An AI-based power equipment health management system and method

Abstract

This invention discloses an artificial intelligence-based power equipment health management system, comprising: an operation data acquisition module for acquiring the current operating temperature and insulation resistance of the power equipment, and acquiring historical operating temperature sequences and historical insulation resistance sequences at multiple historical moments within a preset historical time range; an alternating influence analysis module for analyzing the operating temperature and insulation resistance in conjunction with the current power operation parameters of the power equipment; an abnormal fluctuation analysis module for analyzing and obtaining a historical abnormal superposition sequence based on the historical operating temperature sequence and historical insulation resistance sequence, and obtaining a compensation abnormal superposition number; and an equipment monitoring and management module for generating health parameters of the power equipment based on the compensation abnormal superposition number, and performing health maintenance management of the power equipment. This invention solves the shortcomings of existing technologies, which are difficult to automatically analyze power equipment operation data for health status assessment and fault prediction, resulting in low maintenance efficiency and reliability.

Description

Technical Field

[0001] This invention relates to the field of power equipment maintenance technology, and more specifically to a power equipment health management system and method based on artificial intelligence. Background Technology

[0002] In power system operation, the health status of electrical equipment is crucial to power supply reliability. Traditional power equipment maintenance mainly relies on periodic inspections and post-fault repairs. This approach has many limitations, such as the difficulty in obtaining accurate real-time operating status of equipment and the inability to predict potential faults in advance. For key indicators of power equipment, such as insulation performance and temperature, existing technologies often can only monitor them in isolation, failing to fully consider the alternating effects between them. In terms of data analysis, there is a lack of effective automated means to deeply mine the potential information in operational data, making it difficult to achieve accurate assessment of equipment health status and accurate prediction of faults.

[0003] Existing technologies suffer from technical problems such as difficulty in automatically analyzing power equipment operating data for health status assessment and fault prediction, resulting in low maintenance efficiency and reliability. Summary of the Invention

[0004] To address the aforementioned problems, this invention proposes an artificial intelligence-based power equipment health management system, which solves the shortcomings of existing technologies, such as difficulty in automatically analyzing power equipment operating data for health status assessment and fault prediction, resulting in low maintenance efficiency and reliability.

[0005] The technical solution adopted in this invention is as follows:

[0006] An artificial intelligence-based power equipment health management system includes:

[0007] A data acquisition module is provided, which is used to acquire the current operating temperature and insulation resistance of the power equipment, and to acquire the historical operating temperature sequence and historical insulation resistance sequence of multiple historical moments within a preset historical time range;

[0008] An alternating influence analysis module is used to perform an alternating superposition influence analysis of operating temperature and insulation resistance based on the operating temperature and insulation resistance, combined with the current power operating parameters of the power equipment, to obtain the number of abnormal superpositions.

[0009] An abnormal fluctuation analysis module is used to analyze and obtain a historical abnormal superposition number sequence based on the historical operating temperature sequence and historical insulation resistance sequence, calculate the fluctuation amplitude of the abnormal superposition number, compensate for the abnormal superposition number, and obtain the compensated abnormal superposition number; and

[0010] A device monitoring and management module is provided, which is used to generate health parameters of the power equipment based on the number of compensation anomaly superpositions, and to perform health maintenance management on the power equipment.

[0011] Optionally, the power equipment includes a transformer.

[0012] Optionally, the alternating influence analysis module is also used to train influence analysis channels, wherein the influence analysis channels include a temperature influence analysis path and a resistance influence analysis path, the temperature influence analysis path includes a temperature influence analysis branch that integrates the number of influence analyses, and the resistance influence analysis path includes a resistance influence analysis branch that integrates the number of influence analyses.

[0013] Optionally, the alternating influence analysis module performs an integrated influence analysis of the insulation resistance based on the operating temperature and the power operating parameters to obtain a first influencing insulation resistance change value; the insulation resistance and the first influencing insulation resistance change value are superimposed to obtain the first influencing insulation resistance.

[0014] Optionally, the alternating influence analysis module performs an integrated influence analysis of the operating temperature based on the insulation resistance and the power operating parameters to obtain a first influence operating temperature change value; the operating temperature and the first influence operating temperature change value are superimposed to obtain the first influence operating temperature.

[0015] Optionally, the alternating influence analysis module continues to perform the influence analysis of alternating superposition of operating temperature and insulation resistance until the influence on operating temperature is greater than or equal to the operating temperature threshold or the influence on insulation resistance is less than or equal to the insulation resistance threshold. The influence analysis then converges, and the number of rounds of alternating superposition influence analysis is output to obtain the number of abnormal superpositions.

[0016] Optionally, the alternating influence analysis module is also used to collect a set of sample operating temperatures, a set of sample power operating parameters, and a set of sample influence insulation resistance change values ​​based on the operating data of similar power equipment over a historical period, as training data for temperature influence.

[0017] Optionally, based on the operating data of similar power equipment over a historical period, a set of sample insulation resistance, a set of sample power operating parameters, and a set of sample values ​​affecting operating temperature changes can be collected as training data on the influence of insulation resistance.

[0018] Optionally, the temperature effect training data and insulation resistance effect training data are divided according to the number of integrated effect analyses, and the temperature effect analysis branch and insulation resistance effect analysis branch of the number of integrated effect analyses are trained respectively based on the multiple sets of training data of the divided number of integrated effect analyses.

[0019] Optionally, the temperature influence analysis branch and the resistance influence analysis branch of the integrated influence analysis completed by training can be combined to obtain the temperature influence analysis path and the resistance influence analysis path.

[0020] Optionally, the abnormal fluctuation analysis module is used to randomly select multiple sets of historical abnormal overlaps within the historical abnormal overlap sequence, calculate the deviation percentage of each set of historical abnormal overlaps, and obtain multiple deviation percentages. Each set of historical abnormal overlaps includes two historical abnormal overlaps. The ratio of the difference between the two historical abnormal overlaps to the larger historical abnormal overlap is calculated as the deviation percentage. The mean of the multiple deviation percentages is calculated to obtain the fluctuation amplitude of the abnormal overlap. The difference between 1 and the fluctuation amplitude of the abnormal overlap is used as a compensation coefficient, which is multiplied by the abnormal overlap to obtain the compensated abnormal overlap.

[0021] Optionally, the equipment monitoring and management module is used to obtain the threshold number of normal abnormal superpositions of normal power equipment; and calculate the ratio of the number of compensated abnormal superpositions to the threshold number of normal abnormal superpositions to obtain the health parameters of the power equipment; when the health parameters are less than the health parameter threshold, the power equipment is subjected to health maintenance management.

[0022] Beneficial effects

[0023] 1. This invention collects the current operating temperature and insulation resistance of power equipment, and collects historical operating temperature sequences and historical insulation resistance sequences from multiple historical moments within a preset historical time range; it performs an impact analysis on the alternating superposition of operating temperature and insulation resistance to obtain the number of abnormal superpositions; based on the historical operating temperature sequence and historical insulation resistance sequence, it analyzes to obtain a historical abnormal superposition frequency sequence, calculates the fluctuation amplitude of the abnormal superposition frequency, compensates for the abnormal superposition frequency, and obtains the compensated abnormal superposition frequency; and performs health maintenance management on the power equipment. This achieves the technical effect of assessing the health status of equipment and predicting faults, thereby improving the efficiency and reliability of power equipment maintenance. Attached Figure Description

[0024] Figure 1 This is a schematic diagram of the structure of the artificial intelligence-based power equipment health management system provided in Embodiment 1 of the present invention;

[0025] Figure 2 This is a flowchart illustrating the artificial intelligence-based power equipment health management method provided in Embodiment 2 of the present invention.

[0026] Figure labeling: 10 for running data acquisition module, 20 for alternating influence analysis module, 30 for abnormal fluctuation analysis module, and 40 for equipment monitoring and management module. Detailed Implementation

[0027] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0028] In this invention, any step can be stored as a computer instruction or program in an unrestricted computer memory and can be called and recognized by an unrestricted computer processor, without any additional restrictions.

[0029] Example 1

[0030] The technical solution adopted in this invention is as follows:

[0031] like Figure 1 As shown, an artificial intelligence-based power equipment health management system includes: an operation data acquisition module 10, an alternating influence analysis module 20, an abnormal fluctuation analysis module 30, and an equipment monitoring and management module 40. The operation data acquisition module is used to collect the current operating temperature and insulation resistance of the power equipment, and to collect historical operating temperature sequences and historical insulation resistance sequences for multiple historical moments within a preset historical time range.

[0032] The alternating influence analysis module is used to perform an alternating superposition influence analysis of operating temperature and insulation resistance based on the operating temperature and insulation resistance, combined with the current power operating parameters of the power equipment, to obtain the number of abnormal superpositions.

[0033] The abnormal fluctuation analysis module is used to analyze and obtain the historical abnormal superposition number sequence based on the historical operating temperature sequence and historical insulation resistance sequence, calculate the fluctuation amplitude of the abnormal superposition number, compensate for the abnormal superposition number, and obtain the compensated abnormal superposition number.

[0034] The equipment monitoring and management module is used to generate health parameters of the power equipment based on the number of compensation anomaly superpositions, and to perform health maintenance management on the power equipment.

[0035] In this embodiment, the power equipment includes a transformer.

[0036] Specifically, the core function of the operation data acquisition module is to collect data from the transformer, a type of power equipment. It first acquires two key parameters: the transformer's current operating temperature and insulation resistance. This provides real-time monitoring of the equipment's current operating status. Furthermore, considering the interrelationship between transformer operating temperature and insulation resistance—that is, excessively high operating temperature leads to decreased insulation resistance, which in turn causes current leakage and further increases operating temperature—the operation data acquisition module also collects relevant data from multiple historical moments within a preset historical time range to gain a more comprehensive and in-depth understanding of this dynamic process. Within this historical time range, multiple historical moments are set at certain time intervals, and the corresponding operating temperature and insulation resistance are acquired for each historical moment, thus forming historical operating temperature sequences and historical insulation resistance sequences. This historical data is crucial for subsequent analysis of the transformer's operating status and health condition.

[0037] Specifically, the alternating influence analysis module is mainly responsible for performing complex alternating superposition influence analysis on the transformer's operating temperature and insulation resistance. First, it acquires the current operating temperature and insulation resistance data provided by the operating data acquisition module, and combines this with the current power operating parameters of the electrical equipment. Next, it performs an integrated influence analysis on insulation resistance based on the operating temperature. Specifically, it predicts the change in insulation resistance based on the current operating temperature (since insulation resistance usually decreases with increasing temperature, this is a decreasing value), and then superimposes this change value onto the current insulation resistance (actually, it subtracts the change value because the resistance decreases), obtaining a new insulation resistance value. The rate of decrease in the electrical equipment's insulation resistance mainly depends on the thermal stability of the insulation material, the equipment's heat dissipation efficiency, and the magnitude of the load current, among other actual physical conditions. For example, based on extensive experimental data, for this type of transformer, within a certain range of power operating parameters, for every 1°C increase in temperature, the insulation resistance decreases by approximately k1 (k1 is a coefficient determined based on the actual equipment characteristics, in MΩ / °C). By taking into account the difference between the current operating temperature and the standard temperature, combined with k1 and the influence factors of power operating parameters (such as the correction factors for the temperature coefficient of insulation resistance of load current and voltage), the change in insulation resistance is calculated. This calculation method fully considers the influence of actual physical conditions on insulation resistance.

[0038] Next, the change in operating temperature is predicted based on this new insulation resistance value (since a decrease in insulation resistance may lead to increased current leakage, the temperature will rise, hence the rising temperature), and this is added to the current operating temperature. This process is repeated, performing alternating influence analysis, until one of the parameters (operating temperature or insulation resistance) reaches the fault threshold (in actual operation, this threshold is the minimum safe threshold for oil-immersed transformers; reaching this threshold will cause tripping). Finally, the number of alternating influence cycles is obtained. This number reflects the degree of mutual influence between temperature and resistance. The smaller the number of cycles, the shorter the fault threshold is reached in the alternation process, indicating that the higher the current operating temperature or the lower the insulation resistance, the worse the transformer's health condition. The entire process is an integrated analysis based on the magnitude of operating temperature and insulation resistance. For example, suppose there is an operating transformer. The data collected by the operating data acquisition module shows that its current operating temperature is 60°C, the measured insulation resistance of the high-voltage winding to the casing is 300MΩ, and the current power operating parameters of the electrical equipment are also acquired, such as a load current of 50A and a voltage of 8kV. Subsequently, the alternating influence analysis module 20 starts operating. First, based on the collected operating temperature of 60°C, an integrated impact analysis of the insulation resistance was conducted. Based on long-term monitoring data and established characteristic studies of this transformer model, it is known that for every 5°C increase in temperature, the insulation resistance decreases by approximately 50 MΩ (this is only an example of changes within a certain temperature range and does not represent specific actual operating conditions). The current operating temperature of 60°C represents a 30°C increase compared to the standard operating temperature of 30°C. Based on this, the estimated change in insulation resistance is -300 MΩ. This change is then superimposed on the existing 300 MΩ insulation resistance. In practical engineering applications, considering the physical meaning of insulation resistance, when the calculated result approaches a small positive value, the insulation performance has severely deteriorated. The small positive value is referred to as an extremely low value below. In actual work, the lower limit of insulation resistance can be set as the minimum safety threshold of the transformer. The newly obtained insulation resistance value is an extremely low value. In actual work, the actual current of the power equipment, the change in insulation resistance, and the heat dissipation coefficient are combined. Here, the heat dissipation coefficient is a parameter that reflects the heat dissipation capacity of the equipment and is related to the equipment structure, heat dissipation method, etc., to determine the change in operating temperature. There is a non-linear relationship between insulation resistance and temperature.

[0039] For example:

[0040] in R 0 is the reference temperature T Resistance at 0 kThis is the material coefficient. As temperature rises, the rate of decrease in resistance gradually slows, eventually reaching a stable value. For example, the minimum safety threshold of a typical oil-immersed transformer is usually very low, but it won't reach zero. This is just for ease of discussion; the approximate simulation given here is to obtain the number of alternating influence cycles and does not represent the actual operating conditions. Because the insulation resistance approaches an extremely low value, current leakage is significantly aggravated. According to Joule's law (Q=I²RT, where Q is heat, I is current, R is resistance, and T is time), in this situation, heat will accumulate rapidly, leading to a sharp rise in operating temperature. In actual calculations, the change in operating temperature is determined by combining the actual current of the power equipment, the change in insulation resistance, and the heat dissipation coefficient (a parameter reflecting the equipment's heat dissipation capacity, related to the equipment structure, heat dissipation method, etc.). For example, under current power operating parameters, for every reduction in insulation resistance by R1 (R1 is the change in insulation resistance in MΩ), the operating temperature rises by approximately k2 (k2 is a coefficient determined based on the actual equipment, in °C / (MΩ・t)) within time t. In actual operation, due to the influence of dynamic factors such as equipment heat dissipation and load changes, this module considers the changes of these dynamic factors in real time during the simulation of alternating effects. Assuming that under the current operating conditions, the operating temperature can rise by 15°C every half minute, when this is added to the initial operating temperature of 60°C, it will quickly exceed the preset operating temperature threshold of 90°C. Through the above alternating effect analysis process, the entire process only goes through one round, and the number of abnormal superpositions obtained is 1.

[0041] That is, in one embodiment, the change in insulation resistance caused by temperature change can be calculated using the following formula:

[0042]

[0043] This is the temperature coefficient (unit: MΩ / °C), which needs to be determined based on the characteristics of the equipment.

[0044] The current operating temperature. This is the standard temperature, which is usually taken as 20℃.

[0045] The specific steps of alternating effects analysis are as follows:

[0046] 1) Based on the current temperature Calculate the change in insulation resistance .

[0047] 2) Based on the updated and load current Temperature rise calculated using a thermodynamic model Update temperature .

[0048] 3) When Exceeding the safety threshold of an oil-immersed transformer (e.g., 110°C) or When the value falls below the minimum safety value (e.g., 1 MΩ), a trip is triggered.

[0049] In this thermodynamic equilibrium-based temperature rise model, the temperature change mainly depends on the current heating effect (load current) and heat dissipation efficiency:

[0050]

[0051] in: For load current, Equivalent resistance For time; For the specific heat capacity of the material, For quality; This is the heat dissipation efficiency coefficient, which is related to the heat dissipation structure of the equipment. The ambient temperature.

[0052] The specific steps of the alternating effects analysis in another embodiment are as follows:

[0053] 1) Initialize temperature ;

[0054] 2) Calculate the current resistance , For temperature coefficient, This refers to the temperature of the heat source.

[0055] 3) Update heat generation ;

[0056] 4) Solve the heat balance equation , obtain new temperature ;in The heat transfer coefficient is... For ambient temperature, For heat transfer area, This refers to the temperature of the heat source.

[0057] 5) Iterate to Among them, the allowable error It is a small positive number representing the allowable error range. This embodiment's scheme is designed to avoid the assumption of constant resistance and converge to a reasonable steady-state temperature.

[0058] It should be noted that this embodiment ultimately obtains the number of alternating influences. This number of influences reflects the degree of mutual influence between temperature and resistance. The smaller the number of influences, the shorter the fault threshold is reached in the shorter alternation process. This means that the higher the current operating temperature or the lower the insulation resistance, the worse the health condition of the transformer. The entire process is based on an integrated analysis of the operating temperature and insulation resistance.

[0059] This clearly demonstrates that, under the current initial conditions of a transformer operating at 60°C and with an insulation resistance of 300 MΩ, its health condition is far from satisfactory, necessitating timely maintenance to effectively mitigate potential fault risks. This rigorous alternating superposition analysis process accurately and intuitively reveals the intrinsic mechanism of the interaction between the transformer's internal operating temperature and insulation resistance, providing a strong basis for refined health management of power equipment.

[0060] Specifically, the abnormal fluctuation analysis module mainly conducts in-depth analysis of historical operating temperature sequences and historical insulation resistance sequences. First, for each set of data in the historical operating temperature and insulation resistance sequences—that is, the operating temperature and insulation resistance corresponding to each historical moment—an alternating superposition influence analysis is performed. This process is similar to the analysis of the current operating temperature and insulation resistance. In this way, the number of historical abnormal superpositions corresponding to each set of data can be obtained, thus forming a historical abnormal superposition sequence. Next, to more accurately grasp the fluctuation characteristics of these historical abnormal superpositions, the fluctuation amplitude of the abnormal superpositions is calculated. Specifically, multiple sets of historical abnormal superpositions are randomly selected from the historical abnormal superposition sequence. For each selected set of historical abnormal superpositions (each set contains two historical abnormal superpositions), the ratio of their difference to the larger historical abnormal superposition is calculated, and this is taken as the deviation percentage. Then, the mean of the deviation percentages of all selected sets is calculated; this mean is the fluctuation amplitude of the abnormal superposition. Finally, the current number of abnormal superpositions is compensated and adjusted based on the calculated fluctuation amplitude of the abnormal superposition count. A compensation coefficient is obtained by subtracting the fluctuation amplitude of the abnormal superposition count from 1, and then multiplying this compensation coefficient by the number of abnormal superpositions to obtain the compensated abnormal doping count. This compensation method allows the data to better reflect the actual operating status of the equipment, providing a more accurate basis for subsequent assessments of the equipment's health status.

[0061] Specifically, firstly, the threshold number of normal abnormal superpositions for normal power equipment is obtained. Then, the ratio of the number of compensated abnormal superpositions to this normal abnormal superposition threshold is calculated, and this ratio is used as the health parameter of the power equipment. This health parameter can intuitively reflect the current health status of the power equipment. When the health parameter is less than the health parameter threshold, it means that there is a health problem with the power equipment. At this time, health maintenance management of the power equipment is carried out. This includes further testing, diagnosis, and a series of measures such as deciding whether planned shutdown maintenance is needed based on specific circumstances, to ensure the normal operation and safety of the power equipment.

[0062] In this embodiment, the runtime data acquisition module is further configured to:

[0063] The system collects the current operating temperature of the power equipment and the insulation resistance between the high-voltage winding and the outer casing. Multiple historical time points are set within a preset historical time range, where the time interval between any two adjacent historical time points is the same. The system acquires the historical operating temperature and historical insulation resistance data for these multiple historical time points to obtain a historical operating temperature sequence and a historical insulation resistance sequence.

[0064] Specifically, firstly, high-precision temperature sensors are used to collect the current operating temperature of the power equipment. These sensors are installed in key parts of the power equipment, enabling them to sense the heat generated during operation in real time and convert the temperature signal into an electrical signal for subsequent processing and analysis. For collecting the insulation resistance between the high-voltage winding and the casing of the power equipment, specialized insulation resistance testing instruments are required. During testing, the instrument applies a specific voltage between the high-voltage winding and the casing, then measures the current flowing through it, and calculates the insulation resistance value according to Ohm's law. This measurement method accurately reflects the insulation performance of the internal insulation materials of the power equipment, providing an important basis for assessing the health status of the equipment.

[0065] Determining a suitable historical time range requires comprehensive consideration of factors such as the operating characteristics of the power equipment, potential failure cycles, and data analysis needs. Within this selected time range, multiple historical moments should be set. To ensure the regularity and comparability of data collection, the time interval between any two adjacent historical moments should be the same. For example, if the selected time range is one day, setting 24 historical moments, with one moment per hour, results in a one-hour time interval between any two adjacent moments. This uniform time interval arrangement ensures that the collected data presents an orderly arrangement over time, facilitating subsequent data analysis and processing, and thus enabling a better understanding of the operating status and trends of the power equipment at different points in time.

[0066] After setting the historical time points, it is necessary to acquire corresponding historical operating temperature and insulation resistance data for each historical time point. For each historical time point, the operating temperature value of the power equipment and the insulation resistance value of the high-voltage winding to the casing are accurately acquired using appropriate sensors and measuring instruments. Arranging the operating temperature values ​​acquired at each historical time point sequentially forms a historical operating temperature sequence, which reflects the temperature changes of the power equipment at different historical times. Similarly, arranging the insulation resistance values ​​acquired at each historical time point sequentially yields a historical insulation resistance sequence, which shows the trend of insulation resistance changes of the power equipment at different historical times. These historical operating temperature and insulation resistance sequences provide an important data foundation for subsequent in-depth analysis of the operating status and health condition of the power equipment.

[0067] In this embodiment, the alternating influence analysis module is further used for:

[0068] The training process includes an impact analysis channel, comprising a temperature impact analysis path and a resistance impact analysis path. The temperature impact analysis path includes a temperature impact analysis branch integrating the number of impact analyses, and the resistance impact analysis path includes a resistance impact analysis branch integrating the number of impact analyses. The system collects the current power operating parameters of the power equipment. Based on the operating temperature, it performs an integrated impact analysis of the insulation resistance based on the operating temperature and power operating parameters to obtain a first impact on the insulation resistance change value. The insulation resistance and the first impact on the insulation resistance change value are then superimposed to obtain a first impact on the insulation resistance. Based on the insulation resistance, it performs an integrated impact analysis of the operating temperature based on the insulation resistance and power operating parameters to obtain a first impact on the operating temperature change value. The operating temperature and the first impact on the operating temperature change value are then superimposed to obtain a first impact on the operating temperature. The system continues to perform alternating superposition impact analysis of the operating temperature and insulation resistance until the impact on the operating temperature is greater than or equal to the operating temperature threshold or the impact on the insulation resistance is less than or equal to the insulation resistance threshold. At this point, the impact analysis converges, and the number of rounds of alternating superposition impact analysis is output to obtain the number of abnormal superpositions.

[0069] Specifically, the training influence analysis channel is a crucial foundation for the entire system's alternating superposition influence analysis of operating temperature and insulation resistance of power equipment. This influence analysis channel consists of two key paths: a temperature influence analysis path and a resistance influence analysis path. The temperature influence analysis path includes a number of integrated influence analysis branches. These branches are units that refine the analysis of temperature effects, each with its own specific analysis mechanism and parameters, aiming to capture the impact of temperature on power equipment operation from different perspectives. Through multiple such branches, a more comprehensive and in-depth understanding of the mechanism by which temperature factors act in the operation of power equipment can be achieved. Similarly, the resistance influence analysis path includes a number of integrated influence analysis branches. These branches focus on analyzing the impact of resistance on power equipment operation. Similar to the temperature influence analysis branches, each resistance influence analysis branch has its unique analysis methods and parameter settings to accurately assess how resistance changes affect the operating state of the power equipment. This combination of multiple branches in the temperature and resistance influence analysis paths constitutes the influence analysis channel, providing an effective analytical framework and foundation for subsequent analysis of the complex interactive effects of operating temperature and insulation resistance of power equipment.

[0070] Next, the current power operation parameters of the power equipment are collected, which will serve as an important basis for subsequent analysis.

[0071] An integrated influence analysis of insulation resistance using a neural network algorithm is employed. First, a suitable neural network structure is constructed for this analysis, comprising an input layer, hidden layers, and an output layer. The number of nodes in the input layer is determined by the characteristic dimensions of the power operating parameters and operating temperature.

[0072] In this embodiment, the neural network algorithm is a computational model that mimics the structure and function of biological neural networks. The basic models include feedforward neural networks (FNNs) (where information flows unidirectionally, suitable for static data modeling), convolutional neural networks (CNNs) (which efficiently process image / time-series data through local connections, parameter sharing, and pooling operations), and recurrent neural networks (RNNs) (which introduce recurrent structures to process sequential data).

[0073] For example, if the power operating parameters include current, voltage, and power factor, plus the factor of operating temperature, the input layer might be set to 4 nodes. The hidden layer can consist of multiple layers, with the number of neurons in each layer determined based on actual conditions. The output layer has only one node, whose function is to output the insulation resistance change value. During the neural network training phase, a large amount of historical data needs to be collected. This includes historical operating temperature sequences, historical power operating parameter sequences, and corresponding historical insulation resistance change value sequences. Using the historical operating temperature and power operating parameter sequences as inputs and the corresponding historical insulation resistance change value sequences as outputs, the neural network is trained using the stochastic gradient descent algorithm. By continuously adjusting the weights and biases of the neural network, it learns the intrinsic relationship between operating temperature and power operating parameters and the insulation resistance change value. When performing real-time analysis to obtain the first influencing factor of insulation resistance change value, the current operating temperature and power operating parameters of the power equipment are first obtained. Then, the operating temperature and power operating parameters are preprocessed to meet the requirements of the neural network input layer, such as data normalization. The preprocessed operating temperature and power operating parameters are then input into the trained neural network, at which point the neural network outputs the required first influencing factor of insulation resistance change value. This neural network-based method can automatically learn the variation law of insulation resistance based on operating temperature and power operating parameters, thereby effectively obtaining the first value of the insulation resistance variation.

[0074] After obtaining the first-influence insulation resistance change value, it needs to be superimposed on the original insulation resistance to obtain the first-influence insulation resistance. First, the original insulation resistance value is defined; it is the insulation resistance value between the high-voltage winding and the casing inside the power equipment, obtained through appropriate measurement methods. Then, this original insulation resistance value is superimposed on the calculated first-influence insulation resistance change value. Since the first-influence insulation resistance change value is based on the prediction of the possible changes in insulation resistance caused by temperature and power operating parameters, and generally, an increase in temperature will decrease the insulation resistance, the superposition method is to subtract the first-influence insulation resistance change value from the original insulation resistance value. The result obtained is the first-influence insulation resistance value after considering the influence of the current operating temperature and power operating parameters.

[0075] A large amount of historical insulation resistance data, power operation parameter data, and corresponding operating temperature change data are collected as a training set. A neural network model containing an input layer, hidden layers, and an output layer is constructed. Input layer nodes correspond to insulation resistance values ​​and power operation parameters (such as current and voltage), while output layer nodes correspond to operating temperature changes. The insulation resistance values ​​and power operation parameters from the training set are used as inputs, and the operating temperature changes are used as outputs to train the neural network. The weights and biases of the neural network are continuously adjusted using the backpropagation algorithm, enabling the model to learn the relationship between insulation resistance, power operation parameters, and operating temperature changes. When the current insulation resistance value and power operation parameters are obtained, they are input into this embodiment. The backpropagation algorithm is the core method for training the neural network; it guides the update of weights and biases by calculating the gradient of the loss function with respect to the network parameters.

[0076] Calculation: Solve for the partial derivatives of the loss function L with respect to the weights W and bias b. and Used for parameter updates:

[0077]

[0078] in, It is the learning rate.

[0079] Backpropagation here is essentially finding the derivative of a composite function. Assume the network's... The layer output is:

[0080]

[0081] Where σ is the activation function, then the loss function is... The gradient is:

[0082]

[0083] In the trained neural network model, the output of the neural network model is the value that first affects the operating temperature change.

[0084] In this embodiment, data acquisition of the power transformer begins with the deployment of various sensors, covering multiple dimensions of parameters including electrical, mechanical, and thermodynamic parameters. For example, electrical parameters such as voltage (U), current (I), power (P, Q), and power factor (pf) can be acquired through voltage transformers (PT) and current transformers (CT). Temperature parameters such as winding temperature, oil temperature, and casing temperature can be acquired using thermocouples, resistance temperature detectors (RTDs), or infrared sensors.

[0085] Taking oil-immersed transformer monitoring as an example, the complete data acquisition process is as follows: 1) Signal sensing: For example, oil temperature sensor (RTD), current transformer (CT), and vibration sensor simultaneously acquire data. 2) Signal conditioning: For example, the CT output is converted into a 0-5V signal by an isolation amplifier, and the vibration signal is filtered by a low-pass filter to remove high-frequency noise.

[0086] 3) Data Acquisition: For example, the LabVIEW platform controls the NI9234 module to capture vibration signals at a 10kHz sampling rate, while the NI9206 module acquires three-phase current at 250kS / s. 4) Real-time Processing: For example, the FPGA performs FFT analysis of the vibration spectrum to detect abnormal frequency components. 5) Data Transmission: For example, data is uploaded to a cloud platform via a 4G module to trigger an over-temperature alarm (e.g., tripping when oil temperature > 85℃). 6) Storage and Display: For example, data is archived to an SQL database, and real-time waveforms and health assessment results are displayed on a PC interface.

[0087] Obtain the current operating temperature value and add it to the first influencing operating temperature change value. The addition method is to add the operating temperature to the first influencing operating temperature change value to obtain the first influencing operating temperature after taking into account the effect of insulation resistance.

[0088] Next, based on the newly obtained first influencing operating temperature and the previously related insulation resistance data, the above process is repeated. That is, based on the magnitude of the first influencing operating temperature, an integrated influence analysis of insulation resistance is performed to obtain new insulation resistance-related data; then, based on the new insulation resistance data, an integrated influence analysis of operating temperature is performed to obtain new operating temperature-related data. This process is repeated cyclically. During each cycle, it is checked whether the influencing operating temperature is greater than or equal to the operating temperature threshold, and whether the influencing insulation resistance is less than or equal to the insulation resistance threshold. If either condition is met, the influence analysis has converged. At this point, the cycle stops, and the number of alternating influence analysis rounds is output. This number of rounds represents the number of abnormal superpositions, reflecting the degree of mutual influence between operating temperature and insulation resistance, as well as the potential state of the equipment.

[0089] For example, from a specific monitoring node, we obtain information such as the insulation resistance of the transformer's high-voltage winding to its casing being 35 MΩ, along with power operating parameters closely related to equipment operation, including a load current of 55 A and a voltage of 8.5 kV. First, based on the current insulation resistance value of 350 MΩ, combined with existing power operating parameters, and using a model derived from extensive data on similar equipment and professional theories, we conduct an integrated impact analysis on the predicted operating temperature. Since the insulation resistance is at this value, according to electrical principles, under the current power operating parameters, the heat generated by the resistance when current flows will cause the operating temperature to rise. Assuming the predicted temperature rise is 10°C every 40 seconds, we obtain the initial first-influence operating temperature change value. This value is then superimposed on a default starting temperature value (e.g., a standard ambient temperature of 25°C) to obtain a new first-influence operating temperature, assumed to be 35°C. Next, based on this newly obtained first-influence operating temperature of 35°C, we conduct an integrated impact analysis on the insulation resistance. Referring again to previously accumulated experimental data and characteristic models of similar transformers, it is known that for every 6°C increase in temperature, the insulation resistance decreases by approximately 60MΩ. At this point, 35°C represents a 10°C increase compared to the standard operating temperature of 25°C, thus the change in insulation resistance can be calculated to be approximately -100MΩ. Adding this change to the existing 350MΩ insulation resistance, the new insulation resistance value becomes 250MΩ, indicating that the insulation performance has begun to change. Then, based on the new insulation resistance value of 250MΩ, the integrated impact analysis of operating temperature based on insulation resistance and power operating parameters is repeated. Due to the decrease in insulation resistance, the current leakage situation changes. According to Joule's law, the temperature is estimated to rise by 12°C every 35 seconds, yielding a new operating temperature change value. Adding this to the current operating temperature of 35°C, the updated operating temperature is 47°C. Subsequently, the crucial convergence judgment stage begins. The system checks whether the current operating temperature of 47℃ is greater than or equal to the preset operating temperature threshold (assumed to be 70℃, set according to the equipment's safety operation standards) and whether the current insulation resistance of 250MΩ is less than or equal to the preset insulation resistance threshold (assumed to be 100MΩ). Finding that the convergence condition has not yet been met, the process continues in a loop. With each round of alternating superposition analysis, the estimated value of the other factor is adjusted based on the latest operating temperature or insulation resistance. This process repeats until, for example, the operating temperature reaches 72℃ (greater than or equal to 70℃) or the insulation resistance decreases to 90MΩ (less than or equal to 100MΩ), indicating that the influence analysis has converged. At this point, the loop stops immediately, and the number of rounds of alternating superposition influence analysis is output. Assuming convergence is achieved after 4 rounds, this 4 represents the number of abnormal superpositions.As a key indicator that accurately reflects changes in the internal state of equipment, a larger value means that the operating temperature and insulation resistance need to go through more rounds of interaction before approaching a fault, indicating that the equipment is currently relatively stable and the possibility of serious problems in the short term is low. A smaller value, on the other hand, warns that the health of the equipment is worrying, that the negative synergistic effect between operating temperature and insulation resistance is strong, and that urgent maintenance measures such as cooling and replacing insulation components are needed to ensure the safe and continuous operation of the power equipment.

[0090] In this embodiment, the alternating influence analysis module is further used for:

[0091] Based on historical operational data of similar power equipment, sample sets of operating temperatures, power operating parameters, and insulation resistance variation values ​​are collected as training data for temperature effects. Similarly, based on historical operational data of similar power equipment, sample sets of insulation resistance, power operating parameters, and temperature variation values ​​are collected as training data for insulation resistance effects. The temperature and insulation resistance training data are divided according to the number of integrated impact analyses. Based on multiple sets of training data from each division, temperature and resistance influence branches of the integrated impact analysis are trained. The trained temperature and resistance influence branches are combined to obtain the temperature and resistance influence analysis paths.

[0092] Specifically, a large number of sample operating temperature values ​​are collected from the historical operating data of similar power equipment, forming a sample operating temperature set. Simultaneously, corresponding power operating parameters, such as current, voltage, and power factor, are obtained to form a sample power operating parameter set. Furthermore, the impact of each sample operating temperature and power operating parameter combination on insulation resistance is determined, forming a sample set of insulation resistance change values.

[0093] Similarly, a set of sample insulation resistance values ​​is collected from historical data, i.e., a large number of insulation resistance values. At the same time, a set of corresponding sample power operation parameters and the change in operating temperature under each sample insulation resistance and power operation parameter combination are obtained, forming a set of sample values ​​affecting the change in operating temperature.

[0094] Determine the number of integrated impact analyses, denoted as n. For the temperature impact training data, divide the sample operating temperature set, sample power operating parameter set, and sample insulation resistance change value set into n groups. For example, if there are 100 sample data points, n=10, then each group has approximately 10 data points. The principle of division is to ensure that each group of data is representative and can cover different operating states and conditions. For the insulation resistance impact training data, divide the sample insulation resistance set, sample power operating parameter set, and sample operating temperature change value set into n groups in the same way. For each group of training data after division, train the temperature impact analysis branch and the resistance impact analysis branch respectively. Taking the temperature impact analysis branch as an example, input a set of data containing part of the sample operating temperature, sample power operating parameters, and sample insulation resistance change values ​​into the corresponding temperature impact analysis branch. This branch employs a neural network, using a defined appropriate loss function (such as the mean squared error loss function) to measure the difference between the predicted results and the true values. Then, it uses stochastic gradient descent to continuously adjust the parameters within the branch, gradually reducing the value of the loss function. This allows the branch to learn the relationship between the sample operating temperature, power operating parameters, and insulation resistance changes. For the resistance influence analysis branch, the training process is similar, except that the input data includes partial sample insulation resistance, sample power operating parameters, and sample temperature change data. The same techniques are used to enable the branch to learn the corresponding relationships.

[0095] In this embodiment, the loss function is a core concept in machine learning and deep learning. It quantifies the difference between the model's prediction and the true value. Mathematically, it maps the prediction error to a non-negative real number, reflecting the model's performance under current parameters. For example, in supervised learning, the loss function measures the difference between the predicted and true distributions; a smaller difference indicates a more accurate model. In statistical decision theory, it represents the "cost" or risk of making a decision under a specific state. Its loss value directly reflects the accuracy of the model's prediction; a smaller loss value means the prediction is closer to the true result. As an optimization objective, model parameters are adjusted using algorithms such as backpropagation and gradient descent to minimize the loss. For example, cross-entropy loss drives classification models to output probabilities closer to the true labels. The changing trend of the loss function provides feedback for model training; for example, mean squared error (MSE) optimizes regression tasks by reducing the squared difference between the prediction and the true value.

[0096] Among them, the mean squared error loss function is one of the most commonly used loss functions in regression tasks, used to measure the average squared difference between the model's predicted values ​​and the true values.

[0097] The specific formula is as follows:

[0098] in, : Sample size;

[0099] : The true value of the i-th sample;

[0100] : The predicted value of the i-th sample.

[0101] During training, a small batch of data (called a mini-batch) is randomly selected from the training dataset each time. The gradient of the loss function on this batch of data is calculated, and then the weights and biases of the neural network are updated based on the gradient.

[0102] In this embodiment, stochastic gradient descent (SGD) is one of the core algorithms used in machine learning and deep learning to optimize model parameters. SGD calculates gradients and updates parameters using only a single sample or a mini-batch of samples in each iteration, making it efficient and flexible, and particularly suitable for large-scale datasets and online learning scenarios.

[0103] Stochastic gradient descent: Each time, a random sample (or mini-batch) is selected to calculate the gradient, and the update formula is:

[0104]

[0105] in It is the loss function for the i-th sample. It is the learning rate.

[0106] The specific algorithm flow includes: 1) Initializing parameters: Randomly initializing the model parameters θ. 2) Iterative update: 2.1 Random sampling: Randomly selecting a sample or mini-batch from the training set. 2.2 Calculating gradient: Calculating the gradient of the loss function with respect to the parameters using the sampled data. 2.3 Updating parameters: Updating the parameters in the opposite direction of the gradient, with the step size determined by the learning rate. Control. 4) Termination condition: The maximum number of iterations is reached or the loss converges (e.g., the change in loss is less than a threshold).

[0107] After training the temperature and resistance influence branches for integrated influence analysis, these branches need to be combined to form temperature and resistance influence analysis paths. For the temperature influence analysis path, all trained branches are integrated according to their specific logic and structure. Each branch has learned the relationship between temperature and insulation resistance under different conditions. Combined, the temperature influence analysis path can comprehensively consider various situations and analyze how temperature affects insulation resistance and the overall operating status of the power equipment. Similarly, for the resistance influence analysis path, all trained branches are combined. These branches each understand the influence of resistance on operating temperature under different conditions. The combined resistance influence analysis path can systematically analyze the impact of resistance on operating temperature and power equipment operation from the perspective of resistance, thus providing an effective analysis path for subsequent alternating superposition influence analysis.

[0108] In this embodiment, the alternating influence analysis module is further used for:

[0109] Calculate the ratio of the operating temperature to the operating temperature threshold, multiply it by the number of integrated influence analyses, and round it to obtain the number of temperature influence analysis branches. Input the operating temperature and power operating parameters into a randomly selected temperature influence analysis branch within the number of temperature influence analysis branches, and output the first set of values ​​affecting the change in insulation resistance. Calculate the average of these values ​​to obtain the first value affecting the change in insulation resistance.

[0110] Specifically, the operating temperature threshold is first defined, a key value set based on the operating characteristics and experience of the power equipment. After obtaining the current operating temperature of the power equipment, the ratio of this operating temperature to the operating temperature threshold is calculated. Next, this ratio is multiplied by a pre-set number of integrated influence analyses; this number determines the level of detail in the analysis. Finally, the result of the multiplication is rounded down; the integer obtained is the number of temperature influence analysis branches, which determines the number of branches used to analyze changes in insulation resistance.

[0111] Based on the previously calculated number of temperature influence analysis branches, a corresponding number of branches are randomly selected from all temperature influence analysis branches. Then, the current operating temperature of the power equipment and relevant power operating parameters (such as current, voltage, and power factor) are used as input data and input into these randomly selected temperature influence analysis branches. Each temperature influence analysis branch is constructed using a neural network algorithm. Upon receiving the input data, it processes it according to its own rules and outputs a predicted value of insulation resistance change. These predicted insulation resistance change values ​​output by each branch collectively constitute the first set of insulation resistance change values. Finally, by calculating the average of all elements in this set, the first set of insulation resistance change values ​​is obtained, which comprehensively reflects the possible changes in insulation resistance under the current operating temperature and power operating parameter conditions.

[0112] In this embodiment, the abnormal fluctuation analysis module is further used for:

[0113] Based on the historical operating temperature sequence and the historical insulation resistance sequence, an impact analysis of the alternating superposition of historical operating temperature and historical insulation resistance is performed to obtain a sequence of historical abnormal superposition times. Multiple sets of historical abnormal superposition times are randomly selected from this sequence, and the deviation percentage of each set is calculated to obtain multiple deviation percentages. Each set of historical abnormal superposition times includes two sets of historical abnormal superposition times. The ratio of the difference between the two sets of historical abnormal superposition times to the larger set of historical abnormal superposition times is calculated as the deviation percentage. The mean of these multiple deviation percentages is calculated to obtain the fluctuation amplitude of the abnormal superposition times. The difference between 1 and the fluctuation amplitude of the abnormal superposition times is used as a compensation coefficient, which is multiplied by the number of abnormal superposition times to obtain the compensated abnormal superposition times.

[0114] Specifically, historical operating temperature and insulation resistance values ​​are sequentially obtained from historical operating temperature and insulation resistance sequences. Simultaneously, relevant power operating parameters are extracted for each historical moment. For each pair of historical operating temperature and insulation resistance values, relevant parameters for integrated impact analysis are set based on their magnitude. For example, the number of integrated impact analyses and corresponding branch structures involved in the temperature and resistance impact analysis paths are determined. Based on the historical operating temperature values ​​and the corresponding power operating parameters, integrated impact analysis of insulation resistance is performed according to the set analysis method. After calculating the insulation resistance change value, it is superimposed on the original historical insulation resistance value to obtain a new insulation resistance value. Then, based on the new insulation resistance value and the power operating parameters, integrated impact analysis of operating temperature is performed to obtain the operating temperature change value, which is then superimposed on the original historical operating temperature value to obtain a new operating temperature value. This alternating superposition process is repeated until the convergence condition is met. The convergence condition is that the affected operating temperature is greater than or equal to the operating temperature threshold set for the corresponding historical moment, or the affected insulation resistance is less than or equal to the insulation resistance threshold set for the corresponding historical moment. Each time the convergence condition is met, the number of rounds of this alternating superposition impact analysis is recorded. Repeat the above process for all historical operating temperatures and historical insulation resistance values, and arrange the recorded alternating superposition effect analysis rounds in sequence to form a sequence of historical anomaly superposition times.

[0115] Multiple data sets are randomly selected from the historical anomaly overlap sequence. For each selected data set, there are two historical anomaly overlap counts, denoted as n1 and n2. Then, the absolute value of the difference between these two historical anomaly overlap counts is calculated. Next, determine the larger of n1 and n2, denoted as . Finally, calculate This yields the percentage deviation for the number of historical anomaly stacks in each group. Repeating this process will yield multiple percentage deviations.

[0116] Calculate the mean of these deviation percentages; this mean is the fluctuation amplitude of the number of anomaly overlaps. It reflects the degree of fluctuation in the historical number of anomaly overlaps.

[0117] The compensation coefficient is obtained by subtracting the fluctuation amplitude of the number of abnormal superpositions from 1. Finally, the number of abnormal superpositions obtained from the current analysis is multiplied by this compensation coefficient to obtain the compensated number of abnormal superpositions. The compensated number of abnormal superpositions takes into account the fluctuation of historical data and can more accurately reflect the actual status of the equipment.

[0118] In this embodiment, the device monitoring and management module is further used for:

[0119] Obtain the threshold for the number of normal abnormal superpositions of normal power equipment; calculate the ratio of the number of compensated abnormal superpositions to the threshold for the number of normal abnormal superpositions to obtain the health parameters of the power equipment; when the health parameters are less than the health parameter threshold, perform health maintenance management on the power equipment.

[0120] Specifically, to obtain the threshold for the number of normal anomaly superpositions for normal power equipment, it is first necessary to collect a large amount of data from a large number of similar transformers operating normally. This data should cover the equipment's operating information under different operating conditions and environmental conditions. The collected data is then analyzed. The historical operating temperature sequence and historical insulation resistance sequence for each device within a preset historical time range are obtained through the operating data acquisition module, along with the corresponding power operating parameters. Then, following the method of the alternating influence analysis module in the system, an alternating influence analysis of operating temperature and insulation resistance is performed on each device to obtain the number of anomaly superpositions for each device. Next, statistical analysis is performed on the number of anomaly superpositions for these normal devices, calculating statistics such as the mean and standard deviation. Finally, based on actual needs and experience, this statistical information is comprehensively considered to determine the threshold for the number of normal anomaly superpositions. This threshold will serve as an important standard for measuring whether the equipment is operating normally.

[0121] After obtaining the number of compensated abnormal stackings, it is compared with the threshold for the number of normal abnormal stackings. The ratio between the two is calculated, i.e., health parameter = number of compensated abnormal stackings / threshold for the number of normal abnormal stackings. This ratio reflects the current health status of the power equipment relative to normal equipment.

[0122] A health parameter threshold is set, determined based on equipment reliability and maintenance requirements. When the calculated health parameter falls below this threshold, it indicates a potential problem with the equipment's health, necessitating health maintenance management. Health maintenance management measures may include further inspection, repair, and adjustment of equipment operating parameters to ensure normal equipment operation and prevent malfunctions and unplanned downtime.

[0123] Example 2

[0124] like Figure 2 As shown, this invention also discloses an artificial intelligence-based method for health management of power equipment, which includes the following steps:

[0125] Step S100: Collect the current operating temperature and insulation resistance of the power equipment, and collect the historical operating temperature sequence and historical insulation resistance sequence of multiple historical moments within a preset historical time range, wherein the power equipment is a transformer.

[0126] Step S200: Based on the operating temperature and insulation resistance, and combined with the current power operating parameters of the power equipment, perform an impact analysis of the alternating superposition of operating temperature and insulation resistance to obtain the number of abnormal superpositions. Among them, an integrated impact analysis is performed based on the magnitude of operating temperature and insulation resistance.

[0127] Step S300: Based on the historical operating temperature sequence and historical insulation resistance sequence, analyze and obtain the historical abnormal superposition number sequence, calculate the fluctuation amplitude of the abnormal superposition number, compensate for the abnormal superposition number, and obtain the compensated abnormal superposition number.

[0128] Step S400: Based on the number of compensation anomaly superpositions, generate health parameters for the power equipment and perform health maintenance management on the power equipment.

[0129] In this embodiment, step S100 further includes the following steps:

[0130] Step S110: Collect the current operating temperature of the power equipment and the insulation resistance between the high-voltage winding inside the power equipment and the outer casing.

[0131] Step S120: Set multiple historical moments within a preset historical time range, wherein the time interval between two adjacent historical moments in each group is the same.

[0132] Step S130: Obtain the historical operating temperature and historical insulation resistance at the multiple historical moments to obtain the historical operating temperature sequence and historical insulation resistance sequence.

[0133] In this embodiment, step S200 further includes the following steps:

[0134] Step S210: Train the influence analysis channel, wherein the influence analysis channel includes a temperature influence analysis path and a resistance influence analysis path, the temperature influence analysis path includes a temperature influence analysis branch that integrates the number of influence analyses, and the resistance influence analysis path includes a resistance influence analysis branch that integrates the number of influence analyses.

[0135] Step S220: Collect the current power operation parameters of the power equipment.

[0136] Step S230: Based on the operating temperature, perform an integrated influence analysis on the insulation resistance based on the operating temperature and power operating parameters to obtain the first value affecting the change in insulation resistance.

[0137] Step S240: Superimpose the insulation resistance and the first influencing insulation resistance change value to obtain the first influencing insulation resistance.

[0138] Step S250: Based on the magnitude of the insulation resistance, perform an integrated influence analysis on the operating temperature based on the insulation resistance and the power operating parameters to obtain the first value affecting the change in operating temperature.

[0139] Step S260: Superimpose the operating temperature and the first influencing operating temperature change value to obtain the first influencing operating temperature.

[0140] Step S270: Continue the influence analysis of alternating superposition of operating temperature and insulation resistance until the influence on operating temperature is greater than or equal to the operating temperature threshold or the influence on insulation resistance is less than or equal to the insulation resistance threshold. The influence analysis converges, and the number of rounds of alternating superposition influence analysis is output to obtain the number of abnormal superpositions.

[0141] In this embodiment, step S200 further includes the following steps:

[0142] Step S280: Based on the historical operating data of similar power equipment, collect sample operating temperature sets, sample power operating parameter sets, and sample insulation resistance change value sets as temperature influence training data; based on the historical operating data of similar power equipment, collect sample insulation resistance sets, sample power operating parameter sets, and sample operating temperature change value sets as insulation resistance influence training data; according to the number of integrated influence analyses, divide the temperature influence training data and insulation resistance influence training data respectively, and train the temperature influence analysis branch and resistance influence analysis branch of the integrated influence analysis number respectively based on the multiple sets of training data of the divided integrated influence analysis number; combine the trained temperature influence analysis branch and resistance influence analysis branch of the integrated influence analysis number to obtain the temperature influence analysis path and resistance influence analysis path.

[0143] In this embodiment, step S200 further includes the following steps:

[0144] Step S290: Calculate the ratio of the operating temperature to the operating temperature threshold, multiply it by the number of integrated influence analyses and round it to obtain the number of temperature influence analysis branches; input the operating temperature and power operating parameters into the randomly selected temperature influence analysis branches of the number of temperature influence analysis branches, output the first set of values ​​affecting the change in insulation resistance, and calculate the average value to obtain the first value affecting the change in insulation resistance.

[0145] In this embodiment, step S300 further includes the following steps:

[0146] Step S310: Based on the historical operating temperature sequence and the historical insulation resistance sequence, perform an impact analysis on the alternating superposition of historical operating temperature and historical insulation resistance to obtain a sequence of historical abnormal superposition times.

[0147] Step S320: Randomly select multiple sets of historical anomaly overlap times within the historical anomaly overlap times sequence, calculate the deviation percentage of each set of historical anomaly overlap times, and obtain multiple deviation percentages. Each set of historical anomaly overlap times includes two historical anomaly overlap times. Calculate the ratio of the difference between the two historical anomaly overlap times to the larger historical anomaly overlap time as the deviation percentage.

[0148] Step S330: Calculate the mean of the multiple deviation percentages to obtain the fluctuation amplitude of the number of abnormal superpositions.

[0149] Step S340: Use the difference between 1 and the fluctuation amplitude of the number of abnormal superpositions as a compensation coefficient, multiply it by the number of abnormal superpositions, and obtain the number of compensated abnormal superpositions.

[0150] In this embodiment, step S400 further includes the following steps:

[0151] Step S410: Obtain the threshold for the number of times normal abnormalities are superimposed on normal power equipment.

[0152] Step S420: Calculate the ratio of the number of compensation abnormal superpositions to the threshold of the number of normal abnormal superpositions to obtain the health parameters of the power equipment.

[0153] Step S430: When the health parameter is less than the health parameter threshold, perform health maintenance management on the power equipment.

[0154] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0155] The above description is merely a preferred embodiment of the present invention and does not limit the scope of patent protection of the present invention. Any equivalent structural transformations made based on the description and drawings of the present invention, whether directly or indirectly applied to other related technical fields, are similarly included within the scope of protection of the present invention.

[0156] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this invention without departing from its scope. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this invention is intended to include such modifications and variations.

Claims

1. An artificial intelligence-based power equipment health management system, characterized in that, include: A data acquisition module is provided, which is used to acquire the current operating temperature and insulation resistance of the power equipment, and to acquire the historical operating temperature sequence and historical insulation resistance sequence of multiple historical moments within a preset historical time range; An alternating influence analysis module is used to perform an alternating superposition influence analysis of operating temperature and insulation resistance based on the operating temperature and insulation resistance, combined with the current power operating parameters of the power equipment, to obtain the number of abnormal superpositions. An abnormal fluctuation analysis module is used to analyze and obtain a historical abnormal superposition number sequence based on the historical operating temperature sequence and historical insulation resistance sequence, calculate the fluctuation amplitude of the abnormal superposition number, compensate for the abnormal superposition number, and obtain the compensated abnormal superposition number. as well as A device monitoring and management module is used to generate health parameters of the power equipment based on the number of compensation anomaly superpositions, and to perform health maintenance management on the power equipment; the alternating influence analysis module performs integrated influence analysis on insulation resistance based on the operating temperature and power operating parameters to obtain the value affecting the change in insulation resistance. The insulation resistance and the value affecting the change in insulation resistance are superimposed to obtain the value affecting the insulation resistance; the alternating influence analysis module performs an integrated influence analysis on the operating temperature based on the magnitude of the insulation resistance and the power operating parameters to obtain the value affecting the change in operating temperature. The operating temperature and the value affecting the change in operating temperature are superimposed to obtain the operating temperature. The alternating influence analysis module continues to perform the influence analysis of alternating superposition of operating temperature and insulation resistance until the operating temperature is greater than or equal to the operating temperature threshold or the insulation resistance is less than or equal to the insulation resistance threshold. The influence analysis converges and the number of rounds of alternating superposition influence analysis is output to obtain the number of abnormal superpositions.

2. The artificial intelligence-based power equipment health management system according to claim 1, characterized in that, The electrical equipment includes transformers.

3. The artificial intelligence-based power equipment health management system according to claim 1, characterized in that, The alternating influence analysis module is also used to train influence analysis channels, wherein the influence analysis channels include a temperature influence analysis path and a resistance influence analysis path, the temperature influence analysis path includes a temperature influence analysis branch that integrates the number of influence analyses, and the resistance influence analysis path includes a resistance influence analysis branch that integrates the number of influence analyses.

4. The artificial intelligence-based power equipment health management system according to claim 1, 2, or 3, characterized in that, The alternating influence analysis module is also used to collect a set of sample operating temperatures, a set of sample power operating parameters, and a set of sample influence insulation resistance changes based on the operating data of similar power equipment over a historical period, as training data for temperature influence. Based on the historical operating data of similar power equipment, a set of sample insulation resistance, a set of sample power operating parameters, and a set of sample values ​​affecting operating temperature changes were collected as training data for the influence of insulation resistance. According to the number of integrated impact analyses, the temperature impact training data and insulation resistance impact training data are divided into categories, and the temperature impact analysis branch and insulation resistance impact analysis branch of the integrated impact analysis are trained according to the multiple sets of training data of the divided integrated impact analysis categories. By combining the temperature influence analysis branch and the resistance influence analysis branch of the integrated influence analysis completed by the combined training, the temperature influence analysis path and the resistance influence analysis path are obtained.

5. The artificial intelligence-based power equipment health management system according to claim 1, 2, or 3, characterized in that, The abnormal fluctuation analysis module is used to randomly select multiple sets of historical abnormal overlaps within the historical abnormal overlap sequence, calculate the deviation percentage of each set of historical abnormal overlaps, and obtain multiple deviation percentages. Each set of historical abnormal overlaps includes two historical abnormal overlaps. The ratio of the difference between the two historical abnormal overlaps to the larger historical abnormal overlap is calculated as the deviation percentage. The mean of the multiple deviation percentages is calculated to obtain the fluctuation amplitude of the abnormal overlap. The difference between 1 and the fluctuation amplitude of the abnormal overlap is used as a compensation coefficient, which is multiplied by the abnormal overlap to obtain the compensated abnormal overlap.

6. The artificial intelligence-based power equipment health management system according to claim 1, 2, or 3, characterized in that, The equipment monitoring and management module is used to obtain the threshold number of normal abnormal superpositions of normal power equipment; and to calculate the ratio of the number of compensated abnormal superpositions to the threshold number of normal abnormal superpositions to obtain the health parameters of the power equipment. When the health parameter is less than the health parameter threshold, health maintenance management is performed on the power equipment.

7. A method for health management of power equipment based on artificial intelligence, characterized in that, This method is implemented based on the artificial intelligence-based power equipment health management system according to any one of claims 1-6, and includes the following steps: The current operating temperature and insulation resistance of the power equipment are collected, and historical operating temperature sequences and historical insulation resistance sequences are collected for multiple historical moments within a preset historical time range, wherein the power equipment is a transformer; Based on the operating temperature and insulation resistance, and combined with the current power operating parameters of the power equipment, an impact analysis of the alternating superposition of operating temperature and insulation resistance is performed to obtain the number of abnormal superpositions. Among them, an integrated impact analysis is performed based on the magnitude of operating temperature and insulation resistance. Based on the historical operating temperature sequence and historical insulation resistance sequence, the historical abnormal superposition number sequence is obtained by analysis, the fluctuation amplitude of the abnormal superposition number is calculated, and the abnormal superposition number is compensated to obtain the compensated abnormal superposition number. Based on the number of compensation anomaly superpositions, health parameters of the power equipment are generated, and health maintenance management of the power equipment is performed.

Images