Power supply facility diagnosis and early warning method and system for smart power grid

Abstract

The invention belongs to the technical field of power supply facility monitoring, and discloses a power supply facility diagnosis and early warning method and system for an intelligent power grid. The method comprises the steps that a platform end receives long-term operation data collected by a facility end; the platform end analyzes the long-term operation data, obtains analysis data and judges whether an abnormal instruction is generated or not; if the abnormal instruction is generated, the platform end receives real-time operation data collected by the facility end; the platform end analyzes the analysis data and the real-time operation data and judges whether a descending instruction is generated or not; if a descending instruction is generated, the platform end diagnoses the residual life of the power grid facility according to the analysis data and the real-time operation data; the platform end generates an early warning instruction, and sends the early warning instruction, the analysis data, the real-time operation data and the residual life to the corresponding facility end; according to the invention, deep learning and dynamic decision-making of power grid facility life management are realized, and the management efficiency and operation safety of the power grid facility are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power supply facility monitoring, and more specifically, to a power supply facility diagnosis and early warning method and system for a smart grid. Background Art

[0002] Life monitoring and diagnosis of power grid facilities are key to ensuring safe and reliable operation. They can reduce maintenance costs, enable personalized management, and allow for proactive planning. Traditional methods for monitoring the life of power grid facilities rely primarily on manual monitoring, which lacks real-time performance and automation, is inefficient, and often subject to significant subjective factors.

[0003] With the development of the Internet, smart grids have become one of the main development directions of power systems, and therefore intelligent methods for monitoring and diagnosing the life of power grid facilities have emerged. For example, the Chinese patent application with publication number CN108181580A discloses an intelligent diagnostic device and a life assessment method for solid-insulated switchgear. The device includes solid-insulated switchgear, a solid-insulated switchgear monitoring terminal, and a life assessment system. Based on the online real-time monitoring of the operating status of the switchgear, the physical data (including electrical parameters and non-electrical parameters) monitored by each sensor are compared and analyzed, and the reference model established by historical data is used to timely detect and eliminate faults. At the same time, the electrical parameters at the time of the fault are recorded and uploaded for equipment reliability and life assessment. The device can effectively grasp the operating status of the high-voltage switchgear and realize intelligent fault diagnosis of the intelligent switchgear, thereby improving the reliability of the power grid operation. Through the assessment of the equipment life, the system maintenance is reasonably arranged, and the human resource cost is effectively reduced.

[0004] However, when making abnormality decisions, these technologies all directly compare collected data with preset thresholds. However, the performance of power grid facilities changes over time, making it difficult to dynamically optimize the preset thresholds and unable to adapt to the individual differences of different power grid facilities. Furthermore, the life assessment methods only mention mechanical and electrical life assessment methods, without detailing the specific steps involved. This means that the specific life monitoring and diagnostic logic cannot be understood.

[0005] In view of this, the present invention proposes a power supply facility diagnosis and early warning method and system for smart grid to solve the above problems. Summary of the Invention

[0006] In order to overcome the above-mentioned defects of the prior art and to achieve the above-mentioned objectives, the present invention provides the following technical solution: a power supply facility diagnosis and early warning method for a smart grid, comprising:

[0007] The platform receives long-term operation data collected by the facility;

[0008] The platform analyzes long-term running data, obtains analysis data, and determines whether abnormal instructions are generated;

[0009] If an abnormal instruction is generated, the platform receives real-time operation data collected by the facility;

[0010] The platform analyzes the analytical data and real-time operation data to determine whether to generate a descending instruction;

[0011] If a descent command is generated, the platform diagnoses the remaining life of the power grid facilities based on analytical data and real-time operating data;

[0012] The platform generates early warning instructions and sends the early warning instructions, analysis data, real-time operation data and residual life to the corresponding facility end.

[0013] Furthermore, the long-term operation data includes current value and oil quality index;

[0014] Preset the collection interval and collect the current value according to the collection interval;

[0015] The oil quality indicators are calculated according to the collection interval. The calculation method of the oil quality indicators is as follows: collect oil quality impact data according to the collection interval, and the oil quality impact data include total acid value, mineral content and lead content; preset a lead content threshold, and compare the lead content with the lead content threshold; if the lead content is less than or equal to the lead content threshold, adjust the lead content in the oil quality impact data to 0; if the lead content is greater than the lead content threshold, subtract the lead content threshold from the lead content to obtain the lead content difference, and adjust the lead content in the oil quality impact data to the lead content difference; calculate the oil quality indicators according to the oil quality impact data.

[0016] Furthermore, the method for obtaining analysis data includes:

[0017] The current change curve is constructed based on the obtained n current values, and the oil quality index change curve is constructed based on the obtained n oil quality indexes; the current change curve is a curve showing the continuous change of current during the operation of the power grid facilities, and the oil quality index change curve is a curve showing the continuous change of oil quality index during the operation of the power grid facilities; the moment corresponding to the nth current value in the current change curve is marked as the predicted moment; the previous moment in the current change curve is marked as the predicted moment. Input the trained current prediction model into the current value, predict the current value corresponding to the prediction moment, and mark it as the predicted current value; The oil quality index is input into the trained index prediction model, and the oil quality index corresponding to the prediction time is predicted and marked as the predicted oil quality index;

[0018] Divide the nth current value in the current change curve by the predicted current value to obtain the current error rate, and divide the nth oil quality index in the oil quality index change curve by the predicted oil quality index to obtain the index error rate;

[0019] Subtract the nth current value from the current change curve The current value is used to obtain the current change amplitude, and the oil index in the oil index change curve is subtracted from the nth oil index. The change range of each oil product index is obtained;

[0020] The current change amplitude, index change amplitude, current error rate and index error rate are used as analysis data.

[0021] Furthermore, the method for determining whether an abnormal instruction is generated includes:

[0022] Input the analysis data into the trained probability analysis model to predict the abnormal probability of power grid facilities;

[0023] Preset probability threshold ; The abnormal probability With probability threshold Make a comparison;

[0024] like , no exception instruction is generated;

[0025] like , an exception instruction is generated.

[0026] Furthermore, the real-time operation data includes polarization rate, number of hot spots and number of defects;

[0027] The polarization rate is the degree of imbalance of the power grid facility's vibration in different directions;

[0028] The number of hot spots is the number of abnormal temperature points on the surface of power grid facilities;

[0029] The number of defects refers to the number of physical defects on the surface of the power grid facilities.

[0030] Furthermore, the method for obtaining the polarization rate is as follows: obtaining three-dimensional vibration values, which include X-axis vibration values, Y-axis vibration values, and Z-axis vibration values; marking the vibration value with the largest value in the three-dimensional vibration values ​​as the maximum vibration value, and marking the vibration value with the smallest value as the minimum vibration value; obtaining a vibration deviation value based on the difference between the maximum vibration value and the minimum vibration value; adding and averaging each vibration value in the three-dimensional vibration values ​​in sequence to obtain a vibration average value; and obtaining the polarization rate based on the ratio of the vibration deviation value to the vibration average value.

[0031] The method for obtaining the number of hot spots is as follows: obtaining a temperature distribution image; presetting a temperature threshold, comparing the temperature value of each pixel in the temperature distribution image with the temperature threshold, marking pixels with temperature values ​​greater than the temperature threshold as hot spots, and not marking pixels with temperature values ​​less than or equal to the temperature threshold; and counting the number of hot spots to obtain the number of hot spots.

[0032] The method for obtaining the number of defects includes:

[0033] A three-dimensional coordinate system is established for the power grid facility, and m coordinate points corresponding to the surface of the power grid facility are obtained; a laser sensor is used to scan multiple coordinate points on the surface of the power grid facility, that is, all m coordinate points are scanned; the laser sensor includes a laser transmitter and a laser receiver, the laser transmitter transmits a laser beam, which is a short-pulse laser beam, and the laser receiver receives the laser beam reflected from the surface of the power grid facility; the reflection time of each coordinate point is obtained, and the reflection time is the duration from laser emission to laser reception; the emission angle and reception angle of each coordinate point are then obtained, the emission angle is the angle between the laser transmitter and the horizontal plane when the laser is emitted, and the reception angle is the angle between the laser receiver and the horizontal plane when the laser is received;

[0034] According to the reflection time and light speed of each coordinate point, the laser distance of each coordinate point is calculated. The laser distance is the distance traveled by the laser from laser emission to laser reception. According to the laser distance, emission angle, and reception angle of each coordinate point, the straight-line distance of each coordinate point is calculated. The straight-line distance is the vertical distance between each coordinate point and the laser sensor.

[0035] Point cloud data is formed based on the straight-line distances corresponding to m coordinate points. The polygonal mesh method is used to reconstruct the point cloud data to obtain the three-dimensional surface structure of the power grid facility and mark it as the real-time structure. The real-time structure is compared with the preset power grid facility structure, and the number of areas with different structures is counted to obtain the number of defects.

[0036] Furthermore, the method for determining whether to generate a descending instruction includes:

[0037] The analysis data and real-time operation data are used as test data, and the test data is input into the trained life degradation analysis model to determine whether the life of the power grid facilities has declined;

[0038] Obtain the corresponding judgment result based on the predicted judgment label to determine whether the life of the power grid facilities has decreased under the conditions of the test data;

[0039] If the judgment result is that the life span is reduced, a reduction instruction is generated;

[0040] If the judgment result is that the life is constant, no descending instruction is generated.

[0041] Furthermore, the step of diagnosing the remaining life of the power grid facilities includes:

[0042] Step 1: Encode the residual lifespan into chromosomes and randomly generate N chromosomes to form the initial population;

[0043] Step 2: The fitness is the accuracy, which is determined by the inverse of the life error value, which is the absolute value of the difference between the expected life and the residual life;

[0044] Step 3: Keep the first K chromosomes with the highest fitness and copy them directly to the next generation. Randomly select U chromosomes according to the fitness probability for crossover. After generating new chromosomes, replace the individuals in the original population with the best ones. Select the remaining chromosomes to make up the population capacity N.

[0045] Step 4: Randomly select U chromosomes for crossover. After generating new chromosomes, replace the individuals in the original population with the best ones and randomly exchange the gene values ​​in the chromosomes according to the preset mutation probability.

[0046] Step 5: When the preset maximum number of iterations is reached, or a chromosome with a fitness exceeding the threshold appears in the population, the loop is terminated and the optimal solution is output. The optimal solution is the residual lifespan corresponding to the chromosome with the maximum fitness.

[0047] Furthermore, the platform analyzes the remaining life of power grid facilities and matches corresponding early warning instructions;

[0048] Obtain the initial operating life and importance of the power grid facilities; the initial operating life is the expected number of years that the power grid facilities can operate under normal operating conditions; the importance is the importance of the power grid facilities in the power grid, and the importance range is , J is an integer greater than 1;

[0049] Divide the residual life by the initial operating life to obtain the life change rate; according to the importance of the power grid facilities, the corresponding warning instructions are divided into V different levels of warning instructions. , that is, the corresponding warning instructions include the first warning instruction, ..., the Vth warning instruction; then divide 1 by the importance to obtain the matching division degree. The life change rate range corresponding to each level of warning instruction is , where t is the warning instruction level, , To match the division degree; then compare the life change rate of the power grid facilities with the life change rate range corresponding to each level of warning instructions, and use the warning instructions corresponding to the life change rate within the life change rate range and the highest level warning instructions as the warning instructions that match the power grid facilities.

[0050] A power supply facility diagnosis and early warning system for a smart grid, implementing a power supply facility diagnosis and early warning method for a smart grid, comprising:

[0051] The long-term data receiving module is used for the platform to receive the long-term operation data collected by the facility end;

[0052] The long-term data analysis module is used by the platform to analyze long-term operation data, obtain analysis data, and determine whether abnormal instructions are generated;

[0053] The real-time data receiving module, if an abnormal instruction is generated, the platform side receives the real-time operation data collected by the facility side;

[0054] The lifespan reduction analysis module is used by the platform to analyze the analysis data and real-time operation data to determine whether to generate a reduction instruction;

[0055] Life diagnosis module: If a descent command is generated, the platform will diagnose the remaining life of the power grid facilities based on analysis data and real-time operation data;

[0056] The early warning module is used to generate early warning instructions on the platform side and send the early warning instructions, analysis data, real-time operation data and residual life to the corresponding facility side.

[0057] The technical effects and advantages of the power supply facility diagnosis and early warning method and system for smart grids of the present invention are as follows:

[0058] 1. By collecting long-term operating data from the facility side and analyzing the analytical data, the operating status of the power grid facilities can be monitored in real time and abnormal situations can be discovered in a timely manner. By then collecting real-time operating data and combining it with analytical data for in-depth analysis, it is possible to deeply determine whether the life of the power grid facilities is decreasing, thereby providing early warning and generating corresponding instructions. In addition, through genetic algorithms, individualized residual life diagnosis of power grid facilities can be achieved, which greatly improves the detection accuracy and diagnostic effect, provides a scientific basis for the maintenance and update of power grid facilities, and improves the reliability and stability of power grid operation. It realizes in-depth learning and dynamic decision-making on the life management of power grid facilities, and improves the management efficiency and operational safety of power grid facilities.

[0059] 2. Based on the importance of power grid facilities, early warning instructions are divided into multiple levels, and the life change rate is calculated to match the early warning instructions of the corresponding level; this can improve the accuracy and practicality of early warnings, provide personalized and goal-oriented intelligent decision-making support for operation and maintenance personnel, help relevant operation and maintenance personnel accurately grasp the life changes of power grid facilities, optimize resource allocation and management, and provide more effective support for the management and operation of power grid facilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1This is a schematic diagram of a power supply facility diagnosis and early warning system for a smart grid according to Example 1 of the present invention;

[0061] Figure 2 This is a schematic diagram of the defect quantity acquisition principle of Example 1 of the present invention;

[0062] Figure 3 This is a schematic diagram of a power supply facility diagnosis and early warning system for a smart grid according to Embodiment 2 of the present invention;

[0063] Figure 4 This is a flow chart of a method for diagnosing and warning power supply facilities for a smart grid according to embodiment 3 of the present invention. DETAILED DESCRIPTION

[0064] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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 technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0065] Example 1

[0066] See also Figure 1 As shown, the power supply facility diagnosis and early warning system for smart grids described in this embodiment includes a long-term data receiving module, a long-term data analysis module, a real-time data receiving module, a life degradation analysis module, a life diagnosis module and an early warning module; each module is connected by wired and / or wireless means to realize data transmission between modules.

[0067] The long-term data receiving module is used for the platform to receive the long-term operation data collected by the facility end;

[0068] The platform end is the intelligent network cloud control platform, and one power grid facility is one facility end;

[0069] Long-term operating data includes current value and oil quality index;

[0070] A collection interval is preset, and current values ​​are collected according to the collection interval. The current values ​​are obtained by current sensors built into the power grid facilities. The collection interval is pre-set by technicians in this field based on the current change rate during the operation of the power grid facilities. The greater the current change amplitude, the shorter the collection interval, and the smaller the current change amplitude, the longer the collection interval.

[0071] The oil quality index is calculated based on the collection interval. The oil quality index is calculated as follows: oil quality impact data is collected based on the collection interval, and the oil quality impact data includes total acid value, mineral content, and lead content; a lead content threshold is preset, and the lead content is compared with the lead content threshold; if the lead content is less than or equal to the lead content threshold, the lead content in the oil quality impact data is adjusted to 0; if the lead content is greater than the lead content threshold, the lead content threshold is subtracted from the lead content to obtain the lead content difference, and the lead content in the oil quality impact data is adjusted to the lead content difference; the oil quality index is calculated based on the oil quality impact data, and the expression of the oil quality index is:

[0072] Where, is the oil quality index, is the standardized total acid value, The mineral content after standardization, The denominator is added with 1 to avoid the risk of dividing by zero, ensuring that In the interval [0,1], it is more in line with actual evaluation needs. 、 、 are all preset weight coefficients; the specific values ​​of the weight coefficients in the formula can be set according to actual conditions. The weight coefficients reflect the degree of influence of each oil product influencing data on the oil product indicators. Technical personnel in this field can preset corresponding weight coefficients according to the actual degree of influence of each oil product influencing data on the oil product indicators, so as to accurately evaluate the oil product indicators during the operation of the power grid facilities.

[0073] Among them, the total acid value is the hydrogen ion content in the oil of the power grid facility, and the total acid value is obtained by the hydrogen ion selective electrode installed in the oil tank of the power grid facility. The higher the total acid value, the faster the corrosion rate of the metal components in the power grid facility, the lower the oil quality index, and the shorter the life of the power grid facility, and vice versa. The mineral content is the content of minerals (such as calcium, aluminum, and magnesium) in the oil of the power grid facility, and the mineral content is obtained by a sequential spectrometer installed at the oil production port in the oil tank of the power grid facility. The higher the mineral content, the higher the conductivity of the oil, which reduces the insulation performance and bubble point. Therefore, the lower the oil quality index, the shorter the life of the power grid facility, and vice versa. The lead content is the lead ion content in the oil of the power grid facility, and the lead content is obtained by the lead ion selective electrode installed in the oil tank of the power grid facility. The more the lead content exceeds the lead content threshold, the more serious the arc corrosion problem in the power grid facility, resulting in lower oil quality index and shorter life of the power grid facility, and vice versa. The lead content threshold is pre-set by technical personnel in this field according to the lead content limit set by industry standards for power grid facilities.

[0074] The long-term data analysis module is used by the platform to analyze long-term running data, obtain analysis data, and determine whether abnormal instructions are generated.

[0075] Methods for obtaining analytical data include:

[0076] The current change curve is constructed based on the obtained n current values, and the oil quality index change curve is constructed based on the obtained n oil quality indexes; the current change curve is a curve showing the continuous change of current during the operation of the power grid facilities, and the oil quality index change curve is a curve showing the continuous change of oil quality index during the operation of the power grid facilities; the moment corresponding to the nth current value in the current change curve is marked as the predicted moment; the previous moment in the current change curve is marked as the predicted moment. Input the trained current prediction model into the current value, predict the current value corresponding to the prediction moment, and mark it as the predicted current value; The oil quality indicators are input into the trained indicator prediction model to predict the oil quality indicators corresponding to the prediction moment and mark them as the predicted oil quality indicators. Among them, since the oil quality indicators and current values ​​are both obtained according to the collection interval, the moment corresponding to the nth oil quality indicator in the oil quality indicator change curve is also the prediction moment.

[0077] The specific training process of the current prediction model includes:

[0078] Collect a continuous current value in advance according to the collection interval, ; Preset the sliding step size L and the sliding window length according to the actual experience of those skilled in the art; convert a current values ​​into corresponding training samples using the sliding window method, use the training samples as the input of the current prediction model, predict the current value after the sliding step size L as the output, and use the subsequent current value of each training sample as the prediction target. The prediction results are evaluated for model accuracy using the mean absolute percentage error MAPE. When the calculated MAPE is less than the preset MAPE, the current prediction model training is completed; wherein, the calculation formula of MAPE is: ,in, is the prediction target corresponding to the d-th predicted current value, is the dth predicted current value, and R is the number of predicted current values; a current prediction model is generated for predicting the current value at a future moment based on the current value; wherein the current prediction model is an RNN neural network model; it should be noted that the preset MAPE is pre-set by the staff according to the accuracy required by the model.

[0079] For example, the collected current value data set A contains 10 current values. , is the hth current value, , use a sliding window to construct multiple training samples, define the sliding window length as 3, the sliding step L as 1, each training sample contains 3 consecutive current values, and the next current value of the 3 consecutive current values ​​is used as the prediction target; for example:

[0080] As training data, The corresponding prediction target is ;

[0081] As training data, The corresponding prediction target is ; And so on, it is used to train the current prediction model;

[0082] The specific training process of the indicator prediction model is consistent with that of the current prediction model, and it is also an RNN neural network model.

[0083] Divide the nth current value in the current change curve by the predicted current value to obtain the current error rate, and divide the nth oil quality index in the oil quality index change curve by the predicted oil quality index to obtain the index error rate;

[0084] Subtract the nth current value from the current change curve The current value is used to obtain the current change amplitude, and the oil index in the oil index change curve is subtracted from the nth oil index. The change range of each oil product index is obtained.

[0085] The current change amplitude, index change amplitude, current error rate and index error rate are used as analysis data.

[0086] Methods for determining whether an abnormal instruction is generated include:

[0087] Input the analysis data into the trained probability analysis model to predict the abnormal probability of power grid facilities;

[0088] The specific training process of the probability analysis model includes:

[0089] The abnormal probabilities corresponding to b groups of analysis data are collected in advance, and the analysis data and the corresponding abnormal probabilities are converted into a corresponding set of feature vectors, where b is an integer greater than 1; wherein the abnormal probabilities corresponding to the analysis data are collected by those skilled in the art during the historical power grid facility life monitoring process, b groups of different analysis data are collected, and the corresponding abnormal probabilities are analyzed respectively under the conditions of each group of analysis data in combination with actual experience.

[0090] Each set of feature vectors is used as the input of the probability analysis model. The probability analysis model takes a set of predicted abnormality probabilities corresponding to each set of analysis data as output, and the actual abnormality probability corresponding to each set of analysis data as the prediction target. The actual abnormality probability is the abnormality probability corresponding to the analysis data collected in advance; minimizing the sum of the prediction errors of all analysis data is used as the training goal; the probability analysis model is trained until the sum of the prediction errors reaches convergence and the training is stopped.

[0091] The above-mentioned probability analysis model is specifically a deep neural network model, which includes an input layer, a hidden layer and an output layer; each hidden layer includes multiple neurons, each neuron is connected to the neurons in the next layer, and the connection contains weights, which determine the importance and influence of data transmission in the neural network; an activation function is applied to each neuron between the hidden layer and the output layer, and the activation function reflects nonlinearity, allowing the network to learn more complex patterns and features.

[0092] Preset probability threshold , probability threshold It is pre-set by those skilled in the art according to the required detection accuracy.

[0093] The abnormal probability With probability threshold Make a comparison;

[0094] like , no abnormal instruction is generated; this indicates that the probability of abnormality in the power grid facilities is small, and no further diagnostic analysis is required for the power grid facilities;

[0095] like , an abnormal instruction is generated; this indicates that the probability of an abnormality occurring in the power grid facilities is high, which may affect the life of the power grid facilities and requires subsequent more in-depth diagnostic analysis of the power grid facilities.

[0096] The real-time data receiving module generates an abnormal instruction, and the platform receives the real-time operation data collected by the facility end.

[0097] Real-time operational data includes polarization rate, number of hot spots, and number of defects.

[0098] Polarization rate is the degree of imbalance in the vibration of power grid facilities in different directions. The larger the polarization rate, the more severe the vibration imbalance of the power grid facilities in different directions. Vibration imbalance is usually caused by mechanical problems such as aging materials and damaged components within the power grid facilities. It also means that mechanical stress is concentrated in the power grid facilities, making them more prone to failures such as fracture. Therefore, the life of the power grid facilities is shortened, and vice versa.

[0099] The polarization rate is obtained by using a triaxial vibration sensor installed on the surface of the power grid facility to obtain three-dimensional vibration values, including X-axis vibration value, Y-axis vibration value, and Z-axis vibration value. The vibration value with the largest value in the three-dimensional vibration value is marked as the maximum vibration value, and the vibration value with the smallest value is marked as the minimum vibration value. The vibration deviation value is obtained based on the difference between the maximum vibration value and the minimum vibration value. The expression of the vibration deviation value is: Where, is the vibration deviation value, is the maximum vibration value, is the minimum vibration value; each vibration value in the three-dimensional vibration value is added and averaged to obtain the vibration average value. The expression of the vibration average value is: Where, is the average vibration value, is the X-axis vibration value, is the Y-axis vibration value, is the Z-axis vibration value; the polarization rate is obtained according to the ratio of the vibration deviation value to the vibration average value. The expression of the polarization rate is: Where, is the polarization rate.

[0100] The number of hot spots refers to the number of abnormal temperature points on the surface of the power grid facilities. The greater the number of hot spots, the larger the abnormal temperature area on the surface of the power grid facilities, and the presence of problems such as insulation damage or poor contact. Therefore, the life of the power grid facilities is shorter, and vice versa.

[0101] The method for obtaining the number of hot spots is as follows: using a thermal imager aimed at the power grid facilities to obtain a temperature distribution image; presetting a temperature threshold, comparing the temperature value of each pixel in the temperature distribution image with the temperature threshold, marking pixels with temperature values ​​greater than the temperature threshold as hot spots, and not marking pixels with temperature values ​​less than or equal to the temperature threshold; counting the number of hot spots to obtain the number of hot spots; wherein the temperature threshold is obtained by technicians in this field based on the technical manual or parameter table of the power grid facilities to obtain the normal operating temperature range, and the temperature threshold is preset based on the maximum value of the normal operating temperature range.

[0102] The number of defects refers to the number of physical defects on the surface of the power grid facility. The greater the number of defects, the more defects there are on the surface of the power grid facility and the more failure points of the damaged power grid facility. This will destroy the structural continuity, reduce the stiffness and durability of the material, and weaken the insulation protection capability, thus resulting in a shorter life of the power grid facility, and vice versa.

[0103] See also Figure 2 As shown, the method for obtaining the number of defects includes:

[0104] A three-dimensional coordinate system is established for the power grid facility, and m coordinate points corresponding to the surface of the power grid facility are obtained; a laser sensor aimed at the power grid facility is used to scan multiple coordinate points on the surface of the power grid facility, that is, all m coordinate points are scanned; the laser sensor includes a laser transmitter and a laser receiver, the laser transmitter transmits a laser beam, which is a short-pulse laser beam, and the laser receiver receives the laser beam reflected from the surface of the power grid facility; the reflection time of each coordinate point is obtained, and the reflection time is the duration from laser emission to laser reception. The reflection time is obtained by the built-in timer of the laser transmitter. When the laser transmitter emits the laser beam, the timer starts timing, and when the laser receiver receives the laser beam, the timer stops timing. The data collected by the timer is the reflection time; then the emission angle and receiving angle of each coordinate point are obtained, the emission angle is the angle between the laser transmitter and the horizontal plane when the laser is emitted, and the receiving angle is the angle between the laser receiver and the horizontal plane when the laser is received. The emission angle is obtained by the absolute encoder built into the laser transmitter, and the receiving angle is obtained by the absolute encoder built into the laser receiver.

[0105] According to the reflection time and light speed of each coordinate point, the laser distance of each coordinate point is calculated. The laser distance is the distance traveled by the laser from the laser emission to the laser reception. The expression of the laser distance is: Where, is the laser distance, is the reflection time, is the speed of light, which is defined as ; Calculate the straight-line distance of each coordinate point based on the laser distance, emission angle, and receiving angle of each coordinate point. The straight-line distance is the vertical distance between each coordinate point and the laser sensor. The expression for the straight-line distance is: Where, is the straight-line distance, is the emission angle, is the receiving angle;

[0106] Point cloud data is formed based on the straight-line distances corresponding to m coordinate points, and the polygonal mesh method is used to mesh the point cloud data to obtain the three-dimensional surface structure of the power grid facility and mark it as the real-time structure; the real-time structure is compared with the preset power grid facility structure, and the number of areas with different structures is counted to obtain the number of defects; the polygonal mesh method is a prior art and will not be elaborated on here; the power grid facility structure is modeled and set by those skilled in the art based on the design drawings of the power grid facility.

[0107] The lifespan reduction analysis module is used by the platform to analyze the analysis data and real-time operation data to determine whether to generate a reduction instruction;

[0108] Methods for determining whether to generate a descending instruction include:

[0109] The analysis data and real-time operation data are used as test data, and the test data is input into the trained life degradation analysis model to determine whether the life of the power grid facilities has declined;

[0110] The specific training process of the lifespan reduction analysis model includes:

[0111] A corresponding judgment result is set in advance for c groups of test data, where c is an integer greater than 1, and the judgment result includes life reduction and life stability. Different digital labels are set for life reduction and life stability. For example, the digital label for life reduction is set to 0, and the digital label for life stability is set to 1. The judgment result corresponding to the test data is collected by those skilled in the art during the historical power grid facility life monitoring process. Those skilled in the art, based on actual experience, sequentially judge whether the life of the power grid facility decreases under the conditions of the c groups of different test data, and sequentially set the corresponding judgment results for the c groups of different test data.

[0112] Mark the digital label of the judgment result as the judgment label, and convert the test data and the corresponding judgment label into a corresponding set of feature vectors;

[0113] Each set of feature vectors is used as the input of the life reduction analysis model. The life reduction analysis model uses a set of predicted judgment labels corresponding to each set of test data as output, and the actual judgment labels corresponding to each set of test data as the prediction target. The actual judgment labels are the digital labels of the judgment results corresponding to the test data set set above. The training goal is to minimize the sum of the prediction errors of all test data. The calculation formula of the prediction error is: ,in is the prediction error, is the group number of the feature vector corresponding to the test data, For the The predicted judgment label corresponding to the group test data, For the The actual judgment labels corresponding to the test data of the group are obtained; the lifespan reduction analysis model is trained until the sum of the prediction errors reaches convergence and the training is stopped;

[0114] The above-mentioned lifespan reduction analysis model is specifically a deep neural network model;

[0115] Obtain the corresponding judgment result based on the predicted judgment label to determine whether the life of the power grid facilities has decreased under the conditions of the test data;

[0116] If the judgment result is that the life span is reduced, a reduction instruction is generated;

[0117] If the judgment result is that the life span is constant, no descending instruction is generated;

[0118] Life diagnosis module: If a descent command is generated, the platform will diagnose the remaining life of the power grid facilities based on analysis data and real-time operation data;

[0119] The steps to diagnose the residual life of grid facilities include:

[0120] Step 1: Encode the residual lifespan into chromosomes and randomly generate N chromosomes to form the initial population. The residual lifespan range is determined based on the facility technical specifications.

[0121] Step 2: The fitness is the accuracy, which is determined by the inverse of the life error value, which is the absolute value of the difference between the expected life and the residual life;

[0122] Step 3: Keep the first K chromosomes with the highest fitness and copy them directly to the next generation. Randomly select U chromosomes according to the fitness probability for crossover. After generating new chromosomes, replace the individuals in the original population with the best ones. Select the remaining chromosomes to make up the population capacity N.

[0123] Step 4: Randomly select U chromosomes for crossover. After generating new chromosomes, replace the individuals in the original population with the best ones. Randomly exchange the gene values ​​in the chromosomes according to the preset mutation probability (such as 2%) to enhance population diversity.

[0124] Step 5: When the preset maximum number of iterations is reached, or a chromosome with a fitness exceeding the threshold Q appears in the population, the loop is terminated and the optimal solution is output. The optimal solution is the residual lifespan corresponding to the chromosome with the maximum fitness.

[0125] The expected lifespan is calculated in the form of an inverse by weighted integration of negative factors such as the current variation amplitude and current error rate, the index variation amplitude and index error rate, the number of hot spots, the number of defects and the polarization rate. The larger the sum of the independent weight items in the denominator, the shorter the expected lifespan.

[0126] The larger the life error value, the smaller the corresponding accuracy, and vice versa. The expression of expected life is as follows:

[0127] Where, For life expectancy, is the current variation amplitude, is the current error rate, is the index change range, is the indicator error rate, is the number of hotspots, is the number of defects, For the polarization rate, the square root of the current change amplitude is used to suppress the nonlinear effect of extreme values; It represents the coupling effect of the current change amplitude and the error rate (the unit is unified as the square of the current); 、 Respectively express the independent impact of the indicator change range and its error to avoid unit conflicts; 、 、 Still linearly superimposed, which is consistent with the independent defect assumption; 、 、 、 、 、 are all weight factors; the specific values ​​of the weight factors can be set according to actual conditions. The weight factors reflect the impact of each test data on the expected life of the power grid facilities. Through historical data regression analysis or machine learning optimization, the actual correlation between each parameter and life is quantified, and the optimal weight is automatically fitted. Alternatively, tools such as the Analytic Hierarchy Process (AHP) are used to integrate the subjective scores of field experts on the importance of parameters and convert them into objective weights.

[0128] It should be noted that the data in the analysis data are all relevant impact data on the expected life of power grid facilities. The larger each data in the analysis data is, the shorter the expected life of the power grid facilities will be, and vice versa.

[0129] The early warning module is used to generate early warning instructions on the platform side, and send the early warning instructions, analysis data, real-time operation data and residual life to the corresponding facility side, so that the relevant operation and maintenance personnel on the corresponding facility side can promptly learn about the abnormal conditions of the power grid facilities and various related data, and can effectively maintain and repair the power grid facilities.

[0130] This embodiment collects long-term operating data from the facility end and analyzes the analytical data, so as to monitor the operating status of the power grid facilities in real time and detect abnormal situations in a timely manner. It then collects real-time operating data and combines it with the analytical data for in-depth analysis, so as to make an in-depth assessment of whether the life of the power grid facilities is decreasing, thereby providing early warning and generating corresponding instructions. In addition, through genetic algorithms, individualized residual life diagnosis of power grid facilities is achieved, which greatly improves the detection accuracy and diagnostic effect, provides a scientific basis for the maintenance and updating of power grid facilities, and improves the reliability and stability of power grid operation. It realizes in-depth learning and dynamic decision-making on the life management of power grid facilities, and improves the management efficiency and operational safety of power grid facilities.

[0131] Example 2

[0132] See also Figure 3As shown, this embodiment further improves the design based on the first embodiment. In the first embodiment, only early warning instructions are generated. That is, the early warning instructions generated are the same under different life change conditions of the power grid facilities, which cannot help the relevant operation and maintenance personnel to accurately understand the life change of the power grid facilities. In order to help the relevant operation and maintenance personnel understand the life reduction of the power grid facilities more intuitively and clearly, this embodiment provides a power supply facility diagnosis and early warning system for smart grids, which also includes an early warning instruction matching module;

[0133] The early warning instruction matching module is used by the platform to analyze the remaining life of power grid facilities and match corresponding early warning instructions;

[0134] Obtain the initial operating life and importance of the power grid facilities; the initial operating life is the expected number of years that the power grid facilities can operate under normal operating conditions, and the initial operating life is obtained based on the model parameter table provided by the power grid facility manufacturer; the importance is the importance of the power grid facilities in the power grid, and the importance is assessed by technical personnel in the field based on the power grid topology to assess the impact of the safety of each power grid facility on the power grid, and combined with the power system analysis software, through circuit breaker simulation analysis of the impact area of ​​each power grid facility when the power grid fails, and comprehensively assess the importance of each power grid facility. The higher the importance of the power grid facility, the higher the importance, and the lower the importance of the power grid facility, the lower the importance; the importance range is , J is an integer greater than 1, and in this embodiment, J is preferably 5.

[0135] Divide the residual life by the initial operating life to obtain the life change rate; according to the importance of the power grid facilities, the corresponding warning instructions are divided into V different levels of warning instructions. , that is, the corresponding warning instructions include the first warning instruction, ..., the Vth warning instruction; then divide 1 by the importance to obtain the matching division degree. The life change rate range corresponding to each level of warning instruction is , where t is the warning instruction level, , To match the division degree; then compare the life change rate of the power grid facilities with the life change rate range corresponding to each level of warning instructions, and use the warning instructions corresponding to the life change rate within the life change rate range and the highest level warning instructions as the warning instructions that match the power grid facilities.

[0136] For example, the importance of a power grid facility is 4, so the warning instructions corresponding to the power grid facility are divided into 4 levels of warning instructions, that is, the corresponding warning instructions include the first warning instruction, the second warning instruction, the third warning instruction and the fourth warning instruction; the matching division degree is , where the first warning instruction corresponds to the life change rate range of , the second warning instruction corresponds to the life change rate range of , the range of life change rate corresponding to the third warning instruction is , the life span change rate range corresponding to the fourth warning instruction is ; and the residual life of the power grid facility is 20, and the initial operating life is 30, so the life change rate of the power grid facility is ,and , therefore the warning instruction matched by this power grid facility is the third warning instruction.

[0137] It should be noted that the reason why the warning instructions matched with power grid facilities include the highest level warning instructions is that different power grid facilities have different importances, and the number of warning instruction levels divided into corresponding warning instructions is also different. Therefore, the same level of warning instructions have different life change rates in power grid facilities with different importances. If the highest level warning instructions are not included, the operation and maintenance personnel cannot understand the meaning of the generated warning instructions. For example: the importance of a power grid facility is 2, and the importance of a power grid facility is 5. At this time, both power grid facilities generate the first warning instruction. For the power grid facility with an importance of 5, the life change rate is , while for the power grid facilities with importance 2, the life change rate is Therefore, it is necessary to combine it with the highest level of early warning instructions to help operation and maintenance personnel accurately understand the meaning of the early warning instructions matched by the power grid facilities.

[0138] This embodiment divides warning instructions into multiple levels according to the importance of power grid facilities, and calculates the life change rate to match warning instructions of corresponding levels. It can improve the accuracy and practicality of warnings, provide personalized and goal-oriented intelligent decision-making support for operation and maintenance personnel, help relevant operation and maintenance personnel accurately grasp the life changes of power grid facilities, optimize resource allocation and management, and provide more effective support for the management and operation of power grid facilities.

[0139] Example 3

[0140] See also Figure 4 As shown, for parts not described in detail in this embodiment, please refer to the description of Embodiment 1 and Embodiment 2. A method for diagnosing and warning power supply facilities for a smart grid is provided, the method comprising:

[0141] The platform receives long-term operation data collected by the facility;

[0142] The platform analyzes long-term running data, obtains analysis data, and determines whether abnormal instructions are generated;

[0143] If an abnormal instruction is generated, the platform receives real-time operation data collected by the facility;

[0144] The platform analyzes the analytical data and real-time operation data to determine whether to generate a descending instruction;

[0145] If a descent command is generated, the platform diagnoses the remaining life of the power grid facilities based on analytical data and real-time operating data;

[0146] The platform generates early warning instructions and sends the early warning instructions, analysis data, real-time operation data and residual life to the corresponding facility end.

[0147] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

[0148] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A power supply facility diagnosis and early warning method for a smart grid, characterized in that: include: The platform receives long-term operation data collected by the facility; The platform analyzes long-term running data, obtains analysis data, and determines whether abnormal instructions are generated; If an abnormal instruction is generated, the platform receives real-time operation data collected by the facility; The platform analyzes the analytical data and real-time operation data to determine whether to generate a descending instruction; If a descent command is generated, the platform diagnoses the remaining life of the power grid facilities based on analytical data and real-time operating data; The platform generates early warning instructions and sends the early warning instructions, analysis data, real-time operation data and residual life to the corresponding facility end.

2. The power supply facility diagnosis and early warning method for smart grid according to claim 1, characterized in that: The long-term operation data includes current value and oil quality index; Preset the collection interval and collect the current value according to the collection interval; The oil quality indicators are calculated according to the collection interval. The calculation method of the oil quality indicators is as follows: collect oil quality impact data according to the collection interval, and the oil quality impact data include total acid value, mineral content and lead content; preset a lead content threshold, and compare the lead content with the lead content threshold; if the lead content is less than or equal to the lead content threshold, adjust the lead content in the oil quality impact data to 0; if the lead content is greater than the lead content threshold, subtract the lead content threshold from the lead content to obtain the lead content difference, and adjust the lead content in the oil quality impact data to the lead content difference; calculate the oil quality indicators according to the oil quality impact data.

3. The power supply facility diagnosis and early warning method for smart grid according to claim 2, characterized in that: The method for obtaining analysis data includes: The current change curve is constructed based on the obtained n current values, and the oil quality index change curve is constructed based on the obtained n oil quality indexes; the current change curve is a curve showing the continuous change of current during the operation of the power grid facilities, and the oil quality index change curve is a curve showing the continuous change of oil quality index during the operation of the power grid facilities; the moment corresponding to the nth current value in the current change curve is marked as the predicted moment; the previous moment in the current change curve is marked as the predicted moment. Input the trained current prediction model into the current value, predict the current value corresponding to the prediction moment, and mark it as the predicted current value; The oil quality index is input into the trained index prediction model, and the oil quality index corresponding to the prediction time is predicted and marked as the predicted oil quality index; Divide the nth current value in the current change curve by the predicted current value to obtain the current error rate, and divide the nth oil quality index in the oil quality index change curve by the predicted oil quality index to obtain the index error rate; Subtract the nth current value from the current change curve The current value is used to obtain the current change amplitude, and the nth oil index in the oil index change curve is subtracted from the nth oil index. The change range of each oil product index is obtained; The current change amplitude, index change amplitude, current error rate and index error rate are used as analysis data.

4. The power supply facility diagnosis and early warning method for smart grid according to claim 3, characterized in that: The method for determining whether an abnormal instruction is generated includes: Input the analysis data into the trained probability analysis model to predict the abnormal probability of power grid facilities; Preset probability threshold ; The abnormal probability With probability threshold Make a comparison; like , no exception instruction is generated; like , an exception instruction is generated.

5. The power supply facility diagnosis and early warning method for smart grid according to claim 4, characterized in that: The real-time operation data includes polarization rate, number of hot spots and number of defects; The polarization rate is the degree of imbalance of the power grid facility's vibration in different directions; The number of hot spots is the number of abnormal temperature points on the surface of power grid facilities; The number of defects refers to the number of physical defects on the surface of the power grid facilities.

6. The power supply facility diagnosis and early warning method for smart grid according to claim 5, characterized in that: The polarization rate is obtained by: obtaining three-dimensional vibration values, which include X-axis vibration values, Y-axis vibration values, and Z-axis vibration values; marking the vibration value with the largest value in the three-dimensional vibration values ​​as the maximum vibration value, and marking the vibration value with the smallest value as the minimum vibration value; obtaining a vibration deviation value based on the difference between the maximum vibration value and the minimum vibration value; adding and averaging each vibration value in the three-dimensional vibration values ​​in sequence to obtain a vibration average value; and obtaining the polarization rate based on the ratio of the vibration deviation value to the vibration average value. The method for obtaining the number of hot spots is as follows: obtaining a temperature distribution image; presetting a temperature threshold, comparing the temperature value of each pixel in the temperature distribution image with the temperature threshold, marking pixels with a temperature value greater than the temperature threshold as hot spots, and not marking pixels with a temperature value less than or equal to the temperature threshold; Count the number of hot spots to obtain the number of hot spots; The method for obtaining the number of defects includes: A three-dimensional coordinate system is established for the power grid facility, and m coordinate points corresponding to the surface of the power grid facility are obtained; a laser sensor is used to scan multiple coordinate points on the surface of the power grid facility, that is, all m coordinate points are scanned; the laser sensor includes a laser transmitter and a laser receiver, the laser transmitter transmits a laser beam, which is a short-pulse laser beam, and the laser receiver receives the laser beam reflected from the surface of the power grid facility; the reflection time of each coordinate point is obtained, and the reflection time is the duration from laser emission to laser reception; the emission angle and reception angle of each coordinate point are then obtained, the emission angle is the angle between the laser transmitter and the horizontal plane when the laser is emitted, and the reception angle is the angle between the laser receiver and the horizontal plane when the laser is received; According to the reflection time and light speed of each coordinate point, the laser distance of each coordinate point is calculated. The laser distance is the distance traveled by the laser from laser emission to laser reception. According to the laser distance, emission angle, and reception angle of each coordinate point, the straight-line distance of each coordinate point is calculated. The straight-line distance is the vertical distance between each coordinate point and the laser sensor. Point cloud data is formed based on the straight-line distances corresponding to m coordinate points. The polygonal mesh method is used to reconstruct the point cloud data to obtain the three-dimensional surface structure of the power grid facility and mark it as the real-time structure. The real-time structure is compared with the preset power grid facility structure, and the number of areas with different structures is counted to obtain the number of defects.

7. The power supply facility diagnosis and early warning method for smart grid according to claim 6, characterized in that: The method for determining whether to generate a descending instruction includes: The analysis data and real-time operation data are used as test data, and the test data is input into the trained life degradation analysis model to determine whether the life of the power grid facilities has declined; Obtain the corresponding judgment result based on the predicted judgment label to determine whether the life of the power grid facilities has decreased under the conditions of the test data; If the judgment result is that the life span is reduced, a reduction instruction is generated; If the judgment result is that the life is constant, no descending instruction is generated.

8. The power supply facility diagnosis and early warning method for smart grid according to claim 7, characterized in that: The step of diagnosing the residual life of the power grid facilities comprises: Step 1: Encode the residual lifespan into chromosomes and randomly generate N chromosomes to form the initial population; Step 2: The fitness is the accuracy, which is determined by the inverse of the life error value, which is the absolute value of the difference between the expected life and the residual life; Step 3: Keep the first K chromosomes with the highest fitness and copy them directly to the next generation. Randomly select U chromosomes for crossover according to the fitness probability. After generating new chromosomes, replace the individuals in the original population with the best ones, and select the remaining chromosomes to make up the population capacity N. Step 4: Randomly select U chromosomes for crossover. After generating new chromosomes, replace the individuals in the original population with the best ones and randomly exchange the gene values ​​in the chromosomes according to the preset mutation probability. Step 5: When the preset maximum number of iterations is reached, or a chromosome with a fitness exceeding the threshold appears in the population, the loop is terminated and the optimal solution is output. The optimal solution is the residual lifespan corresponding to the chromosome with the maximum fitness.

9. The power supply facility diagnosis and early warning method for smart grid according to claim 8, characterized in that: The platform analyzes the remaining life of power grid facilities and matches corresponding early warning instructions; Obtain the initial operating life and importance of the power grid facilities; the initial operating life is the expected number of years that the power grid facilities can operate under normal operating conditions; Importance refers to the importance of the power grid facilities in the power grid, and the importance range is , J is an integer greater than 1; Divide the residual life by the initial operating life to obtain the life change rate; according to the importance of the power grid facilities, the corresponding warning instructions are divided into V different levels of warning instructions. , that is, the corresponding warning instructions include the first warning instruction, ..., the Vth warning instruction; then divide 1 by the importance to obtain the matching division degree. The life change rate range corresponding to each level of warning instruction is , where t is the warning instruction level, , To match the division degree; then compare the life change rate of the power grid facilities with the life change rate range corresponding to each level of warning instructions, and use the warning instructions corresponding to the life change rate within the life change rate range and the highest level warning instructions as the warning instructions that match the power grid facilities.

10. A power supply facility diagnosis and early warning system for a smart grid, implementing a power supply facility diagnosis and early warning method for a smart grid according to any one of claims 1 to 9, characterized in that: include: The long-term data receiving module is used for the platform to receive the long-term operation data collected by the facility end; The long-term data analysis module is used by the platform to analyze long-term operation data, obtain analysis data, and determine whether abnormal instructions are generated; The real-time data receiving module, if an abnormal instruction is generated, the platform side receives the real-time operation data collected by the facility side; The lifespan reduction analysis module is used by the platform to analyze the analysis data and real-time operation data to determine whether to generate a reduction instruction; Life diagnosis module: If a descent command is generated, the platform will diagnose the remaining life of the power grid facilities based on analysis data and real-time operation data; The early warning module is used to generate early warning instructions on the platform side and send the early warning instructions, analysis data, real-time operation data and residual life to the corresponding facility side.

Images