Information fusion method, system and multi-parameter sensing device

Abstract

The present invention relates to an information fusion method, system and multi-parameter sensing device, the method comprising the following steps: obtaining historical collection data of each sensor of a transformer, constructing a characteristic parameter historical data set; obtaining the priority of each characteristic parameter relative to different fault types; selecting characteristic parameters with a larger priority for each fault type according to the priority to form a fault characteristic parameter set of the fault type; obtaining the collection data of each sensor of the transformer within a preset time period before the current sampling moment and the current sampling moment, constructing a characteristic parameter current data set; fusing the characteristic parameter current data set according to the fault characteristic parameter set, and obtaining a fault fusion result and a state fusion result. The present invention obtains fault fusion results for different fault types and real-time state fusion results by fusing the collection data of multiple sensors, which facilitates further application and analysis of the data.

Description

Technical Field

[0001] The present invention relates to the field of multi-parameter sensing technology, and in particular to an information fusion method, system and multi-parameter sensing device. Background Art

[0002] The transformer is one of the core equipment of the power system. It plays the role of energy transfer and voltage conversion in the power system and is crucial to the transmission and distribution of electric energy. Monitoring the status of the transformer can detect latent faults in a timely and accurate manner, avoid the deterioration and expansion of faults, effectively reduce the equipment failure rate, reduce power outage time, save operation and maintenance costs, and help improve the safety, reliability and economy of the power system. The inside of the transformer is a complex multi-physical field environment. The process of the transformer gradually developing from defects to failures will cause changes in the multi-physical fields. By detecting the changes in the spatiotemporal distribution of these physical fields, we can determine the location, type, severity and development trend of the transformer defects, timely control the operating status of the transformer and take corresponding dispatching measures.

[0003] With the development of Internet of Things technology, more and more sensors are used for online monitoring of transformers. At present, the operating status and defect characteristics of transformers are mainly monitored through dissolved gas in oil, partial discharge, ground current, and hot spot temperature. These parameters have slow response speed and cannot directly reflect the degree of defects, resulting in occasional transformer fires or even explosions. In the rapid development stage before the fault, multiple physical parameters such as leakage magnetic flux, strain, flow velocity, pressure and sound field inside the transformer undergo sudden changes. They can quickly, sensitively and intuitively reflect the common harmful defects in the transformer. Increasing the digital perception of these parameters can fully grasp the parameters of the transformer equipment status and fault characteristics, and block the deterioration of transformer faults in advance, but there is currently a lack of relevant sensors.

[0004] The use of multi-parameter sensors can obtain more spatiotemporal distribution characteristics of physical quantities, which is conducive to comprehensive control of the status of the transformer. In addition, multiple physical quantities can be jointly diagnosed and cross-complemented, making the transformer status assessment more accurate and reliable. However, the existing multi-parameter sensing technology obtains a large amount of data and only obtains and stores the original data, which poses a serious obstacle to further application and analysis in the future. Summary of the invention

[0005] In view of this, the present invention provides an information fusion method, system and multi-parameter sensing device to achieve the fusion of original data of multi-parameter sensing and facilitate further application and analysis of the data.

[0006] To achieve the above object, the present invention provides the following solutions:

[0007] An information fusion method for transformer multi-parameter perception, the method comprising the following steps:

[0008] Obtain historical collected data of each sensor of the transformer and construct a characteristic parameter historical data set; wherein each sensor in the transformer corresponds to a characteristic parameter, the collected data of the sensor is used as the data of the characteristic parameter, and the change of the collected data of the sensor along with the evolution process of the defect of the transformer is used as the data change of the characteristic parameter;

[0009] Calculate the information gain of each characteristic parameter for different fault types of the transformer based on the characteristic parameter historical data set, and obtain the priority of each characteristic parameter relative to different fault types;

[0010] According to the priority of each characteristic parameter relative to different fault types, for each fault type, a characteristic parameter with a larger priority is selected to form a fault characteristic parameter set of the fault type;

[0011] Acquire the collected data of each sensor of the transformer at the current sampling moment and the preset time period before the current sampling moment, and construct a current data set of characteristic parameters;

[0012] For each fault type, the current data set of the characteristic parameters is fused according to the fault characteristic parameter set of the fault type to obtain a fault fusion result of the fault type; the fault fusion result of the fault type is the data change of each characteristic parameter in the fault characteristic parameter set of the fault type represented in the current data set of the characteristic parameters;

[0013] The data of all feature parameters at the current sampling time in the current data set of the fused feature parameters are taken as the state fusion result.

[0014] Optionally, the acquiring of historical data collected by each sensor of the transformer and the construction of a characteristic parameter historical data set specifically include:

[0015] Acquire historical data collected by each sensor of the transformer as historical data of each characteristic parameter;

[0016] The historical data collected by each sensor is analyzed by using wavelet transform method, adaptive filtering method or continuous refinement Fourier spectrum analysis method, and the changes of the historical data collected by each sensor along with the evolution process of the defects of the transformer are obtained as the historical data changes of each characteristic parameter;

[0017] Construct a feature parameter historical data set including the historical data of each feature parameter and the changes in the historical data.

[0018] An information fusion system for transformer multi-parameter sensing, the system comprising:

[0019] A characteristic parameter historical data set construction module is used to obtain the historical collection data of each sensor of the transformer and construct a characteristic parameter historical data set; wherein each sensor in the transformer corresponds to a characteristic parameter, the collection data of the sensor is used as the data of the characteristic parameter, and the change of the collection data of the sensor along with the evolution process of the defect of the transformer is used as the data change of the characteristic parameter;

[0020] A priority determination module, used to calculate the information gain of each characteristic parameter for different fault types of the transformer based on the characteristic parameter historical data set, and obtain the priority of each characteristic parameter relative to different fault types;

[0021] Determine the fault characteristic parameter set, according to the priority of each characteristic parameter relative to different fault types, for each fault type, select the characteristic parameter with a larger priority to form the fault characteristic parameter set of the fault type;

[0022] The characteristic parameter current data set construction module is used to obtain the collected data of each sensor of the transformer within the preset time period before the current sampling time and the current sampling time, and construct the characteristic parameter current data set;

[0023] A fault fusion module is used to fuse the current data set of characteristic parameters for each fault type according to the fault characteristic parameter set of the fault type to obtain a fault fusion result of the fault type; the fault fusion result of the fault type is the data change of each characteristic parameter in the fault characteristic parameter set of the fault type represented in the current data set of characteristic parameters;

[0024] The state fusion module is used to fuse the data of all feature parameters at the current sampling moment in the current data set of feature parameters as the state fusion result.

[0025] Optionally, the feature parameter historical data set construction module specifically includes:

[0026] The historical data acquisition submodule of the characteristic parameter is used to acquire the historical data collected by each sensor of the transformer as the historical data of each characteristic parameter;

[0027] The historical data feature acquisition submodule is used to analyze the historical data collected by each sensor by using wavelet transform method, adaptive filtering method or continuous refinement Fourier spectrum analysis method, and obtain the changes of the historical data collected by each sensor along with the evolution process of the transformer defects as the historical data changes of each characteristic parameter;

[0028] The feature parameter historical data set construction submodule is used to construct a feature parameter historical data set including the historical data of each feature parameter and the changes in the historical data.

[0029] A transformer multi-parameter sensing device, the device comprising a plurality of sensors, a data acquisition terminal, a data transmission device and a server;

[0030] A plurality of the sensors are arranged on the transformer based on a sensor arrangement mode;

[0031] The plurality of sensors are connected to the acquisition terminal; the acquisition terminal is used to acquire the collected data of each sensor;

[0032] The acquisition terminal sends the acquired data of each sensor to the server through a data transmission device;

[0033] The server is used to fuse the collected data of each sensor using the above method.

[0034] Optionally, the sensor arrangement method includes the following steps:

[0035] Determine a first installation position of a first type of sensor; the first type of sensor is a sensor with a fixed installation position, including an oil chromatography sensor, a top oil temperature sensor, a core grounding current sensor, a high-frequency current sensor, a vibration sensor, and a flow rate sensor;

[0036] Based on the spatiotemporal distribution law of the second type of sensor, the second installation position of the second type of sensor is determined; the second type of sensor is a sensor with a non-fixed installation position, including a partial discharge ultra-high frequency sensor, a pressure sensor, a leakage magnetic sensor, an arc discharge ultrasonic sensor and an arc discharge audible sound sensor; the second installation position includes a mandatory installation position and an optional installation position, and the spatiotemporal distribution law is, for each second type of sensor, a corresponding distribution relationship between the size of the signal response amount to different fault types of the transformer and the installation position of the second type of sensor.

[0037] Optionally, determining the second installation position of the second type of sensor based on the spatiotemporal distribution law of the second type of sensor specifically includes:

[0038] For each second type of sensor, determining the permitted installation position of the second type of sensor according to the performance parameters of the second type of sensor; the performance parameters include: sensing principle, probe structure, and signal extraction method;

[0039] For each second-type sensor, according to the temporal and spatial distribution law of the second-type sensor, determine the permitted installation position where the second-type sensor can generate corresponding signals when different fault types occur inside the transformer as the mandatory installation position of the second-type sensor;

[0040] For each second-category sensor, based on the spatiotemporal distribution law of the second-category sensors, with the goal of maximizing the detection range, simplifying the installation, and minimizing the installation cost, an exhaustive method is used to determine the optional installation locations of the second-category sensors among the permitted installation locations other than the mandatory installation locations.

[0041] Optionally, the data acquisition terminal acquires analog signals by triggering optical signal pulses or electrical signal pulses, and acquires digital signals by message switching, packet switching or circuit switching.

[0042] Optionally, the data acquisition terminal includes a pulse generator, a data acquisition module, a back-end demodulation device and a switch;

[0043] The sensor outputting analog signals is connected to the data acquisition module; the pulse generator is connected to the acquisition module, and the pulse generator is used to output optical signal pulses or electrical signal pulses to trigger the data acquisition module to collect analog signals;

[0044] The sensor that outputs the digital signal is connected to the data acquisition module through a back-end demodulation device and a switch.

[0045] Optionally, the data transmission device transmits the collected data of the sensor in the following manner:

[0046] Extracting time information from the collected data of each sensor respectively as the first time information of each sensor;

[0047] Respectively acquiring time information when a data collection instruction is sent to each of the sensors as second time information of each sensor;

[0048] respectively calculating the difference between the first time information and the second time information of each of the sensors as the action time of each of the sensors;

[0049] Performing time correction on the collected data of each sensor by using terminal timing or clock timing according to the action time of each sensor;

[0050] The corrected collected data of each sensor is sent to the server.

[0051] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0052] The present invention discloses an information fusion method for transformer multi-parameter perception, the method comprising the following steps: obtaining historical acquisition data of each sensor of the transformer, constructing a characteristic parameter historical data set; calculating the information gain of each characteristic parameter for different fault types of the transformer based on the characteristic parameter historical data set, and obtaining the priority of each characteristic parameter relative to different fault types; selecting characteristic parameters with a larger priority for each fault type to form a fault characteristic parameter set of the fault type according to the priority of each characteristic parameter relative to different fault types; obtaining the acquisition data of each sensor of the transformer within a preset time period before and before the current sampling time, and constructing a characteristic parameter current data set; for each fault type, fusing the characteristic parameter current data set according to the fault characteristic parameter set of the fault type, and obtaining the fault fusion result of the fault type; fusing the data of all characteristic parameters at the current sampling time in the characteristic parameter current data set as a state fusion result. The present invention obtains fault fusion results for different fault types and real-time state fusion results by fusing the acquisition data of multiple sensors, which facilitates further application and analysis of the data.

[0053] The present invention provides a sensor arrangement method to achieve an optimized arrangement of installation positions of multi-parameter sensors, thereby reducing the number of sensors and improving the detection effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technical personnel in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0055] Figure 1 A flowchart of an information fusion method for transformer multi-parameter perception provided in Example 1 of the present invention;

[0056] Figure 2 A schematic diagram of characteristic parameter acquisition provided in Example 1 of the present invention;

[0057] Figure 3 A schematic diagram of the selection of a characteristic parameter acquisition method provided in Example 1 of the present invention;

[0058] Figure 4 A schematic diagram of obtaining the evolution law of characteristic parameters along with the development of defects provided in Example 1 of the present invention;

[0059] Figure 5 A flowchart of information gain calculation provided in Example 1 of the present invention;

[0060] Figure 6 A schematic diagram of the corresponding relationship between typical faults and fault characteristic parameters provided in Example 1 of the present invention;

[0061] Figure 7 A schematic diagram of feature parameter fusion provided in Example 1 of the present invention;

[0062] Figure 8 A schematic diagram of a sensor arrangement method provided in Example 2 of the present invention;

[0063] Fig. 9 A schematic diagram of sensor types provided in Example 2 of the present invention;

[0064] Fig.10 A schematic diagram of the principle of the spatiotemporal distribution law provided in Example 2 of the present invention;

[0065] Fig.11 A schematic diagram of the selection of the required installation position provided in Example 2 of the present invention;

[0066] Fig.12 A schematic diagram of the selection of optional installation positions provided in Embodiment 2 of the present invention;

[0067] Fig.13 A flowchart of selecting a sensor installation solution provided in Example 2 of the present invention;

[0068] Fig.14 A schematic diagram of the exhaustive method provided in Example 2 of the present invention;

[0069] Fig.15 This is a schematic diagram of the sensor timing synchronization provided in Example 2 of the present invention. DETAILED DESCRIPTION

[0070] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technical personnel in this field without creative work are within the scope of protection of the present invention.

[0071] The purpose of the present invention is to provide an information fusion method, system and multi-parameter sensing device to achieve the fusion of original data of multi-parameter sensing and facilitate further application and analysis of the data.

[0072] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0073] Example 1

[0074] like Figure 1 As shown, Embodiment 1 of the present invention provides an information fusion method for transformer multi-parameter perception, the method comprising the following steps:

[0075] Step 101, obtain the historical collection data of each sensor of the transformer and construct a historical data set of characteristic parameters; wherein each sensor in the transformer corresponds to a characteristic parameter, the collection data of the sensor is used as the data of the characteristic parameter, and the change of the collection data of the sensor along with the evolution process of the defects of the transformer is used as the data change of the characteristic parameter.

[0076] Step 101, the historical data of each sensor of the transformer is obtained to construct a historical data set of characteristic parameters, specifically including: obtaining the historical data of each sensor of the transformer as the historical data of each characteristic parameter; using wavelet transform method, adaptive filtering method or continuous refinement Fourier spectrum analysis method to analyze the historical data of each sensor, and obtaining the change of the historical data of each sensor along with the evolution process of the defects of the transformer as the historical data change of each characteristic parameter; constructing a historical data set of characteristic parameters including the historical data of each characteristic parameter and the change of the historical data.

[0077] Step 102: Calculate the information gain of each characteristic parameter for different fault types of the transformer based on the characteristic parameter historical data set, and obtain the priority of each characteristic parameter relative to different fault types.

[0078] Step 103 : According to the priority of each characteristic parameter relative to different fault types, for each fault type, a characteristic parameter with a larger priority is selected to form a fault characteristic parameter set of the fault type.

[0079] Steps 102-103 are specifically as follows:

[0080] Step 1: According to the defect development mechanism of the transformer, the relationship between the defect development process and the change law of each parameter is established according to the process shown in the figure below, and the characteristics of each parameter corresponding to defects of different locations, types and degrees are obtained.

[0081] like Figure 2 As shown, the transmission data of each sensor is represented by parameters 1, 2, ..., n. Each parameter changes with the defects of different positions, types and degrees inside the transformer. These changes can be extracted as characteristic signals to achieve tracking control, and then the collected raw data is analyzed to obtain the data of each characteristic parameter.

[0082] Step 2: According to the relationship between the signal characteristics of the parameters and the characteristic parameters, the extraction method of each characteristic parameter of each sensor is established;

[0083] Feature extraction is a further processing of the sensor detection signal. It extracts parameters related to state changes from a large number of signals and uses mathematical statistics, signal processing (FFT, wavelet analysis, fractal and chaos) and other methods to extract signal features. On the one hand, it can improve the signal-to-noise ratio of the signal, and on the other hand, it can reduce the amount of data processing in the later stage and improve the operation speed. The quality of the extracted feature parameters has a very important impact on the performance and reliability of the monitoring system.

[0084] like Figure 3 As shown in the figure, characteristic parameters are selected according to the principles of high sensitivity, good stability, good correspondence, small calculation amount and small database capacity. The parameters are analyzed and detected based on wavelet transform, adaptive filtering technology and continuous refinement Fourier spectrum analysis. The frequency domain characteristics of wavelet multi-scale analysis are used to accurately locate the time of parameter changes. Stable parameter signals are found for adaptive filtering and continuous refinement Fourier transform to perform spectrum analysis and extract their change characteristics.

[0085] Step 3: Based on the relationship between defects and faults in the first step and the features extracted in the second step, the state space of transformer defect evolution is established;

[0086] like Figure 4 As shown in the figure, after the first and second steps obtain the extraction method of the characteristic parameters of each sensor, the data obtained by the sensor is analyzed and processed, the similarities and differences of the parameters in each state of the transformer are summarized, and the evolution law of each parameter with the development of the defect is obtained, which is represented by Ln. These change laws together constitute the state space of the transformer defect evolution. The state space analysis and evaluation of the transformer defect evolution is used to provide a criterion for equipment maintenance decisions. According to the evolution law, the time, location and degree of possible faults are predicted, and the faults that have occurred are located and the degree of fault development is judged according to the fault characteristics.

[0087] Step 4: Use information gain to quantify the priority of multi-sensor multi-feature parameters, and sort the parameters from high to low according to the priority to form a decision tree. The decision tree construction algorithm is as follows:

[0088] like Figure 5 As shown, the calculation of information gain includes the following steps:

[0089] (1) Calculate the initial entropy of the system;

[0090] The overall entropy of the dataset is:

[0091]

[0092] The entropy of feature A for data set D:

[0093]

[0094] Information gain of feature A:

[0095] g(D, A) = H(D) - H(D|A)

[0096] (2) List the corresponding tables according to the entropy value of the entire system corresponding to each parameter (i.e., detection method) and the obtained entropy reduction (i.e., information gain);

[0097] (3) The parameter with the largest entropy reduction of the entire system indicates that its information gain is the largest, and it serves as the root node of the decision tree. Similarly, other parameters corresponding to the lower levels can be calculated.

[0098] Step 5: According to the parameter priority of the fourth step, establish a feature parameter sub-set for each state of the transformer state space in the third step. Specifically, according to the analysis results of the characteristic parameters of the power transformer in the fourth step, for the defect states of different positions, types and degrees of the transformer, each characteristic parameter can obtain the corresponding information gain, sort them according to their priority, summarize and organize the fault detection methods involved, take each characteristic parameter as a fault characteristic parameter, and encode each characteristic parameter in sequence to obtain a fault characteristic parameter set. A fault is often reflected by several fault characteristic quantities, and a fault characteristic quantity may come from multiple fault states, so the selection and extraction of fault characteristic quantities is one of the difficult problems in fault diagnosis. There is ambiguity in the fault characteristic quantity, and representative and effective fault characteristic parameters need to be selected. The membership of various fault causes can be calculated based on the membership of certain features, and then the fault cause of the object can be determined by the threshold principle, the maximum membership principle or the proximity principle. On this basis, according to the corresponding relationship between each typical fault of the transformer and the fault characteristic parameter, the corresponding relationship between each typical fault and the fault characteristic parameter can be established, such as Figure 6 shown.

[0099] Step 104, acquiring the collected data of each sensor of the transformer during the current sampling moment and a preset time period before the current sampling moment, and constructing a current data set of characteristic parameters.

[0100] Step 105, for each fault type, fuse the current data set of characteristic parameters according to the fault characteristic parameter set of the fault type to obtain a fault fusion result of the fault type; the fault fusion result of the fault type is the data change of each characteristic parameter in the fault characteristic parameter set of the fault type represented in the current data set of characteristic parameters.

[0101] Step 106, fusing the data of all feature parameters at the current sampling moment in the current data set of feature parameters as the state fusion result.

[0102] Steps 105 and 106 specifically include:

[0103] According to the perception system instructions, the required multi-sensor multi-parameter feature parameter sub-set data is transmitted, including current data and historical data.

[0104] like Figure 7 As shown in the figure, the characteristic parameter subsets of each state in the transformer state space together constitute the corresponding state space subset. The state quantity of the transformer is obtained in real time by using computer technology, communication technology and sensor technology. Through their analysis and processing, they are sorted and summarized. After issuing the corresponding instructions, the data is transmitted. The transmission mainly consists of two parts. The first part is the data change of each parameter during the development of the transformer defect, and the second part is the data of each sensor characteristic parameter under each state. Then the operating state of the transformer is predicted. Through these data parameters, its current state can be predicted. If there is a fault, the fault type can be determined. This can not only avoid the further expansion of the fault, but also eliminate the fault in time to improve reliability.

[0105] Embodiment 1 of the present invention further provides an information fusion system for transformer multi-parameter sensing, the system comprising:

[0106] The characteristic parameter historical data set construction module is used to obtain the historical collection data of each sensor of the transformer and construct the characteristic parameter historical data set; wherein, each sensor in the transformer corresponds to a characteristic parameter, the collection data of the sensor is used as the data of the characteristic parameter, and the change of the collection data of the sensor with the evolution process of the defects of the transformer is used as the data change of the characteristic parameter.

[0107] The characteristic parameter historical data set construction module specifically includes: a characteristic parameter historical data acquisition submodule, which is used to acquire the historical collection data of each sensor of the transformer as the historical data of each characteristic parameter; a historical data feature acquisition submodule, which is used to analyze the historical collection data of each sensor by adopting a wavelet transform method, an adaptive filtering method or a continuous refinement Fourier spectrum analysis method, and acquire the changes of the historical collection data of each sensor with the evolution process of the defects of the transformer as the historical data changes of each characteristic parameter; a characteristic parameter historical data set construction submodule, which is used to construct a characteristic parameter historical data set including the historical data of each characteristic parameter and the changes of the historical data.

[0108] A priority determination module, used to calculate the information gain of each characteristic parameter for different fault types of the transformer based on the characteristic parameter historical data set, and obtain the priority of each characteristic parameter relative to different fault types;

[0109] Determine the fault characteristic parameter set, according to the priority of each characteristic parameter relative to different fault types, for each fault type, select the characteristic parameter with a larger priority to form the fault characteristic parameter set of the fault type;

[0110] The characteristic parameter current data set construction module is used to obtain the collected data of each sensor of the transformer within the preset time period before the current sampling time and the current sampling time, and construct the characteristic parameter current data set;

[0111] A fault fusion module is used to fuse the current data set of characteristic parameters for each fault type according to the fault characteristic parameter set of the fault type to obtain a fault fusion result of the fault type; the fault fusion result of the fault type is the data change of each characteristic parameter in the fault characteristic parameter set of the fault type represented in the current data set of characteristic parameters;

[0112] The state fusion module is used to fuse the data of all feature parameters at the current sampling moment in the current data set of feature parameters as the state fusion result.

[0113] Example 2

[0114] Embodiment 2 of the present invention provides a transformer multi-parameter sensing device, which includes multiple sensors, a data acquisition terminal, a data transmission device and a server; multiple sensors are arranged on the transformer based on the sensor arrangement method; multiple sensors are connected to the acquisition terminal; the acquisition terminal is used to acquire the collected data of each sensor; the acquisition terminal sends the collected data of each sensor to the server through the data transmission device; the server is used to fuse the collected data of each sensor using the method of embodiment 1.

[0115] The sensor arrangement method includes the following steps:

[0116] 1) Sensors with fixed installation positions: installed at designated locations without the need for optimized layout;

[0117] 2) Sensors with flexible installation positions: The installation position needs to be optimized;

[0118] like Figure 8 The specific method of optimizing the arrangement of the installation position shown includes the following steps:

[0119] Step 1: Determine the permissible installation position of the sensor based on the sensor's sensing principle, probe structure, and signal extraction method; Fig. 9 As shown in the figure, the sensors installed on the transformer are roughly divided into resistance, capacitance, photoelectric and other principles according to the sensing principle; the probe structure is roughly divided into metal type and sub-metal type; the signal extraction method is divided into several extraction methods such as optical signal extraction and electrical signal extraction. Based on these three conditions, it is decided whether to install it outside or inside the transformer, close to the fault or far away from the fault, and then determine the permitted installation position.

[0120] Step 2: Based on the generation mechanism and statistical results of typical faults inside the transformer, analyze the temporal and spatial distribution law of different physical parameters at the permitted installation position and the changing law of defects at different positions and different severity levels inside the transformer; when a fault occurs inside the transformer, sensors with different parameters at different permitted installation positions will respond to the fault and obtain signals related to the fault based on their own measurement principles. When the location and severity of the fault are different, the strength of the signal obtained by the sensor will also change with the location and severity. Time is reflected in the time of fault development, that is, the severity, and space is reflected in the location of the fault and the installation location of the sensor. This law is the temporal and spatial distribution law of the sensor, such as Fig.10 As shown. Based on this time-space distribution law, it is possible to determine which sensor at the permitted installation location can fully detect all faults and ensure safe and stable operation of the transformer; at the same time, it is possible to determine which sensor is installed to locate the fault. The subsequent three and four steps need to be selected according to the time-space distribution law.

[0121] Step 3: If Fig.11 As shown, on the basis of the second and first steps, the mandatory installation position of the sensor is determined in the permitted installation position of the transformer to ensure the safe and stable operation of the transformer. For the sensor Si that monitors any parameter i, the selection process of its mandatory installation position is as follows. For faults of different positions and severity occurring inside the transformer, the signal strengths generated by the sensors with the same parameters installed at various permitted positions are different due to the faults, and some positions cannot even detect the occurrence of faults. Therefore, through experimental comparison, when different faults occur inside the transformer, which positions of the sensors can all generate more obvious signals and realize the detection of faults, these positions are the mandatory installation positions. Sensors with different parameters determine their mandatory installation positions in turn through this method.

[0122] Step 4: If Fig.12 As shown, similarly, based on the second and third steps, the optional installation positions other than the mandatory installation positions are determined in the permitted installation positions to realize the use of sensors to distinguish different faults; the mandatory installation positions guarantee that all faults occurring inside the transformer can be detected to ensure safe and stable operation of the transformer. The optional positions are in addition to the mandatory installation positions. When the transformer has faults of different positions and severity, although this position cannot detect every fault, it is the position where the change is most obvious when the fault occurs. This position can assist the mandatory installation position in basically determining where the fault occurred. Compared with the mandatory installation position, the optional installation position reflects specificity.

[0123] Step 5: Use the exhaustive method to optimize the number and installation positions of sensors, compare the measurement effects of sensors at different optional installation positions for typical faults, select the position with the largest detection range, the simplest installation, and the lowest economic cost among the optional installations, and at the same time meet the necessary conditions for the selection of optional installation positions. With the goal of using as few sensors as possible and most sensitively responding to the characteristic laws of the physical field when the transformer has internal faults, obtain the installation schemes of sensors with different parameters. The determination process of any sensor installation scheme is as follows: Fig.13 As shown in the figure, the calculation process of the exhaustive method is as follows Fig.14 shown.

[0124] The data acquisition terminal collects analog signals by triggering optical signal pulses or electrical signal pulses, and collects digital signals by message exchange, packet exchange or circuit exchange. The data acquisition terminal includes a pulse generator, a data acquisition module, a back-end demodulation device and a switch; a sensor that outputs analog signals is connected to the data acquisition module; the pulse generator is connected to the acquisition module, and the pulse generator is used to output optical signal pulses or electrical signal pulses to trigger the data acquisition module to collect analog signals; the sensor that outputs digital signals is connected to the data acquisition module through a back-end demodulation device and a switch.

[0125] The specific method of data collection by the data collection terminal is as follows:

[0126] In order to reasonably use the sensor's collection capacity, it is necessary to synchronously collect various sensor signals. According to different sensor types, signal collection methods are divided into the following two categories:

[0127] 1) Analog signal: Select optical signal pulse or electrical signal pulse for triggering according to the type of sensor. The trigger signal includes sensor reset signal, parameter configuration signal, sensor start signal, typical time marking signal, etc., to realize the initialization of each sensor, acquisition parameter adjustment, start and typical time marking, so as to facilitate the data fusion in content (5).

[0128] 2) Digital signal: Digital signals can use message switching, packet switching or circuit switching according to conditions such as signal transmission volume and delay requirements. The communication protocol can be flexibly selected according to the port type of the sensor back-end demodulation device and the switch, and the protocol format is determined based on the sensor with the largest data transmission volume.

[0129] The protocol formats include but are not limited to Modbus, Profibus, and Ethernet, etc. The main differences are the transmission distance and transmission rate. For example, the transmission speed of Modbus communication protocol can reach more than 10Mb / s, and the transmission distance can reach about 1200 meters. Profibus communication protocol is mainly used for high-speed data transmission at the field device level. The master station periodically reads the input information of the slave station and periodically sends output information to the slave station, while the Modbus communication protocol can only process 247 addresses on a data link. In actual applications, the choice can be made based on the data transmission volume and transmission rate.

[0130] First, perform fault diagnosis according to the power system equipment, and then track and collect the operation status of the equipment. During the transmission process, the data is converted into analog and digital, compressed and packaged after certain preliminary processing, and then the corresponding exchange technology is adopted according to the data characteristics. The data of each sensor is compared, and the protocol format with the largest transmission volume is selected to ensure that the data will not be lost, and then transmitted to the corresponding processing and control center. After receiving some analysis signals related to the status of the relevant electronic system equipment, the data information processing center uses different processing methods corresponding to multiple signals to perform relevant analysis and processing.

[0131] Digital equipment has certain logical functions. In order for it to operate normally, each part is required to work according to a pre-defined logical program. There is a predetermined timing relationship between each signal.

[0132] The execution of the program must first take out the instruction code, decode it, and send out the control signal to complete the instruction under the action of the specified control signal. These logical relationships are completed under the action of the controller. The signals in the system are orderly information flows, and there is a strict timing relationship between them. Therefore, the most important thing in the test of digital circuits is to check whether the sequence of digital pulses and the timing relationship of the waveform meet the design requirements.

[0133] The specific manner in which the data transmission device transmits the collected data of the sensor includes:

[0134] The embodiment of the present invention needs to ensure the synchronization of each sensor during data transmission. The synchronization method can adopt the terminal to time the sensor or the clock to time the sensor. Different precision synchronization methods are adopted according to the strictness of the synchronization of the collected signal.

[0135] For the action time of each sensor, Fig.15The process shown is verified. First, the raw data of each sensor is collected and stored, the time information is extracted from the raw data and stored, the time value of the clock module is read, and the raw time value is compared with it to determine the action time of the sensor. On this basis, combined with the time synchronization methods used by different sensors, the update data of multiple sensors is obtained, the correction time data is added to each sensor, and the update data is sent to the required server respectively.

[0136] The solution in the embodiment of the present invention has the following advantages:

[0137] (1) In terms of sensor types, new sensors such as pressure, magnetic leakage, flow velocity, arc discharge audible sound and ultrasound are considered, which can quickly and sensitively respond to several common and harmful faults inside the transformer, so as to monitor the transformer status more comprehensively and improve the safety of the transformer;

[0138] (2) For sensors with flexible installation positions, an optimization method for sensor installation layout is proposed, which is conducive to using as few sensors as possible to most sensitively respond to the characteristic laws of the physical field when the transformer has internal faults;

[0139] (3) A multi-parameter and multi-sensor installation location suitable for common transformers is proposed;

[0140] (4) A method for real-time fusion of multi-sensor and multi-parameter data is proposed to save channels and improve transmission efficiency. It is conducive to realizing iterative perception, reducing the transmission of useless data, reducing the redundant interference of useless data in the cloud, and improving the accuracy of transformer status diagnosis.

[0141] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0142] This article uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only used to help understand the method and core ideas of the present invention. At the same time, for those skilled in the art, according to the ideas of the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. An information fusion method for transformer multi-parameter perception, characterized in that: The method comprises the following steps: Acquire historical collected data of each sensor of the transformer and construct a characteristic parameter historical data set; wherein each sensor in the transformer corresponds to a characteristic parameter, the collected data of the sensor is used as the data of the characteristic parameter, and the change of the collected data of the sensor along with the evolution process of the defect of the transformer is used as the data change of the characteristic parameter, and the characteristic parameter historical data set includes the historical data of the characteristic parameter and the historical data change; Calculate the information gain of each characteristic parameter for different fault types of the transformer based on the characteristic parameter historical data set, and obtain the priority of each characteristic parameter relative to different fault types; According to the priority of each characteristic parameter relative to different fault types, for each fault type, a characteristic parameter with a larger priority is selected to form a fault characteristic parameter set of the fault type; Acquire the collected data of each sensor of the transformer at the current sampling moment and within a preset time period before the current sampling moment, and construct a current data set of characteristic parameters; For each fault type, the current data set of the characteristic parameters is fused according to the fault characteristic parameter set of the fault type to obtain a fault fusion result of the fault type; the fault fusion result of the fault type is the data change of each characteristic parameter in the fault characteristic parameter set of the fault type represented in the current data set of the characteristic parameters; The data of all feature parameters at the current sampling time in the current data set of the fused feature parameters are taken as the state fusion result.

2. The information fusion method for transformer multi-parameter perception according to claim 1 is characterized in that: The acquisition of historical data collected by each sensor of the transformer and the construction of a characteristic parameter historical data set specifically include: Obtain historical data collected by each sensor of the transformer as historical data of each characteristic parameter; The historical data collected by each sensor is analyzed by using wavelet transform method, adaptive filtering method or continuous refinement Fourier spectrum analysis method, and the changes of the historical data collected by each sensor along with the evolution process of the defects of the transformer are obtained as the historical data changes of each characteristic parameter; Construct a feature parameter historical data set including the historical data of each feature parameter and the changes in the historical data.

3. An information fusion system for transformer multi-parameter sensing, characterized in that: The system comprises: A characteristic parameter historical data set construction module is used to obtain the historical collection data of each sensor of the transformer and construct a characteristic parameter historical data set; wherein each sensor in the transformer corresponds to a characteristic parameter, the collection data of the sensor is used as the data of the characteristic parameter, and the change of the collection data of the sensor along with the evolution process of the defect of the transformer is used as the data change of the characteristic parameter; the characteristic parameter historical data set includes the historical data of the characteristic parameter and the historical data change; A priority determination module, used to calculate the information gain of each characteristic parameter for different fault types of the transformer based on the characteristic parameter historical data set, and obtain the priority of each characteristic parameter relative to different fault types; Determine the fault characteristic parameter set, according to the priority of each characteristic parameter relative to different fault types, for each fault type, select the characteristic parameter with a larger priority to form the fault characteristic parameter set of the fault type; The characteristic parameter current data set construction module is used to obtain the collected data of each sensor of the transformer within the preset time period before the current sampling time and the current sampling time, and construct the characteristic parameter current data set; A fault fusion module is used to fuse the current data set of characteristic parameters for each fault type according to the fault characteristic parameter set of the fault type to obtain a fault fusion result of the fault type; the fault fusion result of the fault type is the data change of each characteristic parameter in the fault characteristic parameter set of the fault type represented in the current data set of characteristic parameters; The state fusion module is used to fuse the data of all feature parameters at the current sampling moment in the current data set of feature parameters as the state fusion result.

4. The information fusion system for transformer multi-parameter sensing according to claim 3 is characterized in that: The characteristic parameter historical data set construction module specifically includes: The historical data acquisition submodule of the characteristic parameter is used to acquire the historical data collected by each sensor of the transformer as the historical data of each characteristic parameter; The historical data feature acquisition submodule is used to analyze the historical data collected by each sensor by using wavelet transform method, adaptive filtering method or continuous refinement Fourier spectrum analysis method, and obtain the changes of the historical data collected by each sensor along with the evolution process of the transformer defects as the historical data changes of each characteristic parameter; The feature parameter historical data set construction submodule is used to construct a feature parameter historical data set including the historical data of each feature parameter and the changes in the historical data.

5. A transformer multi-parameter sensing device, characterized in that: The device includes a plurality of sensors, a data acquisition terminal, a data transmission device and a server; A plurality of the sensors are arranged on the transformer based on a sensor arrangement mode; The plurality of sensors are connected to the acquisition terminal; the acquisition terminal is used to acquire the collected data of each sensor; The acquisition terminal sends the acquired data of each sensor to the server through a data transmission device; The server is used to fuse the collected data of each sensor using the method described in any one of claims 1-2.

6. The transformer multi-parameter sensing device according to claim 5, characterized in that: The sensor arrangement method comprises the following steps: Determine a first installation position of a first type of sensor; the first type of sensor is a sensor with a fixed installation position, including an oil chromatography sensor, a top oil temperature sensor, a core grounding current sensor, a high-frequency current sensor, a vibration sensor, and a flow rate sensor; Based on the spatiotemporal distribution law of the second type of sensor, the second installation position of the second type of sensor is determined; the second type of sensor is a sensor with a non-fixed installation position, including a partial discharge ultra-high frequency sensor, a pressure sensor, a leakage magnetic sensor, an arc discharge ultrasonic sensor and an arc discharge audible sound sensor; the second installation position includes a mandatory installation position and an optional installation position, and the spatiotemporal distribution law is, for each second type of sensor, a corresponding distribution relationship between the size of the signal response amount to different fault types of the transformer and the installation position of the second type of sensor.

7. The transformer multi-parameter sensing device according to claim 6, characterized in that: The determining the second installation position of the second type of sensor based on the spatiotemporal distribution law of the second type of sensor specifically includes: For each second type of sensor, determining the permitted installation position of the second type of sensor according to the performance parameters of the second type of sensor; the performance parameters include: sensing principle, probe structure, and signal extraction method; For each second-type sensor, according to the temporal and spatial distribution law of the second-type sensor, determine the permitted installation position where the second-type sensor can generate corresponding signals when different fault types occur inside the transformer as the mandatory installation position of the second-type sensor; For each second-category sensor, based on the spatiotemporal distribution law of the second-category sensors, with the goal of maximizing the detection range, simplifying the installation, and minimizing the installation cost, an exhaustive method is used to determine the optional installation locations of the second-category sensors among the permitted installation locations other than the mandatory installation locations.

8. The transformer multi-parameter sensing device according to claim 5, characterized in that: The data acquisition terminal acquires analog signals by triggering optical signal pulses or electrical signal pulses, and acquires digital signals by message switching, packet switching or circuit switching.

9. The transformer multi-parameter sensing device according to claim 8, characterized in that: The data acquisition terminal includes a pulse generator, a data acquisition module, a back-end demodulation device and a switch; The sensor outputting analog signals is connected to the data acquisition module; the pulse generator is connected to the acquisition module, and the pulse generator is used to output optical signal pulses or electrical signal pulses to trigger the data acquisition module to collect analog signals; The sensor that outputs the digital signal is connected to the data acquisition module through a back-end demodulation device and a switch.

10. The transformer multi-parameter sensing device according to claim 5, characterized in that: The data transmission device transmits the collected data of the sensor in the following manner: Extracting time information from the collected data of each sensor respectively as the first time information of each sensor; Respectively acquiring time information when a data collection instruction is sent to each of the sensors as second time information of each sensor; respectively calculating the difference between the first time information and the second time information of each of the sensors as the action time of each of the sensors; Performing time correction on the collected data of each sensor by using terminal timing or clock timing according to the action time of each sensor; The corrected collected data of each sensor is sent to the server.

Images