CN102607643B - Overheat fault diagnosis and early warning method for electrical equipment of traction substation of electrified railway - Google Patents

Abstract

The invention discloses an overheating fault diagnosis and early warning method for electrical equipment in an electrified railway traction substation. The fault warning system is composed of wireless sensors, concentrators, computers and other parts. The wireless sensors include wireless temperature sensors, wireless temperature and humidity sensors, and wireless current sensors to measure the temperature of electrical equipment, ambient temperature and humidity, and transmission line current; the concentrator realizes the reception of data from each wireless sensor monitoring point in the entire traction substation. And upload the collected data to the computer; the computer completes the collection and storage of monitoring data, and completes the intelligent diagnosis and early warning of electrical equipment overheating faults based on the collected data. The system can provide a scientific basis for the operation and maintenance of electrified railway traction substations and ensure the safety of railway power supply.

Description

Diagnosis and early warning method for overheating faults of electrical equipment in traction substations of electrified railways

technical field

The invention relates to a fault diagnosis and early warning system and method, in particular to a fault diagnosis and early warning method for electrical equipment overheating in traction substations of electrified railways.

Background technique

There are a large number of electrical equipment such as high-voltage isolating switches, transformers, circuit breakers, transformers, and high-voltage switchgears in the traction substation. Under normal circumstances, the resistance of the contact point increases due to poor contact at the joint, aging or surface oxidation, etc., and then the temperature rise causes excessive heat and burns out. If the temperature of these heat-generating parts is not monitored in time and effectively , which may eventually lead to equipment damage, resulting in power outages or even fire accidents, which will bring great safety hazards to the operation of electrified railways. Therefore, necessary detection measures should be taken to understand the current operating status of each electrical equipment, and its change trend should be analyzed according to the real-time operating status data of the equipment and combined with historical data, so as to reduce the hidden danger of accidents in the power supply system, reduce the incidence of accidents, and shorten the time for fault finding and maintenance. The purpose of ensuring the safe and reliable operation of the power supply system.

At present, there are two main types of electrical equipment temperature monitoring technologies used at home and abroad: one is to use infrared technology (infrared thermometer or infrared imager) to detect temperature by manual inspection; the other is to arrange wired sensor networks, such as Optical fiber is used for temperature detection. However, none of these methods can meet the temperature monitoring requirements of traction substations. The main reason is that traction substations are different from general substations. When there is a train passing through the power supply line section, the relevant electrical equipment has current passing through, and the hidden danger of equipment overheating failure can be revealed. When there is no train passing, the load current of the electrical equipment is very small At this time, even if the temperature of the relevant equipment is monitored, hidden troubles cannot be found. Because infrared technology temperature measurement is manual detection, real-time online detection cannot be realized, and the temperature inside the device cannot be measured and there are large errors, so it is impossible to realize timely and accurate detection of overheating faults. Optical fiber temperature measurement technology is not conducive to large-scale due to complex wiring. Large-scale applications, and are easily contaminated, resulting in a significant drop in temperature measurement accuracy.

In recent years, temperature measurement technology based on wireless sensors has been widely used. However, it has the following limitations in the monitoring of electrical equipment in substations: the detection object is only the temperature of the monitored point, and it is determined whether an electrical equipment overheating fault occurs by simply setting a temperature threshold. According to common sense, we know that under different ambient temperature and humidity, the same electrical equipment temperature often indicates different overheating fault results, and different line currents will also lead to different operating temperature results, so whether there is an overheating fault is not only related to the The temperature of the equipment at the measuring point is related to the ambient temperature, ambient humidity and line current, which is the result of a comprehensive judgment.

Contents of the invention

The technical problem to be solved by the present invention is to provide a method for diagnosing and early warning of electrical equipment overheating faults in traction substations of electrified railways. The online detection of electrical equipment temperature, ambient temperature, ambient humidity and line current is realized through wireless sensors, and the method is established based on least squares Support vector machine (least square support vector machines referred to as LSSVM) fault diagnosis and early warning model, so as to realize accurate diagnosis and early warning of electrical equipment overheating faults, provide a basis for reliable operation of electrical equipment, and greatly reduce the incidence of failure.

To achieve the above object, the present invention adopts the following technical means:

The electric railway traction substation electrical equipment overheating fault diagnosis and early warning system is characterized in that it is mainly composed of wireless sensors, concentrators and computers, wherein the wireless sensors include wireless temperature sensors for measuring the temperature of electrical equipment, and for measuring ambient temperature and humidity. Wireless environmental temperature and humidity sensors, wireless current sensors used to measure line current; the concentrator is responsible for collecting wireless sensor data and uploading it to the computer, and the computer uses the least squares support vector machine overheating fault early warning model to realize the overheating fault detection of electrical equipment according to the uploaded data. Intelligent diagnosis and online early warning.

One of the wireless temperature sensors is installed on each electrical equipment that requires overheating fault monitoring; one of the wireless environmental temperature and humidity sensors is installed in each area with common environmental characteristics; the wireless current sensor is installed on each of the common environmental characteristics. Install one on each line with current characteristics;

The various wireless sensors are based on the CC2430 chip based on ZIGBEE technology as the core; the wireless temperature sensor uses DS18B20 to sense the temperature of electrical equipment; the wireless temperature and humidity sensor uses DHT21 to sense the temperature and humidity of the environment; the wireless current sensor uses an inductive The coil obtains the current signal;

A method for diagnosing and early warning of overheating faults of electrical equipment in traction substations of electrified railways, comprising the following steps:

(1) The wireless sensor measures the electrical equipment monitoring point temperature T _equipment , ambient temperature T _environment , ambient humidity H _environment and line current I _line , and then transmits these data to the concentrator;

(2) After the concentrator receives the data transmitted by the wireless sensor, it transmits the data to the computer via Ethernet, and then returns the reporting cycle of the above data as a response data packet to the corresponding sensor;

(3) After the computer receives the reported data sent by the concentrator, it first stores it, and then processes it according to the following rules according to the data type: if the data is ambient temperature, ambient humidity or line current, it will not be processed; if the data is electrical equipment Temperature, the position of the measuring point is determined according to the temperature data of the electrical equipment, and then the ambient temperature data, ambient humidity data, and line current data are called out from the computer to form measurement data {T _equipment , T _environment , H _environment , I _line };

(4) Input the measurement data obtained in step (3) {T _equipment , T _environment , H _environment , I _line } into the established least squares support vector machine overheating fault early warning model to obtain fault diagnosis and early warning results.

The least squares support vector machine overheating fault early warning model is established according to the following steps:

(1) Determine the type of kernel function: radial basis function As a kernel function, where x is the current input data, x _n is the training sample set, and δ is the width parameter of the radial basis function;

(2) Use the standard search algorithm to determine the hyperparameters {δ ² ,γ} of the model, where γ is the regularization parameter;

(3) Use the samples in the sample database as the training data set to train the model and obtain the model parameters {α _n ,b}, where α _n is the Lagrangian operator (also known as the support factor), and b is the partial set value;

(4) Get the fault diagnosis model:

Before fault diagnosis, firstly, normalize the acquired data;

The computer adjusts the reporting period of the temperature data of the electrical equipment according to the result of the fault diagnosis; the reporting of the line current adopts a real-time calling method, that is, when the computer receives the temperature data of the wireless electrical equipment, it calls the line current data;

After the least squares support vector machine overheating fault early warning model has collected a certain number of new fault samples, the model is updated and trained.

Compared with the prior art, the system and method for overheating fault diagnosis and early warning of electrical equipment in electrified railway traction substations have at least the following advantages: the invention uses wireless sensors to measure the temperature of electrical equipment, which effectively solves the problem that high-voltage equipment is difficult to detect online in real time The use of a variety of information comprehensive diagnosis and early warning can effectively eliminate the temperature changes of electrical equipment caused by changes in ambient temperature, humidity and line carrying capacity, thereby realizing accurate diagnosis of overheating faults and providing a reliable guarantee for power supply safety.

Description of drawings

Fig. 1 is a schematic structural diagram of the online diagnosis and early warning system for overheating faults of electrical equipment in traction substations of the present invention;

Fig. 2 is a hardware structural diagram of a wireless temperature sensor;

Fig. 3 is a wireless temperature sensor software work flow chart;

Figure 4 is a flowchart of fault diagnosis and early warning.

Detailed ways

An embodiment of the electrical equipment overheat fault diagnosis and early warning system and method of the electrified railway traction substation of the present invention will be described in detail below in conjunction with the accompanying drawings:

The present invention performs on-line diagnosis and early warning for overheating faults of electrical equipment in traction substations of electrified railways. Sensors, wireless current sensors for measuring line current, concentrators for collecting sensor data, and computers for fault diagnosis and unified monitoring.

In a traction substation, install a wireless temperature controller for each electrical equipment that needs to be monitored for overheating faults (such as equipment clamps at transformers, isolating switches, circuit breakers, current transformers, etc., busbars in switch cabinets, etc.) Sensors; install a wireless ambient temperature and humidity sensor in areas with common environmental characteristics, and the measured ambient temperature and humidity data results are shared by all electrical equipment monitoring points in the area for fault diagnosis. The ambient temperature and humidity sensor, its measurement results are used for overheating fault diagnosis of all outdoor electrical equipment. An environmental temperature and humidity sensor is installed in the high-voltage room, and its measurement results are used for all overheating fault diagnosis of electrical equipment in the high-voltage room; each line has a common current characteristic A wireless current sensor is installed on each line, and its data results are shared by all electrical equipment monitoring points on the line for fault diagnosis.

The wireless sensor completes the measurement of the electrical equipment monitoring point temperature T _equipment , ambient temperature T _environment , environmental humidity H _environment and line current I _line ; and transmits the measurement results to the concentrator by wireless; the concentrator can be installed according to the specific conditions of the traction substation. For example, one is installed outdoors, which is responsible for the collection of outdoor sensor data; one is installed in the high-pressure room, which is responsible for the collection of sensor data in the high-pressure room.

After the concentrator receives the wireless sensor data, it transmits the data to the computer through the industrial Ethernet, and returns the reporting period of the above sensor as a response data packet to the corresponding sensor.

After the computer receives the reported data, it first stores the data, and then processes it according to the following rules according to the data type: if the data is ambient temperature, ambient humidity or line current, it will not be processed; if the data is the temperature of electrical equipment, perform overheating Fault diagnosis and early warning procedures, specifically:

After the computer receives the electrical equipment temperature data and completes the data storage, it first executes the data matching program: according to the received electrical equipment temperature data, determine the location of the measuring point (that is, the ID of the electrical equipment temperature sensor), and then call out the Match the ambient temperature and humidity sensor ID and the line current sensor ID, obtain the line current by calling, and obtain the environmental temperature and humidity data by querying the latest data from the historical database to form measurement data {T _equipment , T _environment , H _environment , I _line } ;

Input the obtained measurement data set into the established LSSVM overheating fault early warning model, obtain fault diagnosis and early warning results, and display the results on the monitoring software interface.

The following is combined with a specific application example to illustrate:

As shown in Figure 2, the wireless sensor used to measure the temperature of electrical equipment in traction substations is implemented using the CC2430 chip based on ZIGBEE technology, which integrates a low-power microcontroller, power management module, analog-to-digital conversion module, radio frequency module and storage Modules, etc., can realize data collection, processing and wireless transmission with one chip. The temperature acquisition is realized by DS18B20, the temperature chip is a digital temperature sensor, and the temperature data can be obtained through the digital bus, and its error is within ±1.5°C. The wireless sensor designed by this technology has a communication distance of up to 500 meters, which fully meets the needs of electrical equipment temperature monitoring.

For the wireless sensor for environmental temperature and humidity and line current measurement, the only difference in its hardware structure is that the selected front-end sensor is different. Among them, DHT21 is used to measure the environmental temperature and humidity. The sensor is a temperature and humidity composite sensor with digital signal output. The line current adopts the self-made induction coil to obtain the current signal.

As shown in Figure 3, the working process of the wireless electrical equipment temperature sensor is as follows: firstly, it is judged whether the dormancy cycle is over, and when the dormancy cycle is reached, it enters into the temperature collection program, completes data conversion, filtering and other processing procedures, and then sends the collection results to The concentrator waits for the response packet; after receiving the response packet from the concentrator, it is determined whether it is a parameter change packet (that is, whether to change the reporting period), and if so, change the corresponding parameter according to the requirements. Wireless electrical equipment temperature sensor, its reporting period can be adaptively adjusted according to the degree of overheating fault. A typical set of values is: 30 minutes for no fault, 10 minutes for minor fault, and 5 for moderate fault minutes, and the reporting cycle is 1 minute for serious faults.

For environmental temperature and humidity sensors, the reporting period can be manually set, and the typical value is to report once every 30 minutes. For current sensors, it is obtained by summoning.

After the computer obtains the electrical equipment monitoring point temperature _Tequipment , ambient temperature _Tenvironment , ambient humidity _Henvironment , and line current _Iline , the fault diagnosis model based on LSSVM is used for overheating fault diagnosis and early warning. The workflow is shown in Figure 4.

First of all, it is determined whether to carry out fault model training or fault diagnosis. Model training is carried out in two cases. One is to use it when the model is first established; the other is to collect new fault samples reaching a certain number as the running time becomes longer New collection of 10 fault samples) to update the model training.

The fault diagnosis results are divided into four categories, namely no fault, minor fault, moderate fault, moderate fault and severe fault, which correspond to the output of the LSSVM fault model as 0, 1, 2, and 3, respectively. In the process of establishing the training sample set, T _equipment , T _environment , H _environment , and I _line are automatically obtained by the online monitoring system, and the corresponding fault types are specified by the operator based on experience and analysis results.

After entering the model training program, first determine the type of kernel function and select the radial basis function As the kernel function of LSSVM, where x is the current input data, x _n is the training sample set, and δ is the width parameter of the radial basis function; then use the standard search algorithm in the Matlab mathematical toolbox of LSSVM to determine the hyperparameters of the LSSVM model {δ ² ,γ}, where γ is the regularization parameter; then, the LSSVM model is trained with samples in the sample database as the training data set to obtain model parameters {α _n ,b}, where α _n is the Grangian operator (also known as support factor), b is the bias value, and the fault diagnosis model can be obtained: $Y\left(x\right)=\underset{\mathrm{no}=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\α}}_{\mathrm{no}}K(x,x_{\mathrm{no}})+b.$

In the normal monitoring process, the model enters the fault diagnosis process. First, the acquired data is normalized, and then the trained model is used Obtain the fault diagnosis result and display it on the monitoring interface, and determine whether to change the reporting cycle parameters of the wireless electrical equipment temperature sensor according to the severity of the fault, and store the relevant fault data as samples in the sample database. When the number of new samples reaches 10 When , enter the model update process to update the existing fault diagnosis model.

The online diagnosis method and early warning method for overheating faults of electrical equipment in traction substations provided by the present invention provide fault diagnosis results through the fusion analysis of various data such as the temperature of the monitoring point, the current size of the location, the ambient temperature and humidity of the location; The diagnosis results can meet the accurate diagnosis and early warning under various environments and external conditions of power transmission conditions, which satisfies the intelligent demand for traction substation monitoring and ensures the safety of railway power supply.

The above is only one embodiment of the present invention, not all or the only embodiment. Any equivalent transformation of the technical solution of the present invention adopted by those of ordinary skill in the art by reading the description of the present invention is the right of the present invention. covered by the requirements.

Claims (6)

1. the overheat fault diagnostic of an electric railway traction Substation Electric Equipment overheat fault diagnostic and early warning system and method for early warning, it is characterized in that: electric railway traction Substation Electric Equipment overheat fault diagnostic and early warning system are primarily of wireless senser, concentrator and computing machine composition, and wherein wireless senser comprises radio temperature sensor, the wireless environment Temperature Humidity Sensor for measurement environment humiture, the radio flow sensor for measuring circuit electric current for measuring electrical equipment temperature;

Described overheat fault diagnostic and method for early warning comprise the following steps:

(1) wireless senser measures electrical equipment monitoring point temperature T _equipment, environment temperature T _environment, ambient humidity H _environmentwith line current I _circuit, then these data are transferred to concentrator, described line current I _circuitreport and adopt real-time calling mode, namely after computing machine receives the temperature data of cordless electrical appliance, calling line current data;

(2) concentrator receive wireless senser transmission data after, these data are sent to computing machine with Ethernet, then by the report cycle of above-mentioned data responsively packet be back to corresponding sensor;

(3), after computing machine receives the reported data of concentrator transmission, first store, then process according to following rule according to data type: if data are environment temperature, ambient humidity or line current, then do not process; If data are electrical equipment temperature, then according to the temperature data determination point position of electrical equipment, then recall from computing machine and the ambient temperature data of this location matches, ambient humidity data, and line current data, composition measurement data { T _equipment, T _environment, H _environment, I _circuit;

(4) by measurement data { T that step (3) obtains _equipment, T _environment, H _environment, I _circuitinput the least square method supporting vector machine overheating fault Early-warning Model set up, obtain fault diagnosis and early warning result;

Described least square method supporting vector machine overheating fault Early-warning Model is set up according to following steps:

(1) definite kernel function kind: with radial basis function as kernel function, wherein, x is present input data, x _nfor training sample set, δ is the width parameter of radial basis function;

(2) hyper parameter { δ of standard search algorithms Confirming model is adopted ², γ }, wherein, γ is regularization parameter;

(3) using the sample in sample database as training dataset, to model training, model parameter { α is obtained _n, b}, wherein, α _nfor Lagrangian, b is bias;

(4) fault diagnosis model is obtained:

2. overheat fault diagnostic as claimed in claim 1 and method for early warning, is characterized in that: before carrying out fault diagnosis, first, is normalized the data obtained.

3. overheat fault diagnostic as claimed in claim 1 and method for early warning, is characterized in that: computing machine adjusts according to the report cycle of the result of fault diagnosis to electrical equipment temperature data.

4. overheat fault diagnostic as claimed in claim 1 and method for early warning, is characterized in that: least square method supporting vector machine overheating fault Early-warning Model, collecting after new fault sample reaches some, is carried out to model renewals and trained.

5. overheat fault diagnostic according to claim 1 and method for early warning, is characterized in that: described radio temperature sensor installs one on each electrical equipment needing to carry out overheating fault monitoring; The region that described wireless environment Temperature Humidity Sensor has common environmental feature respectively installs one; Respectively on described radio flow sensor has common current feature circuit at every bar install one.

6. overheat fault diagnostic according to claim 5 and method for early warning, is characterized in that: described various wireless senser all with the CC2430 chip based on ZIGBEE technology for core; Described radio temperature sensor adopts DS18B20 induction electric device temperature; Described wireless humiture sensor adopts DHT21 induced environment humiture; Described radio flow sensor adopts inductive circular loop to obtain current signal.