Self-adaptive cooperative power transmission line fault monitoring system

Abstract

The invention discloses a self-adaptive cooperative power transmission line fault monitoring system, and relates to the technical field of power grid fault monitoring. Compared with a previous power transmission line fault monitoring system, the problem that information provided by a single sensor is limited is solved; a traditional monitoring system lacks self-adaptive adjustment capability; and the collaboration between different monitoring devices and functional modules is poor. Data collected by various sensors are integrated through a sensor network, and fusion analysis is performed by using an intelligent diagnosis algorithm. The diagnosis accuracy can be greatly improved, the system has adaptive adjustment capability, and the monitoring strategy can be optimized in real time according to the fault development situation; the sampling frequency of the sensor can be adaptively adjusted according to the real-time operation state and the fault risk of the line; according to factors such as line aging degree, geographical environment and historical fault distribution, monitoring resources are dynamically allocated, and monitoring key points are intelligently transferred; all the modules of the system cooperatively operate and are matched with one another; misjudgment and missed judgment can be reduced, and the overall stability of the system is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid fault monitoring, and in particular to an adaptive coordinated transmission line fault monitoring system. Background Art

[0002] With the rapid development of the economy, the demand for electricity continues to rise. As the key infrastructure for power transmission, the stable operation of transmission lines is crucial to ensuring the normal order of social production and life. Traditional transmission line fault monitoring methods have gradually exposed many limitations in the face of increasingly complex operating environments and growing reliability requirements, which provides an opportunity for the birth of adaptive collaborative transmission line fault monitoring systems.

[0003] In the early days, fault monitoring of transmission lines mainly relied on manual inspections. With the advancement of technology, sensor-based monitoring technology began to be applied. A single type of sensor was installed on the transmission line, such as using only a current sensor to monitor the line current changes to determine the fault. However, the types of transmission line faults are complex and diverse, including short circuit faults, open circuit faults, ground faults, insulator flashovers, etc. The information provided by a single sensor is limited, and it can often only reflect abnormalities in one aspect, and it is impossible to fully and accurately diagnose the fault. For example, when a fault occurs in the line due to local overheating, it may not be detected by the current sensor alone, because the current change is not necessarily significant, and the coordinated monitoring of the temperature sensor is required to accurately locate the problem.

[0004] Traditional monitoring systems lack adaptive adjustment capabilities. Both the sampling frequency of sensors and the key monitoring areas are usually fixed. It is obviously unreasonable to use the same monitoring strategy under normal operating conditions and special working conditions such as bad weather and high load. For example, in stormy weather, the line is affected by factors such as wind and rain, and the probability of failure increases significantly. At this time, if the conventional low sampling frequency is still maintained, some key fault precursor information may be missed; similarly, during the period of long-term stable operation of the transmission line, maintaining a high sampling frequency will cause energy waste and data redundancy, and increase the system operation and maintenance costs.

[0005] There is poor coordination between different monitoring devices and functional modules. After the data collected by the sensor is transmitted to the background, the data analysis, fault diagnosis and other links are independent of each other, and fail to fully share information and work together. This can easily lead to inaccurate diagnostic results, and misjudgments and missed judgments often occur. For a simple example, when the current sensor detects abnormal current fluctuations, if there is no real-time collaborative analysis with the voltage sensor and temperature sensor, and the fault is judged based on the single current information, it is likely to draw an incorrect diagnostic conclusion, delaying the repair time.

[0006] In order to solve the above problems, the present invention proposes an adaptive coordinated transmission line fault monitoring system. Summary of the invention

[0007] The purpose of the present invention is to propose an adaptive coordinated transmission line fault monitoring system to solve the problems raised in the background technology:

[0008] The information provided by a single sensor is limited; the traditional monitoring system lacks adaptive adjustment capabilities; and the coordination between different monitoring devices and functional modules is poor.

[0009] In order to achieve the above object, the present invention adopts the following technical solutions:

[0010] An adaptive coordinated transmission line fault monitoring system, comprising:

[0011] Network building module: used to build a sensor network for power transmission lines based on several types of sensors;

[0012] Collection and transmission module: used to collect and transmit data based on the sensor network;

[0013] The acquisition and transmission module includes:

[0014] Data acquisition unit: used to collect the raw data monitored by each sensor and convert the analog signal into digital monitoring data through the edge device;

[0015] Data transmission unit: used to transmit the collected digital monitoring data to the background control center through a combination of several communication methods;

[0016] Analysis and processing module: used to receive data transmitted by the data communication module, and perform pre-processing and fault diagnosis on the data;

[0017] The analysis and processing modules include:

[0018] Data preprocessing unit: used for denoising and normalizing the received digital monitoring data;

[0019] Feature extraction unit: used to extract the transmission line operation status and environmental conditions from the pre-processed data;

[0020] Fault diagnosis unit: used to perform fault diagnosis on the transmission line based on the extracted operation status of the transmission line and the intelligent diagnosis algorithm, and obtain the fault diagnosis result and fault severity;

[0021] Adaptive collaborative control module: It is used to dynamically control the collaborative work between sensors and different functional modules based on the adaptive adjustment mechanism according to the real-time operating status, environmental conditions, fault diagnosis results and fault severity of the transmission line.

[0022] Preferably, the sensor network is composed of a plurality of current sensors, voltage sensors, temperature sensors, humidity sensors, vibration sensors, wind speed sensors, wind direction sensors, tower tilt sensors and insulator leakage current sensors.

[0023] Preferably, the operating status of the transmission line extracted by the feature extraction unit includes the amplitude, peak value and period corresponding to the current signal collected by the current sensor; the amplitude, peak value and period corresponding to the voltage signal collected by the voltage sensor; the temperature change trend and temperature change gradient of the transmission line temperature data collected by the temperature sensor; the acceleration, velocity, displacement and frequency of the transmission line vibration signal collected by the vibration sensor; the lateral inclination angle and longitudinal inclination angle of the tower collected by the tower tilt sensor, and the insulator leakage current collected by the insulator leakage current sensor; the environmental conditions include the ambient temperature collected by the temperature sensor, the ambient humidity collected by the humidity sensor, the wind speed collected by the wind speed sensor, and the wind direction angle collected by the wind direction sensor.

[0024] Preferably, the intelligent diagnosis algorithm is as follows:

[0025] For each operating state, a fuzzy set F is defined, and a membership function μ is defined for each fuzzy set F. F (x), the membership function μ F (x) are as follows:

[0026]

[0027] Among them, x is the operating state in the fuzzy set F, x0 and x1 are the membership function parameters, when μ F When (x)≥A, it is determined that a fault occurs, where A is the preset fault threshold;

[0028] Substitute the running state into the membership function, calculate the membership degree of each running state to each fuzzy set, and obtain the membership matrix

[0029] Construct a fuzzy rule base consisting of premise conditions and fault conclusions based on historical fault data and expert experience;

[0030] Based on the membership matrix Determine whether the prerequisite of each rule in the fuzzy rule base is met, that is, whether the membership of each fuzzy set reaches the preset membership threshold; if so, trigger the fault conclusion corresponding to the fuzzy rule;

[0031] Determine each rule R that is triggered k The corresponding fault subset F k , and according to the rule credibility c k The basic probability distribution function m is calculated based on the membership degree of the running state k; The details are as follows:

[0032]

[0033] For the fault subset F k The other fault subsets in the fault set Θ are assigned corresponding probability values, and the sum of the probability values ​​of all fault subsets in the fault set Θ is 1;

[0034] Fusion using Dempster's synthesis rule:

[0035] For two evidence bodies m consisting of basic probability distribution functions a and m b The fused evidence body m is obtained by fusion ab , for the fused evidence body m ab The corresponding basic probability distribution of the fault subset D is calculated as follows:

[0036]

[0037] in, is the conflict coefficient, indicating the degree of conflict between evidences; D is the fused evidence m ab The corresponding fault subset; B is the evidence body m a The corresponding fault subset; C is the evidence body m b The corresponding fault subset;

[0038] After several fusions, the fused comprehensive evidence body m is obtained. final ;

[0039] Based on the comprehensive evidence m obtained after fusion final The trust function Bel(G) is calculated as follows:

[0040]

[0041] Among them, m final (H) is the weight of evidence after fusion, the basic probability distribution for the fault subset H;

[0042] According to the maximum trust criterion, the fault subset with the largest trust function value is selected as the diagnosis result:

[0043]

[0044] Among them, F diagmosis is the fault diagnosis result;

[0045] After determining the fault diagnosis result, the fault diagnosis result is converted into a fault diagnosis description through a defuzzification operation, and the fault diagnosis description is input into the adaptive collaborative control module.

[0046] Preferably, in the fault diagnosis unit, the fault severity is further divided according to the fault credibility through a trapezoidal membership function.

[0047] Preferably, the adaptive adjustment mechanism is specifically as follows:

[0048] Define an intelligent agent, and define a fault diagnosis unit including an intelligent diagnosis algorithm as an intelligent agent;

[0049] A state space S is defined, wherein a state vector s in the state space S includes an operating state of the transmission line, environmental conditions, a fault diagnosis result, and a fault severity;

[0050] Define the action space A, which contains the action vector a = {a 11 ,Λ,a 1u ,Λ,a 1U , a 21 ,Λ,a 2v ,Λ,a 2V , a 31 ,Λ,a 3k ,Λ,a 3L}; a 1u Corresponding sampling frequency proportional coefficient adjustment action; a 2v Corresponding sensor monitoring range adjustment action; a 3k Adjustment actions of computing resources and communication resources corresponding to the fault type;

[0051] The reward function R(s,a) is defined as follows:

[0052] R(s,a)=w1R accuracy (s,a)+w2R energy (s,a)+w3R response (s,a)

[0053] Among them, w1, w2 and w3 are weight coefficients; R accuracy (s,a) is the fault diagnosis accuracy reward; R energy (s,a) is the energy consumption reward; R response (s,a) is the fault response timeliness reward;

[0054]

[0055] Among them, k1 is the accuracy importance coefficient; p acc To take action a to make the next fault diagnosis accuracy; p acc0 is the fault accuracy rate before taking action a;

[0056]

[0057] Among them, k2 is the energy consumption importance coefficient; E0 is the energy consumption before taking action a; E is the energy consumption after taking action a;

[0058]

[0059] Among them, k3 is the timeliness importance coefficient; t d0 is the time of first fault detection before taking action a; t d is the first fault detection time after taking action a;

[0060] Use the Q learning algorithm to learn the optimal strategy, initialize the Q value function as a table, the rows of the table correspond to different environmental states, and the columns correspond to different actions;

[0061] Use the ε-greedy strategy to select actions:

[0062]

[0063] Among them, ε is the exploration rate; is the environmental state s t The maximum Q value corresponding to all possible actions;

[0064] The agent performs the selected action a in the system t The system state transfers to the next state s according to the physical laws of line operation, environmental changes and action influences t+1 , and at the same time calculate and feed back the reward value R to the agent based on the reward function R(s,a) t ;

[0065] Based on the Q learning algorithm, the formula is updated with the newly acquired experience (s t ,a t ,R t ,s t+1 )Update the Q value:

[0066]

[0067] Where: Q'(s t ,a t ) is the updated Q value; Q(s t ,a t ) is the environmental state s t Take action a t Q value; α is the learning rate; γ is the discount factor; a' is the next environment state s t+1 The best action among all possible actions; is the next environment state s t+1The maximum Q value corresponding to all possible actions is the maximum expected cumulative reward obtained from all optional actions after transferring from the current environment state to the next environment state;

[0068] The Q-value function is repeatedly updated until the Q-value converges to obtain the corresponding optimal Q-value function. When the fault diagnosis unit detects a fault in the transmission line, the agent uses the optimal Q-value function to select and execute the optimal action from the action space based on the state vector.

[0069] Compared with the prior art, the present invention provides an adaptive coordinated transmission line fault monitoring system, which has the following beneficial effects:

[0070] The present invention integrates data collected by various sensors such as current, voltage, temperature, vibration, humidity, etc. through a sensor network, and uses intelligent diagnostic algorithms for fusion analysis. It can greatly improve the accuracy of diagnosis. The system has adaptive adjustment capabilities and can optimize monitoring strategies in real time according to the development trend of faults; it can adaptively adjust the sensor sampling frequency according to the real-time operation status and fault risk of the line; it can dynamically allocate monitoring resources and intelligently shift the focus of monitoring according to factors such as the degree of line aging, geographical environment, and historical fault distribution; the modules of the system work together and adapt to each other; it can reduce misjudgments and missed judgments and improve the overall stability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 This is a system block diagram mentioned in Embodiment 1 of the present invention;

[0072] Figure 2 This is a diagram of the architecture of the adaptive adjustment mechanism mentioned in Example 1 of the present invention.

[0073] Meaning of the symbols in the figure:

[0074] 100. Network building module; 200. Collection and transmission module; 210. Data collection unit; 220. Data transmission unit; 300. Analysis and processing module; 310. Data preprocessing unit; 320. Feature extraction unit; 330. Fault diagnosis unit; 400. Adaptive collaborative control module. DETAILED DESCRIPTION

[0075] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0076] The present invention integrates data collected by various sensors such as current, voltage, temperature, vibration, humidity, etc. through a sensor network, and uses an intelligent diagnostic algorithm for fusion analysis. It can greatly improve the accuracy of diagnosis. The system has the ability to adaptively adjust and can optimize the monitoring strategy in real time according to the development trend of the fault; it can adaptively adjust the sensor sampling frequency according to the real-time operation status and fault risk of the line; it can dynamically allocate monitoring resources and intelligently shift the focus of monitoring according to factors such as the degree of line aging, geographical environment, and historical fault distribution; the various modules of the system work together and adapt to each other; it can reduce misjudgments and missed judgments and improve the overall stability of the system. Specifically include the following contents.

[0077] Embodiment 1:

[0078] See also Figure 1 The present invention provides an adaptive coordinated transmission line fault monitoring system, comprising:

[0079] Network building module 100: used to build a sensor network for power transmission lines based on several types of sensors;

[0080] The sensor network consists of various types of sensors, which are widely distributed in key parts of the transmission line, such as towers, conductors, insulators, etc.

[0081] Current sensor: monitors the current size and changes in the transmission line in real time, and converts the current signal into an electrical signal through a high-precision sensing device for subsequent collection and analysis. Once the current fluctuates abnormally, such as a sudden increase (which may indicate a short circuit fault) or a sharp decrease (which may be a line break, etc.), the corresponding signal can be captured in time, providing a key basis for fault judgment.

[0082] Voltage sensor: Responsible for measuring the voltage value of different nodes of the transmission line. Its working principle is based on electromagnetic induction or voltage division principle, etc., which converts the line voltage into a weak current signal that can be processed. Abnormal voltage changes, such as voltage sag and overvoltage, are often closely related to line faults (such as voltage shock caused by lightning strikes, voltage abnormalities caused by line insulation damage, etc.). Voltage sensors can accurately sense these changes and feed back to the system.

[0083] Temperature sensor: installed in the parts that are prone to heat, such as conductors, connection points, and towers, to detect the real-time temperature and ambient temperature of these locations. Because the transmission line will have a local temperature rise in the event of overload operation, poor contact, and other faults, the temperature sensor can keenly detect the abnormal temperature rise and assist in determining whether there is a hidden danger of overheating of the line, such as increased resistance and heating caused by loose conductor joints.

[0084] Vibration sensor: monitors the vibration of transmission lines caused by external forces such as wind vibration and dancing, including acceleration, speed, displacement and frequency. Under normal circumstances, the vibration of the line is within a certain range. When a fault occurs (such as broken wires, loose tower foundations, etc.), the vibration characteristics will change. Vibration sensors can help determine whether there is an abnormality in the line structure by collecting vibration signals.

[0085] Humidity sensor: monitors ambient humidity. Humidity is an important factor affecting the external insulation performance of transmission lines. A high humidity environment may reduce the insulation performance of insulators and increase the risk of insulator flashover. Monitoring ambient humidity through humidity sensors can provide an important basis for evaluating the external insulation status of transmission lines. When the humidity exceeds a certain threshold, the system can strengthen the monitoring of related parameters such as insulator leakage current, and provide early warning of faults caused by humidity. At the same time, humidity changes may also be related to the corrosion of the line. For some transmission line structures with more metal parts, the data from humidity sensors can help analyze the corrosion risk of the line.

[0086] Wind speed sensor: used to monitor the ambient wind speed. The impact of wind speed on transmission lines is mainly reflected in wind load. Strong winds may cause problems such as the dancing of transmission line conductors and tilting of towers. Real-time monitoring of wind speed can provide data support for determining whether the line will have wind deviation failures. When the wind speed exceeds the wind speed range allowed by the line design, the system can adjust the monitoring focus in time, such as increasing the monitoring frequency of the conductor vibration amplitude and tower stability, and can further analyze the direction of wind load that the line may be subjected to in combination with wind direction information, and take measures in advance to prevent line failures, such as adjusting line tension.

[0087] Wind direction sensor: used to monitor the environmental wind direction. The combination of wind direction and wind speed information can more accurately assess the impact of wind on the transmission line. It is particularly important for the prediction and analysis of wind deviation faults. Different wind directions will cause the wires to be subjected to wind loads in different directions. When the wind direction is perpendicular to the line direction, the risk of wind deviation faults may increase. Through the information provided by the wind direction sensor, the system can accurately calculate the direction of the wind load, and combine the structural parameters of the transmission line to predict the swing direction and amplitude of the wire, and discover potential wind deviation fault risks in advance, so as to take targeted protective measures, such as installing anti-wind deviation devices.

[0088] Tower tilt sensor: used to monitor the tilt angle of the tower. The tower is an important supporting structure for the transmission line. The tilt of the tower may be caused by foundation settlement, soil loosening, external force damage (such as vehicle collision, landslide, etc.). The tower tilt sensor can monitor the tilt state of the tower in real time. Once the tilt angle exceeds the set threshold, it can issue an alarm in time to prompt the operation and maintenance personnel to inspect and repair the tower. At the same time, the tower tilt data can also be combined with other sensor data (such as wind speed, vibration, etc.) to comprehensively analyze the reasons for the tower tilt and determine whether there are other potential fault risks, such as unbalanced line tension.

[0089] Insulator leakage current sensor: used to measure the leakage current. The insulator leakage current sensor is usually placed around the grounding conductor of the insulator. When leakage current passes through the grounding conductor, according to the law of electromagnetic induction, an induced current will be generated in the secondary winding of the sensor. By measuring this induced current, the leakage current of the insulator can be obtained. Insulator leakage current is a key parameter reflecting the insulation performance of the insulator. Under normal circumstances, the leakage current of the insulator is very small. When the surface of the insulator is polluted, damp or aged, the leakage current will increase. By monitoring the insulator leakage current, the degradation of the insulation performance of the insulator can be discovered in time, which is an important means of predicting insulator flashover failures. When the leakage current exceeds a certain safety threshold, the system can determine that the insulator has a hidden fault and promptly notify the operation and maintenance personnel to take maintenance measures such as cleaning and replacement to avoid power outages of transmission lines caused by insulator flashovers.

[0090] The data collected by the sensor network can be found in Table 1:

[0091] Table 1 Sensor network collection data

[0092]

[0093]

[0094] Collection and transmission module 200: used to collect and transmit data based on the sensor network;

[0095] The acquisition and transmission module 200 includes:

[0096] Data acquisition unit 210: used to collect the raw data monitored by each sensor, with the functions of sampling, quantizing and encoding analog signals, converting the continuously changing analog signals output by the sensor into digital monitoring data for subsequent storage, processing and transmission. At the same time, this unit regularly acquires data according to the set sampling frequency, which can ensure the accuracy and real-time nature of the collected data.

[0097] Data transmission unit 220: used to transmit the collected digital monitoring data to the background control center through a combination of several communication methods; the collected digital monitoring data is transmitted to the background monitoring center, and its transmission method can be selected according to the actual application scenario, such as wired transmission (such as optical fiber communication, which has the advantages of high transmission rate and strong anti-interference ability, suitable for long-distance and large-capacity data transmission requirements) or wireless transmission (such as ZigBee, GPRS, 4G / 5G and other wireless communication technologies, suitable for power transmission line areas with difficult wiring and complex geographical environment, convenient and flexible networking). Through the above transmission methods, the stability and integrity of data transmission can be ensured, data loss or errors can be avoided, and the monitoring center can be guaranteed to receive accurate on-site monitoring data in a timely manner.

[0098] Analysis and processing module 300: used to receive data transmitted by the data communication module, and perform pre-processing and fault diagnosis on the data;

[0099] The analysis and processing module 300 includes:

[0100] Data preprocessing unit 310: used to perform preprocessing operations such as denoising and normalization on the received raw monitoring data. Since the sensor may be affected by external electromagnetic interference, environmental noise and other factors during the data collection process, there will be certain noise components in the data. These interference information can be removed through denoising algorithms such as filtering and smoothing to improve data quality. The normalization operation is to unify data of different magnitudes and ranges into a specific standard range, which is convenient for subsequent unified analysis and comparison. For example, the voltage, current, temperature and other data collected by various sensors are mapped to the [0,1] interval.

[0101] Feature extraction unit 320: used to extract key information that can reflect the operation status and fault characteristics of the transmission line from the preprocessed data. This includes using time domain analysis methods to extract the amplitude, peak value, period and other characteristics of current and voltage signals, using frequency domain analysis (such as fast Fourier transform, etc.) to obtain vibration frequency, spectrum amplitude and other characteristics from vibration signals, and analyzing the change trend and gradient of temperature data. These extracted features will serve as an important basis for subsequent fault diagnosis.

[0102] Fault diagnosis unit 330: used to perform fault diagnosis on the transmission line based on the extracted feature information using an intelligent diagnosis algorithm to obtain a fault diagnosis result and fault severity.

[0103] The intelligent diagnosis algorithm is as follows:

[0104] For each operating state, a fuzzy set F is defined, and a membership function μ is defined for each fuzzy set F. F (x), the membership function μ F (x) are as follows:

[0105]

[0106] Among them, x is the operating state in the fuzzy set F, x0 and x1 are the membership function parameters, when μ F When (x)≥A, it is determined that a fault occurs, where A is the preset fault threshold.

[0107] Substitute the running state into the membership function, calculate the membership degree of each running state to each fuzzy set, and obtain the membership matrix Defining fuzzy sets and determining membership functions for each operating state can be the basis for converting accurate sensor data into fuzzy information. By setting membership function parameters reasonably, the membership degree of different operating states in each fuzzy set can be accurately reflected. For example, when judging whether the current is abnormal, the current value can be fuzzified into "normal", "high", "too high" and other states according to the set membership function, providing more realistic information for subsequent fault judgment.

[0108] Based on historical fault data and expert experience, a fuzzy rule base consisting of premise conditions and fault conclusions is constructed. The rule form is: IF (preconditions consisting of fuzzy language variables) THEN (fault conclusion); Based on historical fault data and expert experience, a fuzzy rule base is constructed, covering a variety of possible fault scenarios, such as "IF (current is high and temperature rises) THEN (short circuit fault possibility is high)" and other rules. According to the calculated membership matrix, it is judged whether the premise conditions of the rules are met. When they are met, the corresponding fault conclusion is triggered. This reasoning logic based on knowledge and experience helps to quickly locate possible fault types and improve diagnostic efficiency.

[0109] Based on the membership matrix Determine whether the prerequisite of each rule in the fuzzy rule base is met, that is, whether the membership of each fuzzy set reaches the preset membership threshold; if so, trigger the fault conclusion corresponding to the fuzzy rule;

[0110] Determine each rule R that is triggered k The corresponding fault subset F k , and according to the rule credibility c k The basic probability distribution function m is calculated based on the membership degree of the running state k ; The details are as follows:

[0111]

[0112] For the fault subset F kThe other fault subsets in the fault set Θ are assigned corresponding probability values, and the sum of the probability values ​​of all fault subsets in the fault set Θ is 1; the fault types set in the fault set Θ generally include short circuit faults, ground faults, lightning faults, wind deviation faults and icing faults.

[0113] Fusion using Dempster's synthesis rule:

[0114] For two evidence bodies m consisting of basic probability distribution functions a and m b The fused evidence body m is obtained by fusion ab , for the fused evidence body m ab The corresponding basic probability distribution of the fault subset D is calculated as follows:

[0115]

[0116] in, is the conflict coefficient, indicating the degree of conflict between evidences; D is the fused evidence m ab The corresponding fault subset; B is the evidence body m a The corresponding fault subset; C is the evidence body m b The corresponding fault subset;

[0117] After several fusions, the fused comprehensive evidence body m is obtained. final ;

[0118] Based on the comprehensive evidence m obtained after fusion final The trust function Bel(G) is calculated as follows:

[0119]

[0120] Among them, m final (H) is the weight of evidence after fusion, which is the basic probability distribution for the fault subset H. The basic probability distribution function is calculated according to the credibility of the trigger rule and the membership of the operating state, and the probabilities are allocated to different fault subsets. Then, the Dempster synthesis rule is used to fuse multiple evidence bodies, which fully considers the contribution of different information sources (i.e., different fuzzy rules) to fault diagnosis, and can continuously integrate to reduce information uncertainty and enhance the diagnostic credibility of this fault type.

[0121] According to the maximum trust criterion, the fault subset with the largest trust function value is selected as the diagnosis result:

[0122]

[0123] Among them, F diagmosis This is the fault diagnosis result.

[0124] After determining the fault diagnosis result, the fault diagnosis result is converted into a fault diagnosis description through a defuzzification operation, and then input into the adaptive collaborative control module 400, thereby completing the entire intelligent diagnosis process. The trust function is calculated and the diagnosis result is determined according to the maximum trust criterion. Finally, the result is converted into a clear fault diagnosis description through a defuzzification operation, which can make the diagnosis result have practical application value. The fault severity is further divided according to the fault credibility through a trapezoidal membership function, which can further provide data support for the later adaptive adjustment mechanism.

[0125] In the fault monitoring of power transmission lines, the data obtained by sensors often contain uncertainty. Fuzzy algorithms can fuzzify these imprecise data, while DS evidence theory can fuse and infer these fuzzified information, effectively deal with the fuzziness and uncertainty of fault characteristics, and more accurately determine whether a fault has occurred and the type of fault. The diagnostic algorithm that combines DS evidence theory with fuzzy algorithms can conduct a detailed analysis of faults by constructing a reasonable fuzzy rule base and evidence fusion mechanism. By fusing the evidence bodies corresponding to each fuzzy rule and calculating the trust function, according to the maximum trust criterion, it is possible to more accurately determine whether the fault type is a short circuit fault or other types of faults, and to locate the approximate area where the fault occurs, such as the line section between several towers.

[0126] Adaptive collaborative control module 400: used to dynamically control the collaborative work between sensors and different functional modules based on the adaptive adjustment mechanism according to the real-time operating status, environmental conditions, fault diagnosis results and fault severity of the transmission line. Figure 2 , the adaptive adjustment mechanism is as follows:

[0127] Define an intelligent agent, and define the adaptive collaborative control module 400 as an intelligent agent;

[0128] A state space S is defined, wherein a state vector s in the state space S includes an operating state of the transmission line, environmental conditions, a fault diagnosis result, and a fault severity;

[0129] Define the action space A, which contains the action vector a = {a 11 ,Λ,a 1u ,Λ,a 1U , a 21 ,Λ,a 2v ,Λ,a 2V , a 31 ,Λ,a 3k ,Λ,a 3L}; a 1u The corresponding sampling frequency proportional coefficient is adjusted to the action of the u-th sampling frequency proportional coefficient; a 2vThe corresponding sensor monitoring range is adjusted to the action of the vth range; a 3k The adjustment of computing resources and communication resources corresponding to the fault type is the action of the kth adjustment mode;

[0130] The reward function R(s,a) is defined as follows:

[0131] R(s,a)=w1R accuracy (s,a)+w2R energy (s,a)+w3R response (s,a)

[0132] Among them, w1, w2 and w3 are weight coefficients; R accuracy (s,a) is the fault diagnosis accuracy reward; R energy (s,a) is the energy consumption reward; R response (s,a) is the fault response timeliness reward;

[0133]

[0134] Among them, k1 is the accuracy importance coefficient; p acc To take action a to make the next fault diagnosis accuracy; p acc0 is the fault accuracy rate before taking action a;

[0135]

[0136] Among them, k2 is the energy consumption importance coefficient; E0 is the energy consumption before taking action a; E is the energy consumption after taking action a;

[0137]

[0138] Among them, k3 is the timeliness importance coefficient; t d0 is the time of first fault detection before taking action a; t d is the first fault detection time after taking action a;

[0139] Use the Q learning algorithm to learn the optimal strategy, initialize the Q value function as a table, the rows of the table correspond to different environmental states, and the columns correspond to different actions;

[0140] Use the ε-greedy strategy to select actions:

[0141]

[0142] Among them, ε is the exploration rate; is the environmental state s t The maximum Q value corresponding to all possible actions;

[0143] The agent performs the selected action a in the systemt The system state transfers to the next state s according to the physical laws of line operation, environmental changes and action influences t+1 , and at the same time calculate and feed back the reward value R to the agent based on the reward function R(s,a) t .

[0144] Based on the Q learning algorithm, the formula is updated with the newly acquired experience (s t ,a t ,R t ,s t+1 )Update the Q value:

[0145]

[0146] Where: Q'(s t ,a t ) is the updated Q value; Q(s t ,a t ) is the environmental state s t Take action a t Q value; α is the learning rate; γ is the discount factor; a' is the next environment state s t+1 The best action among all possible actions; is the next environment state s t+1 The maximum Q value corresponding to all possible actions is the maximum expected cumulative reward obtained from all optional actions after transferring from the current environment state to the next environment state.

[0147] The Q value function is repeatedly updated until the Q value converges to obtain the corresponding optimal Q value function. When the fault diagnosis unit 330 detects that there is a fault in the transmission line, the intelligent agent uses the optimal Q value function to select and execute the optimal action from the action space according to the state vector.

[0148] The above system is applied to the actual detection of power transmission lines. The specific adaptive adjustment mechanism can be referred to Table 2:

[0149] Table 2 Transmission line adaptive adjustment mechanism

[0150]

[0151]

[0152] The adaptive adjustment mechanism can timely increase the sampling frequency, thereby capturing line status changes more precisely and ensuring that no possible fault precursors are missed; it can reasonably allocate monitoring resources to improve the pertinence and effectiveness of fault monitoring; it can also provide more accurate data support for diagnostic algorithms based on initial signs of faults and current system status. It can guide the monitoring system to focus on key parameters and areas related to the fault and reduce interference from irrelevant information.

[0153] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. An adaptive coordinated transmission line fault monitoring system, characterized in that: include: A network building module (100): used to build a sensor network for a power transmission line based on several types of sensors; A collection and transmission module (200): used for collecting and transmitting data based on a sensor network; The acquisition and transmission module (200) comprises: Data collection unit (210): used to collect raw data monitored by each sensor and convert analog signals into digital monitoring data through edge devices; Data transmission unit (220): used to transmit the collected digital monitoring data to the background control center by combining several communication modes; Analysis and processing module (300): used for receiving data transmitted by the data communication module, and performing pre-processing and fault diagnosis on the data; The analysis and processing module (300) comprises: A data preprocessing unit (310): used for performing denoising and normalization preprocessing on the received digital monitoring data; A feature extraction unit (320): used to extract the transmission line operation status and environmental conditions from the pre-processed data; A fault diagnosis unit (330): used to perform fault diagnosis on the power transmission line based on the extracted power transmission line operation status and an intelligent diagnosis algorithm, and obtain a fault diagnosis result and a fault severity; Adaptive collaborative control module (400): used to dynamically control the collaborative work between sensors and different functional modules based on an adaptive adjustment mechanism and according to the real-time operating status, environmental conditions, fault diagnosis results and fault severity of the transmission line.

2. The adaptive coordinated transmission line fault monitoring system according to claim 1, characterized in that: The sensor network is composed of a plurality of current sensors, voltage sensors, temperature sensors, humidity sensors, vibration sensors, wind speed sensors, wind direction sensors, pole tower tilt sensors and insulator leakage current sensors.

3. The adaptive coordinated transmission line fault monitoring system according to claim 1, characterized in that: The operating status of the transmission line extracted by the feature extraction unit (320) includes the amplitude, peak value and period corresponding to the current signal collected by the current sensor; the amplitude, peak value and period corresponding to the voltage signal collected by the voltage sensor; the temperature change trend and temperature change gradient of the temperature data of the transmission line collected by the temperature sensor; the acceleration, speed, displacement and frequency of the vibration signal of the transmission line collected by the vibration sensor; the lateral tilt angle and longitudinal tilt angle of the tower collected by the tower tilt sensor, and the insulator leakage current collected by the insulator leakage current sensor; the environmental conditions include the ambient temperature collected by the temperature sensor, the ambient humidity collected by the humidity sensor, the wind speed collected by the wind speed sensor, and the wind direction angle collected by the wind direction sensor.

4. The adaptive coordinated transmission line fault monitoring system according to claim 3, characterized in that: The intelligent diagnosis algorithm is as follows: For each operating state, a fuzzy set F is defined, and a membership function μ is defined for each fuzzy set F. F (x), the membership function μ F (x) are as follows: Among them, x is the operating state in the fuzzy set F, x0 and x1 are the membership function parameters, when μ F When (x)≥A, it is determined that a fault occurs, where A is the preset fault threshold; Substitute the running state into the membership function, calculate the membership degree of each running state to each fuzzy set, and obtain the membership matrix Construct a fuzzy rule base consisting of premise conditions and fault conclusions based on historical fault data and expert experience; Based on the membership matrix Determine whether the prerequisite of each rule in the fuzzy rule base is met, that is, whether the membership of each fuzzy set reaches the preset membership threshold; if so, trigger the fault conclusion corresponding to the fuzzy rule; Determine each rule R that is triggered k The corresponding fault subset F k , and according to the rule credibility c k The basic probability distribution function m is calculated based on the membership degree of the running state k ; The details are as follows: For the fault subset F k The other fault subsets in the fault set Θ are assigned corresponding probability values, and the sum of the probability values ​​of all fault subsets in the fault set Θ is 1; Fusion using Dempster's synthesis rule: For two evidence bodies m consisting of basic probability distribution functions a and m b The fused evidence body m is obtained ab , for the fused evidence body m ab The corresponding basic probability distribution of the fault subset D is calculated as follows: in, is the conflict coefficient, indicating the degree of conflict between evidences; D is the fused evidence m ab The corresponding fault subset; B is the evidence body m a The corresponding fault subset; C is the evidence body m b The corresponding fault subset; After several fusions, the fused comprehensive evidence body m is obtained. final ; Based on the comprehensive evidence m obtained after fusion final The trust function Bel(G) is calculated as follows: Among them, m final (H) is the weight of evidence after fusion, the basic probability distribution for the fault subset H; According to the maximum trust criterion, the fault subset with the largest trust function value is selected as the diagnosis result: Among them, F diagmosis is the fault diagnosis result; After the fault diagnosis result is determined, the fault diagnosis result is converted into a fault diagnosis description through a defuzzification operation, and the fault diagnosis description is input into the adaptive collaborative control module (400).

5. The adaptive coordinated transmission line fault monitoring system according to claim 4, characterized in that: In the fault diagnosis unit (330), the fault severity is further divided according to the fault credibility through a trapezoidal membership function.

6. The adaptive coordinated transmission line fault monitoring system according to claim 5, characterized in that: The adaptive adjustment mechanism is specifically as follows: defining an intelligent agent, defining a fault diagnosis unit (330) including an intelligent diagnosis algorithm as an intelligent agent; A state space S is defined, wherein a state vector s in the state space S includes an operating state of the transmission line, an environmental condition, a fault diagnosis description, and a fault severity; Define the action space A, which contains the action vector a = {a 11 ,Λ,a 1u ,Λ,a 1U , a 21 ,Λ,a 2v ,Λ,a 2V , a 31 ,Λ,a 3k ,Λ,a 3L }; a 1u Corresponding sampling frequency proportional coefficient adjustment action; a 2v Corresponding sensor monitoring range adjustment action; a 3k Adjustment actions of computing resources and communication resources corresponding to the fault type; The reward function R(s,a) is defined as follows: R(s,a)=w1R accuracy (s,a)+w2R energy (s,a)+w3R response (s,a) Among them, w1, w2 and w3 are weight coefficients; R accuracy (s,a) is the fault diagnosis accuracy reward; R energy (s,a) is the energy consumption reward; R response (s,a) is the fault response timeliness reward; Among them, k1 is the accuracy importance coefficient; p acc To take action a to make the next fault diagnosis accuracy; p acc0 is the fault accuracy rate before taking action a; Among them, k2 is the energy consumption importance coefficient; E0 is the energy consumption before taking action a; E is the energy consumption after taking action a; Among them, k3 is the timeliness importance coefficient; t d0 is the time of first fault detection before taking action a; t d is the first fault detection time after taking action a; Use the Q learning algorithm to learn the optimal strategy, initialize the Q value function as a table, the rows of the table correspond to different environmental states, and the columns correspond to different actions; Use the ε-greedy strategy to select actions: Among them, ε is the exploration rate; is the environmental state s t The maximum Q value corresponding to all possible actions; The agent performs the selected action a in the system t The system state transfers to the next state s according to the physical laws of line operation, environmental changes and action influences t+1 , and at the same time calculate and feed back the reward value R to the agent based on the reward function R(s,a) t ; Based on the Q learning algorithm, the formula is updated with the newly acquired experience (s t ,a t ,R t ,s t+1 )Update the Q value: Where: Q'(s t ,a t ) is the updated Q value; Q(s t ,a t ) is the environmental state s t Take action a t Q value; α is the learning rate; γ is the discount factor; a' is the next environment state s t+1 The best action among all possible actions; is the next environment state s t+1 The maximum Q value corresponding to all possible actions is the maximum expected cumulative reward obtained from all optional actions after transferring from the current environment state to the next environment state; The Q value function is repeatedly updated until the Q value converges to obtain a corresponding optimal Q value function. When the fault diagnosis unit (330) detects that a fault exists in the transmission line, the intelligent agent uses the optimal Q value function to select and execute the optimal action from the action space according to the state vector.

Images