1 and A2, which can be hardware-programmed for addressing. This allows eight PCF8591 devices to be connected to the same I2C bus without the need for additional hardware. Address, control, and data signals are transmitted serially via the bidirectional I2C bus. The bipolar input chip supports four external voltage inputs (0-5V). It also includes an integrated photoresistor, enabling precise ADC acquisition of ambient light intensity and a thermistor, enabling precise ADC acquisition of ambient temperature. It also includes a 0-5V voltage input, with the input voltage adjusted using a blue potentiometer. The bipolar input chip also features a power indicator, which illuminates when power is supplied. It also features a DA output indicator, which illuminates when the module's DA output voltage reaches a certain level. The higher the voltage, the brighter the indicator. The four-channel sensor module is an all-in-one sensor that can simultaneously detect the total particulate matter concentration, formaldehyde concentration, temperature, and humidity in the air. The detection of particulate matter concentration is based on the principle of laser scattering, which can continuously collect and calculate the number of suspended particles of different particle diameters in the air per unit volume, that is, the particle concentration distribution, and then convert it into mass concentration. The detection of formaldehyde concentration is based on electrochemical principles and has the characteristics of high precision and high stability. The particulate matter concentration number, formaldehyde concentration number, temperature, and humidity values ​​are combined and output in the form of a universal digital interface. The sensor can be embedded in various air quality-related instruments or environmental improvement equipment to provide them with timely and accurate concentration data. Preferably, the first formaldehyde sensor is a CH20 sensor.

[0075] Due to the existence of gas cross-interference, the signal mask output by a certain sensor contains the cross-interference of other gases. For example, the cross-interference coefficient of carbon monoxide gas to hydrogen sulfide sensor is 0.1. That is, if the sensor's reaction value is z in an environment with an oxygen sulfide gas concentration of C, then the reaction value is 0.1z in the same concentration of carbon monoxide gas. The smaller the cross-interference coefficient, the better the selectivity of the gas sensor. The removal of cross-interference by a single gas detection device can only rely on the selectivity of the gas sensor, which fundamentally limits the improvement of the accuracy of a single gas detection device. The four-way sensor module has multiple gas sensors at the same time. Cross-interference can be removed by calculating multiple signal quantities. The minimum resolution diameter of dust of the four-way sensor module is 0.3 microns, and the minimum resolution of formaldehyde is 0.001 mg / m 3 , ultra-quiet, zero false alarm rate, real-time response, accurate data, all-round shielding, strong anti-interference ability, air inlet and outlet on the same plane, and real-time output of formaldehyde data.

[0076] In an embodiment of the present invention, the second formaldehyde sensor applies automotive-grade fuel cell solid electrolyte to formaldehyde detection, which is truly solid, leak-free, and does not dry up. It is not affected by interferences such as low-concentration alcohols in the environment, and can achieve selective detection of formaldehyde gas in normal living environments. It is suitable for scenarios with high storage temperatures such as air conditioners and automobiles.

[0077] In an embodiment of the present invention, the K210 main control module includes a RISC-V 64-bit dual-core CPU, and each core has an independent built-in FPU. The K210 main control module has a fast Fourier transform accelerator, which can perform high-performance complex FFT calculations. The K210 main control module has built-in AES and SHA256 algorithm accelerators, which can provide basic security functions. The K210 main control module has high-performance, low-power SRAM, and powerful DMA, and has excellent performance in data throughput. The K210 main control module has peripheral units, namely: DVP, JTAG, OTP, FPIOA, GPIO, UART, SPI, RTC, I 2 S, I 2 C. WDT, Timer and PWM.

[0078] In an embodiment of the present invention, the central processing unit is equipped with a dual-core 64-bit high-performance, low-power CPU based on RISC-V ISA.

[0079] The neural network processor supports fixed-point models trained using mainstream training frameworks according to specific constraints. It also supports individual configuration of parameters for each convolutional neural network layer, including the number of input and output channels, and the row and column widths and heights of each layer. The neural network processor supports two convolution kernels, 1x1 and 3x3, and supports a maximum neural network parameter size of 5.5 MiB to 5.9 MiB in real-time operation. In an embodiment of the present invention, the central processing unit includes a logistic regression unit that returns the probability of exceeding the gas content standard for dry-type transformer characteristic decomposition in a mixed gas. The logistic regression unit uses a cross-entropy loss function and employs stochastic gradient descent to iteratively determine the parameters that minimize the cross-entropy loss.

[0080] The formula for output probability is:

[0081]

[0082] Among them, x1, x2, x3, and x4 are the data collected by the four gas sensors, w1, w2, w3, w4, and b are the parameters of the model. The purpose of training the model is to obtain these parameters. When P(x) is close to 1, it can be considered that the characteristic decomposition gas content of the dry-type transformer has exceeded the standard. The closer P(x) is to 1, the more serious the characteristic decomposition gas content exceeds the standard. It is necessary to confirm the operating status of the dry-type transformer and check whether there is overheating fault or quality problem.

[0083] The central processing unit includes the classification algorithm, and the performance of the algorithm is evaluated through confusion matrix, precision, recall rate, and F1 score.

[0084] Confusion matrix: Each column of the confusion matrix represents the predicted category, and the total number of each column represents the number of data predicted to be of that category; each row represents the true category of the data, and the total number of data in each row represents the number of data instances of that category.

[0085] Precision: It indicates the proportion of sample data predicted to be alarms that are actually alarm samples, that is, the probability that the characteristic decomposition gas content in the environment exceeds the standard when the device alarms.

[0086] Call rate: It indicates the proportion of sample data that are actually alarm samples that are predicted to be alarm samples, that is, the probability that the device will alarm when the characteristic decomposition gas content in the environment exceeds the standard.

[0087] F1 score: represents the harmonic mean of precision and recall rate.

[0088]

[0089] From the formula of the F1 score, we can see that it takes into account both the precision and recall rate of the classification model. The closer the F1 score is to 1, the better the classification effect of the model.

[0090] In one embodiment, a training set is input into the neural network processor. Excluding duplicate samples, the training set contains a total of 785 samples, including 386 alarm data samples and 399 non-alarm data samples.

[0091] Confusion Matrix:

[0092]

[0093] Accuracy: 100.00%

[0094] Recall: 100.00%

[0095] F1 score: 100.00%

[0096] The four model evaluation metrics above indicate that the model correctly evaluated all 785 samples in the training set. The neural network processor then fed the test set, which contained 337 samples after removing duplicates. These included 175 alarm data samples and 162 non-alarm data samples.

[0097] Confusion Matrix:

[0098]

[0099] Accuracy: 99.15%

[0100] Summon rate: 100.00%

[0101] F1 score: 100.00%

[0102] Through the above four model evaluation indicators, it can be seen that the model correctly classified 334 out of 337 samples in the test set; misclassified 3, of which 0 were alarm data predicted as non-alarm data and 3 were non-alarm data predicted as alarm data.

[0103] Preferably, the device further comprises:

[0104] an alarm, connected to the main control module, for issuing an alarm when the environmental state is determined based on the probability to be a state where the characteristic decomposition gas content exceeds the standard;

[0105] Display module, used to display formaldehyde concentration data, carbon monoxide concentration data, output probability of dry-type transformer characteristic decomposition gas content exceeding the standard, environmental status and alarm working status;

[0106] A power supply module is used to provide power support for the dry-type transformer characteristic decomposition gas detection device.

[0107] Preferably, the alarm is a buzzer;

[0108] The display module is a display including a liquid crystal display screen, the liquid crystal screen is an LED resistive touch screen, and the LED resistive touch screen can dynamically display formaldehyde concentration data and carbon monoxide concentration data.

[0109] In the present invention, the dry-type transformer characteristic decomposition gas detection device further includes: an alarm, a display module and a power supply module.

[0110] An alarm, connected to the central processing unit, is capable of issuing an alarm based on the output probability. The alarm includes a buzzer. A display connected to the central processing unit is configured to display the first data, the second data, the third data, the fourth data, and the output probability. The display includes a liquid crystal display. A power module is connected to the central processing unit to provide power to the dry-type transformer characteristic decomposition gas detection device.

[0111] In the present invention, the dry-type transformer characteristic decomposition gas detection device is 1m away from the dry-type transformer in the first 5 minutes. At this time, the sensor acquisition signal is basically a noise signal. After the 5th minute, the transformer ends the temperature rise test. At the same time, the gas sensor is close to the inner side of the transformer AB two-phase winding to detect whether there is formaldehyde gas on the winding epoxy resin. Then, the formaldehyde gas outside the A-phase winding, the transformer core, the rubber gasket, and the busbar are measured in turn. The response curves of the three formaldehyde modules are as follows: Figure 3 As shown in the figure, the larger the sensor response amplitude is, the higher the formaldehyde gas concentration is.

[0112] Figure 4 FIG. 4 is a flow chart of a dry-type transformer characteristic decomposition gas detection method 400 according to an embodiment of the present invention. Figure 4 As shown, the dry-type transformer characteristic decomposition gas detection method 400 provided by the embodiment of the present invention starts from step 401. In step 401, the formaldehyde data detection module detects the formaldehyde concentration in the environment and obtains formaldehyde concentration data.

[0113] Preferably, the formaldehyde data detection module includes: a first formaldehyde sensor, a second formaldehyde sensor and a third formaldehyde sensor; wherein the first formaldehyde sensor is a semiconductor type formaldehyde sensor; the second formaldehyde sensor is equipped with an automotive-grade fuel cell solid electrolyte; the third formaldehyde sensor is a fuel cell type electrochemical sensor, in which formaldehyde and oxygen can undergo corresponding redox reactions on the working electrode and the counter electrode and release charges to form current.

[0114] In step 402 , the carbon monoxide data detection module detects the carbon monoxide concentration in the environment and obtains carbon monoxide concentration data.

[0115] Preferably, the method further comprises:

[0116] The AD conversion module converts the formaldehyde concentration data and the carbon monoxide concentration data into digital signals, and sends the formaldehyde concentration data and the carbon monoxide concentration data converted into digital signals to the main control module.

[0117] In step 403, the main control module determines the output probability of the dry-type transformer characteristic decomposition gas content exceeding the standard according to the formaldehyde concentration data and the carbon monoxide concentration data, and determines the environmental state based on the probability; wherein the environmental state includes: a characteristic decomposition gas content exceeding the standard state and a characteristic decomposition gas content normal state.

[0118] Preferably, the main control module determines the output probability of the dry-type transformer characteristic decomposition gas content exceeding the standard according to the formaldehyde concentration data and the carbon monoxide concentration data, including:

[0119] The central processing unit determines the output probability of the dry-type transformer characteristic decomposition gas content exceeding the standard by the following method, including:

[0120]

[0121] Where P(x) is the output probability of the dry-type transformer characteristic decomposition gas content exceeding the standard; x i is the gas concentration data detected by the i-th sensor, which includes a formaldehyde sensor and a carbon monoxide sensor; ω i is the weight parameter corresponding to the gas concentration data detected by the i-th sensor, and b is the preset coefficient; the weight parameter and preset coefficient are obtained by neural network learning based on historical formaldehyde concentration data, carbon monoxide concentration data and environmental status data using a neural network processor.

[0122] Preferably, the method further comprises:

[0123] When the alarm determines based on the probability that the environmental state is a state where the content of the characteristic decomposition gas of the dry-type transformer exceeds the standard, the alarm issues an alarm;

[0124] The display module displays formaldehyde concentration data, carbon monoxide concentration data, output probability of dry-type transformer characteristic decomposition gas content exceeding the standard, environmental status and alarm working status;

[0125] The power supply module provides power supply support for the dry-type transformer characteristic decomposition gas detection device.

[0126] The dry-type transformer characteristic decomposition gas detection method 400 according to an embodiment of the present invention corresponds to the dry-type transformer characteristic decomposition gas detection device 100 according to another embodiment of the present invention, and will not be described in detail here.

[0127] The invention has been described above with reference to a few embodiments. However, it is readily apparent to a person skilled in the art that other embodiments than the ones disclosed above are equally within the scope of the invention, as defined by the appended patent claims.

[0128] Generally, all terms used in the claims are to be interpreted according to their ordinary meaning in the technical field, unless explicitly defined otherwise herein. All references to "a / the [means, component, etc.]" are to be interpreted openly as referring to at least one instance of the means, component, etc., unless explicitly stated otherwise. The steps of any method disclosed herein do not necessarily need to be performed in the exact order disclosed, unless explicitly stated otherwise.

[0129] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0130] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0131] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0132] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0133] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered by the scope of protection of the claims of the present invention.

Claims

1. A dry-type transformer characteristic decomposition gas detection device, characterized in that: The device includes: a formaldehyde data detection module, a carbon monoxide data detection module and a main control module; wherein, The formaldehyde data detection module is connected to the main control module and is used to detect the formaldehyde concentration in the environment and obtain formaldehyde concentration data; The carbon monoxide data detection module is connected to the main control module and is used to detect the carbon monoxide concentration in the environment and obtain carbon monoxide concentration data; The main control module is used to determine the output probability of the dry-type transformer characteristic decomposition gas content exceeding the standard based on the formaldehyde concentration data and the carbon monoxide concentration data, and determine the environmental state based on the probability; wherein the environmental state includes: a characteristic decomposition gas content exceeding the standard state and a characteristic decomposition gas content normal state; The formaldehyde data detection module includes: a first formaldehyde sensor, a second formaldehyde sensor, and a third formaldehyde sensor; wherein the first formaldehyde sensor is a semiconductor-type formaldehyde sensor; the second formaldehyde sensor has a built-in automotive-grade fuel cell solid electrolyte; and the third formaldehyde sensor is a fuel cell-type electrochemical sensor, in which formaldehyde and oxygen can undergo corresponding redox reactions on the working electrode and the counter electrode, releasing charge to form a current; The main control module determines the output probability of the dry-type transformer characteristic decomposition gas content exceeding the standard according to the formaldehyde concentration data and the carbon monoxide concentration data, including: The central processing unit is configured to determine the output probability of the dry-type transformer characteristic decomposition gas content exceeding the standard by the following method, including: Where P(x) is the output probability of the dry-type transformer characteristic decomposition gas content exceeding the standard; x i is the gas concentration data detected by the i-th sensor, which includes a formaldehyde sensor and a carbon monoxide sensor; ω i is the weight parameter corresponding to the gas concentration data detected by the i-th sensor, and b is the preset coefficient; the weight parameter and preset coefficient are obtained by neural network learning based on historical formaldehyde concentration data, carbon monoxide concentration data and environmental status data using a neural network processor.

2. The device according to claim 1, characterized in that The device further comprises: The AD conversion module is connected to the input ends of the formaldehyde data detection module, the carbon monoxide data detection module and the main control module, respectively, and is used to convert the formaldehyde concentration data and the carbon monoxide concentration data into digital signals, and send the formaldehyde concentration data and the carbon monoxide concentration data converted into digital signals to the main control module.

3. The device according to claim 1, characterized in that The neural network processor includes: a convolution unit, a batch normalization unit, an activation unit, and a pooling operation unit. The neural network processor is a fixed-point model trained based on a mainstream training framework according to specific restriction rules. It supports separate configuration of the parameters of each layer of the convolutional neural network, including the number of input and output channels, and the input and output row width and column height. The neural network processor supports two convolution kernels, 1x1 and 3x3. When working in real time, the maximum supported neural network parameter size is 5.5 MiB to 5.9 MiB.

4. The device according to claim 1, characterized in that The device further comprises: an alarm, connected to the main control module, for issuing an alarm when it is determined based on the probability that the environmental state is a state where the content of characteristic decomposition gas of the dry-type transformer exceeds the standard; Display module, used to display formaldehyde concentration data, carbon monoxide concentration data, output probability of dry-type transformer characteristic decomposition gas content exceeding the standard, environmental status and alarm working status; A power supply module is used to provide power support for the dry-type transformer characteristic decomposition gas detection device.

5. The device according to claim 4, characterized in that The alarm is a buzzer; The display module is a display including a liquid crystal display screen, the liquid crystal screen is an LED resistive touch screen, and the LED resistive touch screen can dynamically display formaldehyde concentration data and carbon monoxide concentration data.

6. A dry-type transformer characteristic decomposition gas detection method, characterized in that: The method comprises: The formaldehyde data detection module detects the formaldehyde concentration in the environment and obtains formaldehyde concentration data; The carbon monoxide data detection module detects the carbon monoxide concentration in the environment and obtains carbon monoxide concentration data; The main control module determines an output probability of the dry-type transformer characteristic decomposition gas content exceeding the standard according to the formaldehyde concentration data and the carbon monoxide concentration data, and determines an environmental state based on the probability; wherein the environmental state includes: a characteristic decomposition gas content exceeding the standard state and a characteristic decomposition gas content normal state; The formaldehyde data detection module includes: a first formaldehyde sensor, a second formaldehyde sensor, and a third formaldehyde sensor; wherein the first formaldehyde sensor is a semiconductor-type formaldehyde sensor; the second formaldehyde sensor has a built-in automotive-grade fuel cell solid electrolyte; and the third formaldehyde sensor is a fuel cell-type electrochemical sensor, in which formaldehyde and oxygen can undergo corresponding redox reactions on the working electrode and the counter electrode, releasing charge to form a current; The main control module determines the output probability of the dry-type transformer characteristic decomposition gas content exceeding the standard according to the formaldehyde concentration data and the carbon monoxide concentration data, including: The central processing unit determines the output probability of the dry-type transformer characteristic decomposition gas content exceeding the standard by the following method, including: Where P(x) is the output probability of the dry-type transformer characteristic decomposition gas content exceeding the standard; x i is the gas concentration data detected by the i-th sensor, which includes a formaldehyde sensor and a carbon monoxide sensor; ω i is the weight parameter corresponding to the gas concentration data detected by the i-th sensor, and b is the preset coefficient; the weight parameter and preset coefficient are obtained by neural network learning based on historical formaldehyde concentration data, carbon monoxide concentration data and environmental status data using a neural network processor.

7. The method according to claim 6, characterized in that The method further comprises: The AD conversion module converts the formaldehyde concentration data and the carbon monoxide concentration data into digital signals, and sends the formaldehyde concentration data and the carbon monoxide concentration data converted into digital signals to the main control module.

8. The method according to claim 6, characterized in that The method further comprises: The alarm device issues an alarm when it is determined based on the probability that the environmental state is a state where the characteristic decomposition gas content exceeds the standard; The display module displays formaldehyde concentration data, carbon monoxide concentration data, output probability of dry-type transformer characteristic decomposition gas content exceeding the standard, environmental status and alarm working status; The power supply module provides power supply support for the dry-type transformer characteristic decomposition gas detection device.

Images