An adaptive data acquisition method and system of an intelligent distribution and transformation monitoring unit

Abstract

The application relates to an adaptive data acquisition method and system of an intelligent distribution transformer monitoring unit, which comprises the following steps: through the intelligent distribution transformer monitoring unit, a plurality of operating parameters in a power distribution network are monitored in real time; the plurality of monitored operating parameters are classified according to the operating state of the power distribution network, and are divided into normal monitoring data and abnormal monitoring data; the current state of the power distribution network is analyzed based on the real-time monitoring data, and it is judged whether the power distribution network is in a normal state, a load state or an abnormal state; a distribution transformer monitoring prediction model is constructed, which is used for identifying the operating state of the power distribution network and predicting potential faults or abnormal conditions; according to the prediction result, the frequency and type of data acquisition are determined; in the normal operating state of the power distribution network, the plurality of operating parameters in the power distribution network are acquired at a low frequency; when the load of the power distribution network changes greatly or is about to fail, the data acquisition frequency is increased; when a fault or an abnormality occurs, a highest-frequency data acquisition mode is started immediately, and detailed power distribution network data is collected in real time.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent monitoring of distribution transformers, and particularly to an adaptive data acquisition method and system for an intelligent distribution transformer monitoring unit. BACKGROUND

[0002] The intelligent monitoring unit of a distribution transformer generally collects data through various sensors (such as current, voltage, temperature, humidity sensors, etc.) to determine the working state, load, fault alarm, etc. of the distribution network. In traditional distribution transformer monitoring systems, the data acquisition frequency is generally fixed. This approach has some significant limitations in distribution network monitoring, especially in the context of increasing intelligence and equipment in the distribution network.

[0003] Traditional intelligent monitoring systems of distribution transformers use fixed frequency data acquisition, such as collecting data once every fixed time (e.g. every 5 minutes, every hour). One major problem with this approach is that the acquisition frequency cannot be flexibly adjusted according to the actual operating state, load level or environmental changes of the distribution network. When the system load is light and the distribution network is working normally, data is still collected at a fixed frequency, which can result in unnecessary data redundancy, increasing the burden of data transmission, storage and processing, and wasting system resources and bandwidth.

[0004] In addition, when a sudden event occurs (such as overload, short circuit, overvoltage, distribution network failure, etc.), the fixed frequency acquisition mode cannot capture key change data in time, resulting in the omission of potential faults or abnormalities. For example, in the case of a sharp change in load and abnormal voltage fluctuations in the distribution network, if the acquisition period is too long, the system may miss important fault diagnosis information, affecting the timeliness of fault location, judgment and repair.

[0005] For some stable distribution networks, fixed frequency acquisition generates a large amount of redundant data, which is of little help to the normal operation of the distribution network, but increases the complexity of data processing. For example, when the distribution network is in a stable working state for a long time, frequent acquisition and uploading of meaningless data can result in excessive storage space occupation, increasing the difficulty of data analysis.

[0006] With the gradual intelligence of the distribution network, more and more devices are connected to the distribution network system, including distribution substations, transformers, feeder automation devices, switchgear, smart meters, and leakage protectors. The status of each device needs to be monitored, and the amount of data increases dramatically. In this case, the traditional static acquisition method increases the data transmission burden, causing network bandwidth pressure, and even causing data loss, transmission delay, etc., affecting the real-time and reliability of the entire distribution network. SUMMARY

[0007] To this end, the technical problem to be solved by the present application is to overcome the limitation of the conventional distribution transformer intelligent monitoring unit in the prior art which adopts fixed frequency data acquisition, and to provide an adaptive data acquisition method and system for an intelligent distribution transformer monitoring unit, which dynamically adjusts the frequency and content of data acquisition to meet the actual needs of different distribution transformer networks, thereby improving the efficiency and reliability of the monitoring system.

[0008] To solve the above technical problems, the present application provides an adaptive data acquisition method for an intelligent distribution transformer monitoring unit, comprising:

[0009] The intelligent distribution transformer monitoring unit is used to monitor a plurality of operating parameters in the power distribution network in real time.

[0010] The monitored plurality of operating parameters are classified according to the operating state of the power distribution network, into routine monitoring data and abnormal monitoring data, and the current state of the power distribution network is analyzed based on the real-time monitoring data to determine whether it is in a normal, load or abnormal state.

[0011] A distribution transformer monitoring and prediction model is constructed to identify the operating state of the power distribution network and predict potential faults or abnormal conditions, and the frequency and type of data acquisition are determined according to the prediction results.

[0012] In the normal operating state of the power distribution network, low-frequency data acquisition is used to acquire a plurality of operating parameters in the power distribution network.

[0013] When the load of the power distribution network changes greatly or is about to fail, the data acquisition frequency is increased.

[0014] When a fault or abnormality occurs, the highest frequency data acquisition mode is immediately started to collect detailed power distribution network data in real time.

[0015] In an embodiment of the present application, the distribution transformer monitoring and prediction model is constructed, comprising:

[0016] The historical data of the distribution transformer network operation are obtained, including current, voltage, load, power factor, temperature and humidity.

[0017] The historical operating data and known power distribution network fault or abnormal information are used to label the data for identifying abnormal events.

[0018] Machine learning algorithms are used to classify real-time data to identify the operating state of the power distribution network, and the power distribution network is divided into a normal state or a fault state.

[0019] A long-term memory network is used to construct an initialized prediction model, the prediction model is trained based on historical data to capture the timing patterns and trends in the operation of the power distribution network, and potential faults are warned in advance to obtain the distribution transformer monitoring and prediction model.

[0020] In an embodiment of the present application, the prediction model is trained based on historical data, including:

[0021] All historical data is divided into K subsets, and K-1 subsets are selected as the training set and the remaining 1 subset is selected as the validation set each time training;

[0022] The loss and accuracy on the validation set are recorded each time the validation is performed;

[0023] The average of K validation results is calculated as the overall performance evaluation of the model, and the average loss and average accuracy are calculated as the evaluation standard of the model.

[0024] In an embodiment of the present application, when the power distribution network is in normal operation state, the collection frequency of each parameter is adjusted by weighted proportion, including:

[0025] Define the parameters to be monitored in the power distribution network, and give each parameter a preliminary weight value;

[0026] According to the frequency, type and severity of the fault in the historical data of each parameter, the correlation of each parameter with the fault is calculated, and the Pearson correlation coefficient is used to measure the linear correlation between each parameter and the occurrence of the fault, assuming that the correlation of each monitoring parameter with the occurrence of the fault is r:

[0027] ;

[0028] Wherein, X i is the actual measurement value of the monitoring parameter at the i-th sampling point; Y i is the fault label of the i-th sampling point, 0 is normal and 1 is fault; X mean is the historical average value of the monitoring parameter X i ; Y mean is the historical average value of the fault label Y i ;

[0029] According to the correlation value of each parameter, the weight is adjusted:

[0030] ;

[0031] Wherein, w' is the adjusted weight, w is the initial weight, r is the correlation coefficient, and a is the adjustment factor;

[0032] According to the adjusted weight, the collection frequency of each parameter is adjusted:

[0033] ;

[0034] Wherein, f' is the adjusted collection frequency, f base is the basic collection frequency, and w' is the adjusted weight.

[0035] In an embodiment of the present application, when the load of the power distribution network changes greatly, the change of the collection frequency is controlled according to the load change rate, including:

[0036] The basic collection frequency is set, that is, the sampling frequency in the normal operation state of the power distribution network;

[0037] The relationship between the collection frequency and the load change rate is:

[0038] ;

[0039] Wherein, f is the current collection frequency, f base is the basic collection frequency, △P is the load change rate, and a is the adjustment factor, indicating the gain coefficient of the load change rate and the collection frequency.

[0040] In an embodiment of the present application, the upper limit and the lower limit of the collection frequency are set.

[0041] In an embodiment of the present application, when the power distribution network is about to fail, the priority of collection is determined according to the predicted fault type, the parameters related to the predicted fault type are taken as higher priority, and the collection frequency thereof is increased.

[0042] In an embodiment of the present application, when the power distribution network fails or abnormally, only the data related to the fault is collected, and after the fault is recovered or the abnormality is eliminated, the collection mode in the normal operation state is restored.

[0043] In an embodiment of the present application, through the intelligent monitoring unit of the distribution transformer, various operating parameters in the power distribution network are monitored in real time, including:

[0044] Current: phase current, zero sequence current, and load current of each power distribution line and device, according to the unbalance of multi-phase current, current overload, overcurrent, and load imbalance can be obtained in real time;

[0045] Voltage: monitoring the voltage quality of each node in the power distribution network, including the voltage of the transformer, the power distribution switch, and the load endpoint, real-time detection of voltage fluctuation, voltage sag, voltage too high or too low, and abnormal situation of harmonic voltage;

[0046] Load parameter: monitoring the total load of the power distribution network and the branch load of each region or device, including load change trend, load peak, load curve, and load prediction, and further judging whether there is overloading or load fluctuation problem;

[0047] Power factor: real-time monitoring of power factor, especially when the load of the power distribution network changes and the equipment operates, the deviation of the power factor can be found in time, and whether there is waste of reactive power or equipment operation efficiency problem can be inferred.

[0048] Temperature: Monitor the ambient temperature of the power distribution network equipment, monitor the working temperature of the equipment in the power distribution network, judge whether the equipment is in overload working state, and identify the fault risk of the equipment in advance;

[0049] Humidity: Monitor the ambient humidity of the power distribution equipment in the power distribution network to determine whether the equipment is in a harmful humidity environment.

[0050] To solve the above technical problems, the application also provides an adaptive data acquisition system of an intelligent distribution transformer monitoring unit, comprising:

[0051] The distribution transformer intelligent monitoring unit is used for real-time monitoring of various operating parameters in the power distribution network, including current, voltage, load, power factor, temperature and humidity;

[0052] The data classification module is used for classifying the various operating parameters monitored in real time according to the operating state of the power distribution network, and dividing them into normal monitoring data and abnormal monitoring data;

[0053] The state judgment module is used for analyzing the current state of the power distribution network based on the real-time monitoring data, and judging whether it is in a normal, load or abnormal state;

[0054] The distribution transformer monitoring and prediction module is used for constructing a distribution transformer monitoring and prediction model, identifying the operating state of the power distribution network, and predicting potential faults or abnormal conditions;

[0055] The data acquisition frequency control module is used for dynamically adjusting the frequency and type of data acquisition according to the prediction result of the distribution transformer monitoring and prediction module, including:

[0056] Low-frequency acquisition mode, when the power distribution network is in normal operation state, the low-frequency data acquisition is automatically started by the data acquisition frequency control module to collect various operating parameters in the power distribution network;

[0057] Frequency adjustment acquisition mode, when the power distribution network load changes greatly or is about to fail, the data acquisition frequency is automatically increased by the data acquisition frequency control module;

[0058] Fault emergency acquisition mode, when the power distribution network fails or abnormally, the highest frequency data acquisition mode is started immediately by the data acquisition frequency control module to collect detailed power distribution network data in real time.

[0059] The above technical solutions of the application have the following advantages compared with the prior art:

[0060] The adaptive data acquisition method and system of the intelligent distribution transformer monitoring unit provided by the application can monitor the running state of the distribution network in real time, intelligently judge the running state of the distribution network, and automatically adjust the data acquisition frequency and content according to the state of the distribution network and the load change, thereby optimizing the performance of the intelligent distribution transformer monitoring unit. BRIEF DESCRIPTION OF DRAWINGS

[0061] In order to make the content of the application more easily understood, the application will be further described in detail below according to specific embodiments of the application and in conjunction with the drawings, in which:

[0062] Figure 1 is a step flow chart of the adaptive data acquisition method of the intelligent distribution transformer monitoring unit of the application;

[0063] Figure 2 is a structural block diagram of the adaptive data acquisition system of the intelligent distribution transformer monitoring unit of the application. DETAILED DESCRIPTION

[0064] The application will be further described below in conjunction with the drawings and specific embodiments, so that those skilled in the art can better understand the application and implement it, but the embodiments are not limiting to the application. Embodiment one

[0065] Referring to Figure 1 the application discloses an adaptive data acquisition method of an intelligent distribution transformer monitoring unit, which comprises:

[0066] S1, through the intelligent distribution transformer monitoring unit, a plurality of running parameters in the distribution network are monitored in real time.

[0067] The distribution network is a very important part of the power system, and its running state directly affects the stability of the power and the power supply quality, therefore, real-time monitoring of a plurality of running parameters (such as voltage, current, power, etc.) in the distribution network is the basis for ensuring the safe and stable operation of the system.

[0068] Real-time monitoring provides immediate data support for the operation of the distribution network, can timely discover potential problems, give early warnings and take corresponding measures to ensure the stability of the power system.

[0069] However, in actual use, the intelligent distribution transformer monitoring unit needs to collect a plurality of running parameters of a plurality of components such as transmission lines and distribution equipment in the distribution network, and frequent data collection causes data redundancy, occupies the computing resources and storage space of the entire system, in order to solve this problem, the following scheme is provided.

[0070] S2, classify the monitored multiple operating parameters according to the operating state of the power distribution network, into normal monitoring data and abnormal monitoring data, analyze the current state of the power distribution network based on real-time monitoring data, and determine whether it is in a normal, load or abnormal state.

[0071] Firstly, according to the analysis of the collected data, the operating state of the power distribution network is usually dynamic, therefore, the monitoring data is classified according to the operating state, different monitoring strategies can be taken for different states, for example, in normal state, data is not needed to be collected too frequently, while in abnormal state, more intensive data collection is needed, classified data collection helps to improve the efficiency of data processing, avoid excessive redundant data collection and analysis, thereby saving computing resources and storage space, concentrating resources for in-depth analysis of abnormal data, and improving the response speed of the system.

[0072] S3, construct a distribution transformer monitoring and prediction model, which is used to identify the operating state of the power distribution network and predict potential faults or abnormal conditions, and determine the frequency and type of data collection according to the prediction results.

[0073] Secondly, by analyzing the operating state and potential fault risk of the power distribution network through the prediction model, the data collection frequency is adjusted, which helps to optimize the data collection and analysis strategy at critical moments, improves the intelligence and adaptability of the system, and reduces the frequency of data collection when the power distribution network is in normal operation, thereby reducing the occupation of computing resources and storage space of the entire system, and prolonging the service life of the distribution transformer intelligent monitoring unit.

[0074] S4-1, in the normal operating state of the power distribution network, the load and operating condition are generally stable, and high-frequency data collection is not needed, low-frequency data collection of multiple operating parameters in the power distribution network can reduce the pressure on the communication network and storage, and save system resources.

[0075] S4-2, when the load of the power distribution network changes greatly or is about to fail, the operating state of the power distribution network becomes unstable, and simple low-frequency data collection may not be able to capture these changes in time, increasing the data collection frequency can help detect small changes and discover potential faults or abnormalities, and sufficient monitoring data can be obtained in time at critical moments, enhancing the fault diagnosis and response ability of the power distribution network and reducing the probability and impact of accidents.

[0076] S4-3, when a fault or an abnormality occurs, the operation condition of the power distribution network changes sharply, a large amount of detailed operation data must be collected in the shortest time, in order to quickly diagnose the root cause of the problem and take corrective measures, therefore, the highest frequency data collection mode is started immediately, real-time detailed power distribution network data is collected, key data at the time of fault occurrence can be captured to the maximum extent, which helps to quickly locate the fault cause, improves the fault handling efficiency, shortens the repair time, and avoids larger range of outage or loss caused by fault expansion.

[0077] Based on the above analysis, the application provides a high-efficiency and flexible data collection method for the intelligent power distribution network monitoring unit, which forms a closed-loop intelligent data collection system from resource conservation, fault prediction to rapid response and system intelligence improvement, which not only improves the accuracy and efficiency of power distribution network data collection, but also provides a strong guarantee for the safe and stable operation of the power distribution network.

[0078] Specifically, through the intelligent monitoring unit of the distribution transformer, a variety of operating parameters in the power distribution network are monitored in real time, including current, voltage, load parameter, power factor, temperature and humidity. These monitoring parameters together constitute a comprehensive and real-time monitoring system for the power distribution network. Each parameter is closely related to the safe, stable and efficient operation of the power distribution network, which is specifically explained as follows:

[0079] Current: The phase current, zero sequence current and load current of each power distribution line and equipment can be obtained in real time according to the unbalance of multi-phase current. Current overload, overcurrent and load imbalance can be obtained in real time. Current imbalance, overload or overcurrent are common causes of power distribution network failure. Current overload or overcurrent can cause equipment damage, overheating or even fire. Therefore, monitoring the real-time state of current is crucial to ensure the safety of the power distribution network.

[0080] Voltage: The voltage quality of each node in the power distribution network, including the transformer, power distribution switch and load endpoint, is monitored. Voltage fluctuation, voltage sag, voltage overload or undervoltage, and abnormal harmonic voltage are detected in real time. Voltage is the basis for stable operation of the power distribution network. Voltage fluctuation, voltage sag, voltage overload or undervoltage and other abnormal conditions will directly affect the normal operation of the equipment, and may cause equipment damage or system shutdown in severe cases.

[0081] Load parameter: The total load of the power distribution network and the branch load of each region or device are monitored, including load change trend, load peak, load curve and load prediction, to determine whether there is an overload or load fluctuation problem. The load change of the power distribution network directly affects the efficiency and safety of the power grid. Overload may cause the power distribution equipment to overload, and load fluctuation may increase the instability of the system.

[0082] Power factor: real-time monitoring of power factor, especially when the power distribution network load changes, the equipment is running, it can timely find the deviation of power factor, and then infer whether there is waste of reactive power or equipment operation efficiency problem. The power factor reflects the utilization efficiency of electric energy, and low power factor means a large amount of reactive power, resulting in energy waste of equipment.

[0083] Temperature: monitoring the ambient temperature of the power distribution network equipment, monitoring the working temperature of the equipment in the power distribution network, to judge whether the equipment is in an overload working state, and identify the fault risk of the equipment in advance. High temperature may cause equipment overheating, insulation damage, and even cause fire and other serious consequences.

[0084] Humidity: monitoring the humidity of the surrounding environment of the power distribution equipment in the power distribution network, to determine whether the equipment is in a harmful humidity environment. High humidity may cause condensation on the surface of the equipment, causing corrosion, short circuit and other problems. Equipment in a high humidity environment for a long time will reduce its service life and reliability.

[0085] By monitoring current, voltage, load, power factor, temperature and humidity and other key indicators, potential problems can be detected in time and appropriate measures can be taken to improve the automation and intelligent level of the power distribution network and ensure its long-term stable and safe operation. This multi-dimensional monitoring method effectively reduces the probability of failure and improves the efficiency of the power system, and provides valuable data support for system optimization and adjustment.

[0086] Specifically, in order to realize the adaptive adjustment of the data acquisition method, the running state of the current power distribution network needs to be obtained, and for this purpose, the application constructs a distribution transformer monitoring and prediction model. The method for constructing the model comprises:

[0087] S31, obtaining historical data of the distribution transformer network, including current, voltage, load, power factor, temperature, humidity; wherein: the historical data is the basis for training the model and making predictions. By collecting long-term operation data of the power distribution network, necessary training samples can be provided for the model. Each data point represents the running state of the power distribution network at a certain time point, and contains multiple dimensions of parameters. These data as input can help the model capture the correlation between different parameters of the power distribution network, and then make accurate fault prediction.

[0088] S32, according to the historical operation data and the known power distribution network fault or abnormal information, label the data for identifying abnormal events;

[0089] Specifically, the labeled data is the core of supervised learning. By matching historical data with fault information, the "abnormal" and "normal" states in the data are marked, providing supervised training data for subsequent machine learning models.

[0090] For example, the data of a certain period of time (such as temperature, load, voltage, etc.) is associated with the actual fault event, and these data are labeled as "abnormal" tags. The labeling process helps the machine learning algorithm to identify potential abnormal patterns from a large amount of historical data, so as to make accurate judgments when facing new data.

[0091] S33, using a machine learning algorithm to classify real-time data, identifying the operating state of the power distribution network, and dividing the power distribution network into normal state or fault state;

[0092] Specifically, real-time data classification is achieved by using a trained machine learning model to predict new real-time monitoring data. Based on the rules learned from historical data, the machine learning algorithm can determine whether the current data conforms to the "normal" pattern or deviates from it, thereby identifying the "fault" state.

[0093] An initialized prediction model is constructed using a long-term memory network (LSTM, Long Short-Term Memory), and the prediction model is trained based on historical data to capture time series patterns and trends in power distribution network operation, providing early warning of potential faults, and obtaining a distribution transformer monitoring prediction model.

[0094] The operating state of the power distribution network has a time dependency, for example, load changes, current and voltage fluctuations usually have regularity in time series, LSTM can capture these time series patterns, and the LSTM-based prediction model can process and analyze the time series data of the power distribution network, identify potential faults in advance, and provide direction for data acquisition.

[0095] Further, when training the LSTM model, the historical data is usually input into the network in time series, and the model learns the relationship between different time points to predict future trends. To ensure that the model can work effectively under different power distribution network operating states and accurately predict potential faults or abnormal conditions, a cross-validation method is used during historical data training, including:

[0096] Data division: all historical data is divided into K subsets, and each time K-1 subsets are selected as the training set and the remaining 1 subset is selected as the validation set to ensure that the model is trained and validated on different data subsets, thereby obtaining a stable and reliable evaluation result;

[0097] Validation process: during each validation, record the loss and accuracy on the validation set. By calculating the loss and accuracy, the performance of the model on each validation set can be quantified, and the adaptability of the model to different data can be further analyzed;

[0098] Performance evaluation: Calculate the average of K validation results as the overall performance evaluation of the model, provide a robust performance evaluation standard, avoid performance fluctuations caused by accidental factors, calculate the average loss and average accuracy as the evaluation standard of the model.

[0099] In this embodiment, for different working states of the power distribution network, the data collection of the power distribution network is further described, including: normal operation state of the power distribution network, when the load of the power distribution network changes greatly, when the power distribution network is about to fail, and when the power distribution network is about to fail.

[0100] In the normal operation state of the power distribution network, the collection frequency of each parameter is adjusted by weighted proportion, which reasonably allocates the collection frequency according to the importance of each parameter, the load state of the equipment and the real-time running environment and other factors, ensures that the key parameters are collected frequently enough, and the other less important parameters are collected less frequently, thereby optimizing the resource utilization and improving the system response ability.

[0101] Specifically, the process of adjusting the collection frequency of each parameter by weighted proportion includes:

[0102] Define the parameters that need to be monitored in the power distribution network, which are current, voltage, load, power factor, temperature and humidity in this embodiment, and give each parameter a preliminary weight value, and the initial weight value can be evenly distributed.

[0103] According to the frequency, type and severity of the fault in the historical data of each parameter, the correlation of each parameter with the fault is calculated, and the Pearson correlation coefficient is used to measure the linear correlation between each parameter and the occurrence of the fault, assuming that the correlation of each monitoring parameter with the occurrence of the fault is r:

[0104] ;

[0105] Where X i is the actual measurement value of the monitoring parameter at the i-th sampling point; Y i is the fault label of the i-th sampling point, 0 is normal and 1 is fault; X mean is the historical average value of the monitoring parameter X i ; Y mean is the historical average value of the fault label Y i ;

[0106] According to the correlation value of each parameter, adjust the weight:

[0107] ;

[0108] Where w' is the adjusted weight, w is the initial weight, r is the correlation coefficient, and a is the adjustment factor.

[0109] According to the adjusted weight, the collection frequency of each parameter is adjusted:

[0110] ;

[0111] Wherein, f' is the adjusted collection frequency, f base is the basic collection frequency, and w' is the adjusted weight.

[0112] In this embodiment, the collection frequency is dynamically adjusted according to the importance of the parameters, which can ensure high-frequency collection of key parameters (such as current and voltage), and ensure that relevant data can be captured and analyzed in time before and after the device fails. By adjusting the collection frequency through weighting, the collection of irrelevant data is reduced, the pressure on network bandwidth and storage resources is reduced, and the system burden is reduced.

[0113] Specifically, in the power distribution network, load change is a key influencing factor. The fluctuation of load not only may affect the operation state of the device, but also may indicate potential failure or system instability. Therefore, when the load change of the power distribution network is large, the change of the collection frequency is controlled according to the load change rate, which can improve the collection accuracy and ensure the timely collection of key data, thereby optimizing the resource utilization of the system and reflecting the possible failure or abnormal situation in real time.

[0114] Specifically, the process of controlling the change of the collection frequency according to the load change rate includes:

[0115] Set the basic collection frequency, that is, the sampling frequency in the normal operation state of the power distribution network;

[0116] The relationship between the collection frequency and the load change rate is:

[0117] ;

[0118] Wherein, f is the current collection frequency, f base is the basic collection frequency, and △P is the load change rate. Alpha is an adjustment factor, which represents the gain coefficient of the load change rate and the collection frequency.

[0119] When the load change rate is small, the system will reduce unnecessary data collection and reduce the frequency. At this time, the collection frequency of key parameters such as current and voltage can be maintained at a low level;

[0120] When the load change rate is large, the system will automatically increase the collection frequency through the preset rules to ensure the capture of possible failures, overloads or voltage abnormalities and other events.

[0121] Specifically, while setting the acquisition frequency to change with the load rate, the upper and lower limits of the acquisition frequency are also set, wherein the lower limit is the basic acquisition frequency when the load rate is zero, and the upper limit cannot exceed the maximum acquisition frequency that the system can provide.

[0122] Specifically, when the power distribution network is about to fail, the priority of acquisition is determined according to the predicted failure type, the parameters related to the predicted failure type are taken as higher priority, and the acquisition frequency is increased, so that the data related to the failure can be captured more accurately and efficiently, and the system can respond quickly and take necessary repair measures when the failure occurs.

[0123] For example, when the system predicts that an overload failure may occur, the acquisition frequency of the current sensor and the load monitoring device should be increased to ensure that current fluctuations can be monitored in a timely manner.

[0124] And when the predicted failure is about to occur, not only the acquisition frequency of the parameters related to the failure is increased, but also the acquisition of other parameters unrelated to the failure is ensured.

[0125] Specifically, when the power distribution network fails or abnormally, only the data related to the failure is collected, the acquisition frequency of these key parameters is increased, and the collection of non-key parameters is suspended, so that the system can respond quickly, potential problems can be found in advance, the failure response time can be reduced, the pressure on network bandwidth and storage resources can be reduced, unnecessary data redundancy collection can be avoided, and the utilization of system resources can be optimized.

[0126] After the failure is recovered or the abnormality is removed, the acquisition mode in the normal operation state is restored, and normal monitoring of all related data is restored. Embodiment two

[0127] Referring to Figure 2 Based on the above embodiment one, in order to realize the above method, the application further discloses an adaptive data acquisition system of an intelligent distribution transformer monitoring unit, comprising:

[0128] The distribution transformer intelligent monitoring unit is used for monitoring a plurality of operating parameters in the power distribution network in real time, including current, voltage, load, power factor, temperature, and humidity.

[0129] The data classification module is used for classifying the plurality of operating parameters monitored in real time according to the operating state of the power distribution network, and dividing them into conventional monitoring data and abnormal monitoring data.

[0130] The state judgment module is used for analyzing the current state of the power distribution network based on the real-time monitoring data, and judging whether it is in a normal, load, or abnormal state.

[0131] The power distribution monitoring and prediction module is configured to construct a power distribution monitoring and prediction model, identify the operating state of the power distribution network, and predict potential faults or abnormal conditions.

[0132] The data acquisition frequency control module is configured to dynamically adjust the frequency and type of data acquisition according to the prediction result of the power distribution monitoring and prediction module, including:

[0133] The low-frequency acquisition mode is configured to automatically enable low-frequency data acquisition to collect various operating parameters in the power distribution network when the power distribution network is in a normal operating state through the data acquisition frequency control module.

[0134] The frequency-adjusted acquisition mode is configured to automatically increase the data acquisition frequency when the load of the power distribution network changes greatly or is about to fail through the data acquisition frequency control module.

[0135] The fault emergency acquisition mode is configured to immediately start the highest-frequency data acquisition mode to collect detailed power distribution network data in real time when the power distribution network fails or is abnormal through the data acquisition frequency control module.

[0136] Those skilled in the art will understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage, etc.) containing computer-usable program code.

[0137] The present application is described with reference to flowcharts and / or block diagrams according to the methods, devices (systems), and computer program products of the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as combinations of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing apparatus to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing apparatus produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in one or more flows and / or blocks. Figure 1 The functions specified in one or more flows and / or blocks.

[0138] These computer program instructions can also be stored in a computer-readable memory that can direct the computer or other programmable data processing apparatus to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including instruction devices that implement the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in one or more flows and / or blocks. Figure 1the function specified in the one or more blocks.

[0139] These computer program instructions can also be loaded into computer or other programmable data processing devices, so that a series of operation steps are performed on the computer or other programmable data processing devices to generate computer-implemented processes, so that the instructions executed on the computer or other programmable devices provide processes for implementing the flowcharts Figure 1 the flowcharts or the flowcharts and / or blocks Figure 1 the steps of the function specified in the one or more blocks.

[0140] Obviously, the above embodiments are only examples for the purpose of clarity, and are not limiting to the embodiments. Based on the above description, other different forms of changes or variations can also be made by those of ordinary skill in the art. Here, it is not necessary and impossible to enumerate all the embodiments. The obvious changes or variations derived therefrom are still within the protection scope of the present application.

Claims

1. An adaptive data acquisition method for intelligent power transformer monitoring unit, characterized in that: Comprise: Through the intelligent monitoring unit of distribution transformer, real-time monitoring of a variety of operating parameters in the distribution network; According to the running state of the distribution network, the monitored operating parameters are classified into normal monitoring data and abnormal monitoring data, and the current state of the distribution network is analyzed based on the real-time monitoring data to determine whether it is in a normal, load or abnormal state; A distribution transformer monitoring and prediction model is constructed to identify the operating state of the distribution network and predict potential faults or abnormal conditions, and the frequency and type of data collection are determined according to the prediction results; In the normal operating state of the distribution network, low-frequency data collection is used to collect a variety of operating parameters in the distribution network, and the collection frequency of each parameter is adjusted by weighted allocation, including: Define the parameters to be monitored in the distribution network, and give each parameter a preliminary weight value; According to the frequency, type and severity of each parameter in the historical data and the occurrence of faults, the correlation of each parameter with the occurrence of faults is calculated, and the Pearson correlation coefficient is used to measure the linear correlation between each parameter and the occurrence of faults, assuming that the correlation of each monitoring parameter with the occurrence of faults is r: ； Wherein, X i is the actual measured value of the monitoring parameter of the i th sampling point; Y i is the fault label of the i th sampling point, 0 is normal, and 1 is fault; X mean is the historical average value of the monitoring parameter X i ; Y mean is the historical average value of the fault label Y i ; According to the correlation value of each parameter, the weight is adjusted: ； Where w' is the adjusted weight, w is the initial weight, r is the correlation coefficient, and a is the adjustment factor; According to the adjusted weight, the collection frequency of each parameter is adjusted: ； where f' is the adjusted acquisition frequency, f base is the base acquisition frequency, and w' is the adjusted weight. When the load of the distribution network changes greatly or is about to fail, the data collection frequency is increased; When a fault or anomaly occurs, the highest frequency data collection mode is started immediately to collect detailed distribution network data in real time.

2. The adaptive data acquisition method of the smart distribution monitoring unit according to claim 1, wherein: A distribution transformer monitoring and prediction model is constructed, including: Obtain the historical data of the distribution transformer network, including current, voltage, load, power factor, temperature, humidity; According to the historical operating data and known distribution network fault or abnormal information, the data is labeled to identify abnormal events; Use machine learning algorithms to classify real-time data, identify the operating state of the distribution network, and divide the distribution network into normal or fault states; Use long-term memory network to build an initialized prediction model, train the prediction model based on historical data, capture the timing patterns and trends in the operation of the distribution network, and give early warning of potential faults to obtain the distribution transformer monitoring and prediction model.

3. The adaptive data acquisition method of the smart distribution monitoring unit according to claim 2, wherein: Training the prediction model based on historical data includes: Divide all historical data into K subsets, and select K-1 subsets as the training set and the remaining 1 subset as the validation set each time; Record the loss and accuracy on the validation set during each validation; Calculate the average of K validation results as the overall performance evaluation of the model, and calculate the average loss and average accuracy as the evaluation standard of the model.

4. The adaptive data collection method of the intelligent distribution transformer monitoring unit according to claim 1, characterized in that: When the load of the distribution network changes greatly, the change of the collection frequency is controlled according to the load change rate, including: Set the basic collection frequency, which is the sampling frequency in the normal operating state of the distribution network; The relationship between the collection frequency and the load change rate is: ； Wherein, f is the current acquisition frequency, f base is the basic acquisition frequency, △P is the load change rate, and α is the adjustment factor, indicating the gain coefficient of the load change rate and the acquisition frequency.

5. The adaptive data acquisition method of the smart distribution monitoring unit according to claim 4, wherein: Set the upper and lower limits of the collection frequency.

6. The adaptive data acquisition method of the smart distribution monitoring unit according to claim 1, wherein: When the power distribution network is about to fail, the priority of the collected data is determined according to the predicted failure type, and the parameters related to the predicted failure type are taken as higher priority to improve the collection frequency.

7. The adaptive data acquisition method of the smart distribution monitoring unit according to claim 1, wherein: When the power distribution network fails or abnormally, only the data related to the failure is collected, and after the failure is recovered or the abnormality is removed, the collection mode under normal operation is restored.

8. The adaptive data acquisition method of the smart distribution monitoring unit according to claim 1, wherein: Through the intelligent monitoring unit of the distribution transformer, various operating parameters in the power distribution network are monitored in real time, including: Current: phase current, zero sequence current, load current of each power distribution line and equipment, according to the imbalance of multi-phase current, the current overload, overcurrent, load imbalance can be obtained in real time; Voltage: monitoring each node in the power distribution network, including: transformer, power distribution switch, voltage quality of load endpoint, real-time detection of voltage fluctuation, voltage sag, voltage too high or too low, harmonic voltage abnormality; Load parameter: monitoring the total load of the power distribution network and the branch load of each region or equipment, including load change trend, load peak, load curve and load prediction, to determine whether there is an overload or load fluctuation problem; Power factor: real-time monitoring of power factor, especially when the power distribution network load changes and the equipment is running, the deviation of power factor can be found in time, and whether there is a problem of waste of reactive power or equipment operation efficiency can be inferred; Temperature: monitoring the ambient temperature of the power distribution equipment and the working temperature of the equipment in the power distribution network to determine whether the equipment is in an overload working state and to identify the fault risk of the equipment in advance; Humidity: monitoring the humidity of the surrounding environment of the power distribution equipment in the power distribution network to determine whether the equipment is in a harmful humidity environment.

9. An adaptive data acquisition system of an intelligent power transformer monitoring unit for implementing the method according to any one of claims 1 to 8, characterized in that: It includes: Intelligent monitoring unit of distribution transformer, for real-time monitoring of various operating parameters in the power distribution network, including current, voltage, load, power factor, temperature, humidity; Data classification module, for classifying the real-time monitoring data according to the operating state of the power distribution network, dividing it into normal monitoring data and abnormal monitoring data; State judgment module, for analyzing the current state of the power distribution network based on real-time monitoring data to determine whether it is in normal, load or abnormal state; Distribution transformer monitoring and prediction module, for building a distribution transformer monitoring and prediction model to identify the operating state of the power distribution network and predict potential failures or abnormal conditions; Data acquisition frequency control module, for dynamically adjusting the frequency and type of data acquisition according to the prediction results of the distribution transformer monitoring and prediction module, including: Low-frequency acquisition mode, in the normal operating state of the power distribution network, the low-frequency data acquisition is automatically enabled by the data acquisition frequency control module to collect various operating parameters in the power distribution network; Frequency adjustment acquisition mode, when the power distribution network load changes greatly or is about to fail, the data acquisition frequency is automatically increased by the data acquisition frequency control module; Fault emergency acquisition mode, when the power distribution network fails or abnormally, the highest frequency data acquisition mode is immediately started by the data acquisition frequency control module to collect detailed power distribution network data in real time.

Images