Power equipment maintenance management system and method based on artificial intelligence

Abstract

The invention relates to the field of power equipment intelligent control, and discloses a power equipment maintenance management system and method based on artificial intelligence, and the method comprises the steps: carrying out the multi-source data collection through a sensor, eliminating the noise of the sensor, generating a three-dimensional thermal field model, displaying the temperature gradient between power equipment, and recognizing a high-temperature aggregation region; the heat conduction coefficient and the convection heat dissipation efficiency are corrected in real time by combining environmental parameters to build a model, historical data temperature changes are analyzed, the temperature abnormal condition is analyzed through a temperature threshold value according to the average temperature change and the highest temperature in a period, and the heat dissipation power is adjusted in time; the temperature of the power equipment is predicted based on the temperature change trend of the power equipment, the future heat dissipation efficiency is adjusted, the heat dissipation efficiency is improved to the minimum, the temperature of the power equipment after the heat dissipation efficiency is adjusted is predicted, the actual temperature and the adjusted predicted temperature are analyzed, the deviation is compared, and the formula for adjusting the heat dissipation power is dynamically compensated. And ensuring that the deviation degree and the deviation proportion meet preset standards.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent control of power equipment, and in particular to an artificial intelligence-based power equipment maintenance management system and method. Background Art

[0002] In existing technologies, traditional temperature monitoring relies heavily on a single type of sensor, resulting in limited data coverage and a lack of high-precision time synchronization mechanisms. Time deviations in data from different sensors can distort three-dimensional temperature field models, blur the positioning of high-temperature areas, and make it difficult to accurately identify microscopic temperature gradient differences. Existing noise suppression methods fail to consider the stratification characteristics of noise frequency bands, leading to signal distortion. For example, the mixed processing of sudden electromagnetic interference and sensor drift makes it difficult for fixed-parameter Kalman filters to adapt to environmental changes, resulting in large fluctuations in temperature measurements and delayed response of dynamic heat dissipation control, which can easily lead to light decay or local overheating.

[0003] Traditional interpolation ignores the physical spatial correlation of the temperature field, resulting in large temperature estimation errors in sparse sensor areas. The model cannot reflect the internal heat conduction path of the power equipment, and the heat dissipation optimization lacks a scientific basis. The model prediction deviation accumulates with environmental changes. The heat dissipation power adjustment strategy is out of touch with actual needs. The identification of high-temperature clustered areas relies on a simple threshold method, which is prone to missing scattered anomalies. The prediction model does not combine real-time deviation feedback for dynamic compensation, resulting in temperature control lag, the risk of overshoot or underadjustment in the heat dissipation power adjustment, and a reduction in the life of the power equipment and the stability of the light efficiency.

[0004] Based on this, the present invention is required to provide an artificial intelligence-based power equipment maintenance management system and method. Summary of the Invention

[0005] The purpose of the present invention is to provide an artificial intelligence-based power equipment maintenance management system and method. To solve the above-mentioned existing technical problems, the present invention is implemented through the following technical solutions:

[0006] In a first aspect, an embodiment of the present invention provides an artificial intelligence-based power equipment maintenance management method, comprising:

[0007] Step 1: Collect multi-source data through sensors, eliminate sensor noise, generate a three-dimensional thermal field model to display the temperature gradient between power equipment, and identify high-temperature concentration areas;

[0008] Step 2: Build a model based on environmental parameters to calibrate the thermal conductivity and convection cooling efficiency in real time. Analyze historical temperature changes. Based on the average temperature change and maximum temperature within the cycle, analyze temperature anomalies using temperature thresholds and adjust the cooling power in a timely manner.

[0009] Step 3: Predict the temperature of the power equipment based on the temperature change trend of the power equipment, adjust the future heat dissipation efficiency, minimize the heat dissipation efficiency, and predict the temperature of the power equipment after adjusting the heat dissipation efficiency. Analyze and compare the deviation between the actual temperature and the adjusted predicted temperature, and dynamically compensate the formula for adjusting the heat dissipation power.

[0010] In a second aspect, an embodiment of the present invention provides an artificial intelligence-based power equipment maintenance management system, comprising:

[0011] Data acquisition and synchronization module: collects multi-source data through sensors and maintains time synchronization;

[0012] Data processing and noise reduction module: Based on the improved Kalman filter, it decomposes the sensor signal into layers, performs adaptive noise reduction, and extracts dynamic features;

[0013] Thermal field modeling and analysis module: Generates a three-dimensional thermal field model to display the temperature gradient between power equipment and identify high-temperature concentration areas;

[0014] Power equipment heat dissipation control module: This module builds a model based on environmental parameters to correct the heat transfer coefficient and convection heat dissipation efficiency in real time. It also analyzes temperature changes in historical data, analyzes temperature anomalies based on the average temperature change and maximum temperature within a period, and uses temperature thresholds to analyze temperature anomalies and adjust the heat dissipation power in a timely manner.

[0015] Heat dissipation feedback compensation module: predicts the temperature of power equipment, minimizes heat dissipation efficiency, and predicts the temperature of power equipment after adjusting the heat dissipation efficiency. Analyzes and compares the deviation between the actual temperature and the adjusted predicted temperature, and dynamically compensates the formula for adjusting the heat dissipation power.

[0016] Beneficial effects of the present invention:

[0017] 1. By hybrid deployment of contact micro-thermocouples, infrared thermal imaging sensors, and ambient temperature and humidity sensors, combined with three-dimensional grid division, high-density, multi-dimensional data collection of the temperature field of power equipment is achieved. Wavelet decomposition and improved Kalman filtering are combined to achieve noise frequency band processing and dynamic parameter adjustment. The Kriging interpolation algorithm, combined with the nugget effect, sill value, and range parameter, constructs a thermal field spatial correlation model with clear physical meaning. Based on Bayesian inference and MCMC sampling, the thermal conductivity coefficient k and convection heat dissipation efficiency h are updated in real time, breaking through the limitations of traditional fixed-parameter models. Z-Score normalization and density clustering are integrated to achieve outlier detection and precise regional boundary demarcation.

[0018] 2. Use the time series model to predict the temperature of power equipment in the next hour. Take measures to adjust the heat dissipation efficiency in advance before the temperature actually changes, which helps to avoid damage to power equipment caused by excessive temperature and extend the service life of power equipment. It can also reduce the light attenuation problem caused by abnormal temperature. After predicting the temperature of the power equipment after adjusting the heat dissipation efficiency, compare it with the actual temperature. Dynamically compensate the formula for adjusting the heat dissipation power according to the deviation, and adapt to the changes in the working environment and its own state of the power equipment in real time. When establishing the model, first perform a stationarity test on the historical temperature data, and stabilize it through the difference method. The optimal model parameters are determined by minimizing the information criterion, making full use of the information in the historical data to improve the prediction accuracy of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0020] Figure 1 This is a flowchart of the steps of an artificial intelligence-based power equipment maintenance management method provided in Examples 1 and 2 of the present invention;

[0021] Figure 2 This is a structural diagram of an artificial intelligence-based power equipment maintenance management system provided in Example 3 of the present invention. DETAILED DESCRIPTION

[0022] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0023] Example 1

[0024] like Figure 1 As shown, an embodiment of the present invention provides an artificial intelligence-based power equipment maintenance management method, which specifically includes the following steps:

[0025] Step 1: Collect multi-source data through sensors, eliminate sensor noise, generate a three-dimensional thermal field model to display the temperature gradient of power equipment, and identify high-temperature concentration areas;

[0026] In step one:

[0027] In a specific embodiment, the specific process of collecting multi-source data through sensors is as follows:

[0028] Divide the power equipment into three-dimensional space units of size αcm 3 A three-dimensional cube grid is formed, and contact micro-thermocouples and infrared thermal imaging sensors are deployed at key nodes of power equipment. Key nodes include but are not limited to: driver chips, heat sinks and circuit boards. N sensor nodes are arranged in the three-dimensional space of power equipment. Sensors include but are not limited to: contact micro-thermocouples, non-contact infrared arrays, ambient temperature and humidity sensors. The node spacing is preset.

[0029] Micro-thermocouples are embedded in driver chips, heat sinks, and circuit boards to directly measure contact point temperatures, and infrared thermal imaging arrays are deployed around power equipment to capture surface heat distribution.

[0030] It should be noted that through fine grid division, temperature anomaly areas can be accurately located, providing a data basis for subsequent temperature analysis and control. The temperature data of each grid node can be obtained by interpolating the measurement values ​​of nearby thermocouple sensors and the thermal image data of infrared thermal imaging sensors, thereby constructing a three-dimensional temperature field model of the entire power equipment;

[0031] Adopting the Precision Time Protocol (PTP) to ensure that the time deviation of all sensor data is less than the preset time deviation;

[0032] Sensor node A receives a time synchronization message from the master clock at time T1, while the local clock of node A is at time T2. Node A uses the offset measurement algorithm to calculate the time deviation between the local clock and the master clock: ΔT = T2 - T1.

[0033] Node A receives time synchronization messages from the master clock source multiple times to further improve synchronization accuracy and filters the calculated time deviation to eliminate the influence of noise and short-term fluctuations.

[0034] Based on the calculated time deviation, the sensor node adjusts its own local clock;

[0035] If the time deviation is positive, it indicates that the local clock lags behind the master clock. The node speeds up the local clock by adjusting the clock frequency or directly modifying the clock count value, gradually approaching the master clock time.

[0036] If the time deviation is negative, the local clock will be slowed down;

[0037] Adopting feedback control mechanism, constantly monitoring time deviation and dynamically adjusting clock to maintain continuous time synchronization;

[0038] It should be noted that it is crucial to ensure the time synchronization of all sensor data. Due to the differences in the location, data transmission path and processing mechanism of different sensors, if effective synchronization is not performed, the data collected by each sensor will deviate in the time dimension, resulting in errors in the constructed three-dimensional temperature field model, affecting the accurate judgment of the actual temperature condition of the power equipment and the implementation of subsequent control strategies. The Precision Time Protocol (PTP) is a high-precision time synchronization protocol that sets a master clock source. The master clock source uses an atomic clock with high stability and accuracy or a high-precision GPS clock module. The master clock source is responsible for generating accurate time signals and sending time synchronization messages to each sensor node distributed in the three-dimensional space of the power equipment through a network link.

[0039] In a specific embodiment, the specific process of eliminating sensor noise is:

[0040] Decompose and layer the sensor signals at different scales and locations;

[0041] Specifically, for the temperature sensor signal, the burst interference has high-frequency characteristics, while the trend part of the signal is concentrated in the low frequency;

[0042] When decomposing into multiple layers, the division of different frequency ranges is based on the study of the frequency characteristics of common noise sources;

[0043] For example, when the sensor signal is decomposed into three layers at different scales and locations, electromagnetic interference in the circuit is typically in the relatively high frequency range of 62.5Hz-125Hz in the D1 layer, environmental interference may be in the slightly lower frequency range of 31.25Hz-62.5Hz in the D2 layer, and sensor drift is relatively low frequency, in the D3 layer of 15.625Hz-31.25Hz. The SURE criterion is used to calculate the adaptive threshold, automatically adjusting the degree of noise reduction based on the characteristics of the signal itself, to avoid signal distortion caused by excessive noise reduction or insufficient noise reduction.

[0044] In the state space model, the temperature value and the temperature change rate are used as state variables to enable the state space model to capture the dynamic change trend of temperature; the state transfer matrix Among them, Δt represents the preset sampling period;

[0045] For example, during the sampling period, the temperature value will change according to its rate of change, and the rate of change itself is relatively stable when there is no drastic change due to external factors, which is reflected in the matrix elements. The observation matrix H =

[10] only observes the temperature value. The actual temperature sensor directly measures the temperature, not the temperature change rate;

[0046] The adaptive Kalman filter is dynamically adjusted by the noise covariance, and the initial value of the process noise Q0 is set to The elements of the diagonal matrix correspond to the initial uncertainty of the temperature value and the temperature change rate, respectively, through the formula Calculate the residual ∈ k ,in, It represents the predicted value, and k represents the number of adjustments;

[0047] Update the formula by process noise Calculate the process noise Q k , where β represents the preset proportional coefficient, K k represents the Kalman gain;

[0048] If the residual between the predicted value and the observed value is large, it means that the uncertainty of the system state prediction increases, and Q needs to be increased. k value, in order to rely more on the observed value to correct the prediction, the observation noise R k Similarly, the initial value R0 = 0.5, and the formula is updated by observing the noise Observation noise R k , by dynamically adjusting the observation noise covariance, the Kalman filter can adapt to the uncertainty of the observation data;

[0049] In a specific embodiment, the specific process of generating a three-dimensional thermal field model by combining the Kriging interpolation algorithm is as follows:

[0050] Through mathematical models Describes the spatial correlation of temperature, where γ(h) represents the semivariogram value at distance h, c1 represents the base number, and b represents the range parameter;

[0051] It should be noted that the nugget effect c0 = 0.2 °C 2 It represents the random variation variance of temperature within a very short distance, reflecting the non-spatial structured fluctuations caused by measurement errors and micro-environmental differences.

[0052] For example, in power equipment, even if two sensors are very close to each other, slight differences in chip manufacturing processes and solder joints can still cause random fluctuations in the measured temperature. This random fluctuation is reflected in the nugget effect.

[0053] Base value c1=1.5℃ 2 Indicates the maximum variance of temperature change. When the distance h exceeds the range b = 10 cm, the spatial correlation of temperature tends to be stable. At this time, the variance of temperature change reaches the maximum, that is, the base value c1 = 1.5 °C 2 At a distance greater than the range, the mutual influence between temperature values ​​reaches a stable state and no longer changes significantly with increasing distance;

[0054] It should be noted that the range b = 10 cm determines the effective range of temperature spatial autocorrelation. Within the range distance, temperature values ​​influence each other. The closer the distance, the stronger the correlation. Beyond the range distance, the correlation weakens significantly.

[0055] For example, when the distance between the monitoring points of two key nodes is less than or equal to 10 cm, the temperature changes of the two key nodes affect each other, and the heating of one key node causes the temperature of the other key nodes nearby to rise; when the distance is greater than 10 cm, the mutual influence relationship becomes very weak;

[0056] In three-dimensional space, solve the weight coefficient λ i , accurately estimate the temperature T at any point (x,y,z) x,y,z , the weight coefficient satisfies the unbiased property Ensure that the interpolation results are statistically accurate and avoid systematic deviations; at the same time, meet the minimum variance requirement so that the estimated value is close to the true value; in this way, the temperature values ​​T of the surrounding known points are used to calculate the interpolation results. i , through the weighted summation formula Accurately estimate the temperature at a point in an area where sensors are sparsely distributed;

[0057] For example, in some areas of power equipment, where there are relatively few sensors, the temperature of other locations in the area can be reasonably inferred based on the limited sensor measurements in the surrounding area by combining the Kriging interpolation algorithm;

[0058] By calculating the temperature gradient, the rate of change of temperature in all directions in space is monitored to understand the direction and intensity of heat transfer. In power equipment, areas with large temperature gradients indicate paths where heat is concentrated or areas with poor heat dissipation.

[0059] For example, if the temperature gradient is large at the connection between the power equipment component and the heat sink, it means that there is an obstacle in the process of heat transfer from the power equipment component to the heat sink, and the heat dissipation structure needs to be optimized;

[0060] The central difference method is used for discretization calculation, based on the gradient in the x-axis direction, through the formula Calculate the temperature gradient in the x-axis direction Among them, T x+Δx,y,z It represents the temperature value at the coordinate x+Δx,y,z, T x-Δx,y,z It represents the temperature value at the coordinate x-Δx,y,z, where Δx represents the grid spacing in the x-axis direction;

[0061] It should be noted that the temperature gradient is calculated by the temperature difference between adjacent grid points. The selection of the grid spacing Δx needs to take into account the calculation efficiency and accuracy requirements. A smaller Δx improves the accuracy and increases the calculation amount; a larger Δx reduces the accuracy and reduces the calculation amount.

[0062] Specifically, the blue area represents a low-temperature region with good heat dissipation, while the red area is a high-temperature region that requires attention. The T = 70°C isothermal surface is drawn to identify the critical high-temperature region. For power equipment, temperatures exceeding 70°C affect the lifespan and performance of power equipment components.

[0063] By rendering the isothermal surface, you can intuitively see which areas have temperatures approaching or exceeding the critical value, making it easier to take cooling measures in a timely manner.

[0064] Specifically, the isothermal surface revealed that the temperature near the power equipment components was close to 70°C. The heat dissipation structure of the power equipment was adjusted in a timely manner, and ventilation conditions were improved to reduce the temperature.

[0065] The gradient vector field is displayed by superimposing arrows. The length of the arrows is proportional to the gradient strength and points to the area with the fastest temperature increase. This allows the direction and intensity of heat flow to be analyzed.

[0066] By observing the gradient vector field, we can find the path of heat transfer from high temperature area to low temperature area, as well as the difference in transfer intensity in different parts;

[0067] For example, the gradient vector field reveals that heat is mainly transferred from the center of the power equipment components to the surrounding areas, and the transfer intensity is strong in the y-axis direction. This provides a basis for optimizing the heat dissipation structure and adding heat dissipation fins in the y-axis direction where the heat transfer intensity is strong.

[0068] The deviation of each grid point temperature from the neighborhood mean is calculated by Z-Score normalization. To determine whether the grid point temperature is abnormal, when Z ≥ 2.5, it is marked as a high temperature gathering point;

[0069] Based on the DBSCAN algorithm, adjacent outliers are merged to define the boundary of the high-temperature cluster area; the neighborhood radius r ∈ =2d and the minimum number of points MinPts=3. The settings of the neighborhood radius and the minimum number of points are determined based on the actual situation and experience of the power equipment;

[0070] It should be noted that a smaller neighborhood radius accurately defines the boundaries of the high-temperature cluster area, but it will cause some adjacent outliers to be unable to be merged; a larger neighborhood radius has the opposite effect. The minimum number of points MinPts = 3 means that at least 3 adjacent outliers are required to form a high-temperature cluster area, avoiding the misjudgment of isolated outliers as high-temperature cluster areas;

[0071] By randomly hiding 10% of the sensor data, interpolating the remaining 90% of the data, and comparing the interpolated result with the actual 10% of data, the root mean square error (RMSE) is calculated. This comprehensively reflects the degree of deviation between the interpolated result and the actual value, and the root mean square error (RMSE) is controlled to RMSE ≤ 0.8°C to ensure the accuracy of the interpolation algorithm.

[0072] For example, if the root mean square error (RMSE) is consistently less than 0.8°C across multiple validations, it indicates that the constructed temperature spatial correlation model and interpolation algorithm accurately estimate temperature values, providing a reliable data foundation for subsequent temperature analysis, identification of high-temperature cluster areas, and formulation of heat dissipation strategies.

[0073] Step 2: Build a model based on environmental parameters to calibrate the thermal conductivity and convection cooling efficiency in real time. Analyze historical temperature changes. Based on the average temperature change and maximum temperature within the cycle, analyze temperature anomalies using temperature thresholds and adjust the cooling power in a timely manner.

[0074] Collect environmental parameters and temperature data of power equipment. Environmental parameters include but are not limited to: ambient temperature, humidity, and wind speed;

[0075] Establish a heat balance equation. For power equipment, the heat balance equation can be expressed as: Among them, C represents the heat capacity of the power equipment, P heat It represents the heating power of the power equipment, k is the thermal conductivity coefficient, h is the convection heat dissipation efficiency, A is the heat dissipation area, ΔT = TT env , T represents the temperature of the power equipment, T env Indicates the ambient temperature;

[0076] Incorporating environmental parameters into the model, humidity affects the thermal conductivity of air, which in turn affects the heat transfer coefficient k; wind speed directly affects the convective heat dissipation efficiency h, and a functional relationship is established;

[0077] Define the prior distribution of thermal conductivity k and convection heat dissipation efficiency h;

[0078] According to Bayes' theorem, through the formula The posterior probability P(k,h|D) is calculated, where D is the collected environmental parameters and temperature data, 2σ 2 represents the variance of the measurement noise, It represents the temperature at the i-th time point predicted by the model. It represents the measured temperature at the i-th time point;

[0079] The likelihood function P(D|k,h) is calculated using the heat balance equation and measurement data. After discretizing the heat balance equation, the temperature change under given k and h is calculated using numerical methods and compared with the actual measured temperature data to obtain the likelihood function value.

[0080] The Markov Chain Monte Carlo (MCMC) method is used to sample from the posterior distribution to obtain updated values ​​of k and h, thereby achieving real-time correction of the thermal conductivity coefficient and convection heat dissipation efficiency.

[0081] It should be noted that Markov Chain Monte Carlo (MCMC) is a statistical method that samples complex probability distributions by constructing Markov chains;

[0082] In a specific embodiment, the method of analyzing temperature anomalies through temperature thresholds and adjusting heat dissipation power in a timely manner is:

[0083] Preprocess the collected historical temperature data to remove outliers and noise;

[0084] Specifically, through the formula Calculate the average temperature within a unit period Where n is the number of temperature measurements in a cycle, T i is the temperature value measured for the i-th time; at the same time, the highest temperature T in the cycle is determined max =maxT i ;

[0085] Set the safety temperature threshold T safe , warning temperature threshold T warning and the dangerous temperature threshold T critical , determined according to the material characteristics and service life requirements of the power equipment;

[0086] By comparing the average temperature and the maximum temperature T max The temperature threshold is used to determine the abnormal temperature. For example, if or T max >T critical , it is considered that temperature anomaly occurs;

[0087] Adjust the heat dissipation power P according to the abnormal temperature cool , using the proportional-integral-differential control algorithm method for adjustment;

[0088] For example, a simple proportional control formula is P cool =K p (TT set ), where K p is the proportional coefficient, T is the average temperature or the maximum temperature of the current temperature, select according to the actual situation, Tset The target temperature is set, close to the safety temperature threshold T safe ;

[0089] When the temperature rises abnormally, increase the heat dissipation power to reduce the temperature;

[0090] The technical solution of the embodiment of the present invention is as follows: by hybrid deployment of contact micro-thermocouples, infrared thermal imaging sensors and ambient temperature and humidity sensors, combined with three-dimensional grid division, high-density, multi-dimensional data collection of the temperature field of power equipment is achieved; by combining wavelet decomposition and improved Kalman filtering, noise frequency band processing and dynamic parameter adjustment are achieved; by adopting the Kriging interpolation algorithm, combining the nugget effect, base value and range parameter, a thermal field spatial correlation model with clear physical meaning is constructed; based on Bayesian inference and MCMC sampling, the thermal conductivity coefficient k and convection heat dissipation efficiency h are updated in real time, breaking through the limitations of traditional fixed parameter models; integrating Z-Score standardization and density clustering, anomaly detection and precise division of regional boundaries are achieved.

[0091] Example 2

[0092] like Figure 1 As shown, an embodiment of the present invention provides an artificial intelligence-based power equipment maintenance management method, which specifically includes the following steps:

[0093] Step 3: Predict the power equipment temperature based on its temperature change trend, adjust the future heat dissipation efficiency to minimize it, and predict the power equipment temperature after the adjustment. Analyze and compare the deviation between the actual temperature and the adjusted predicted temperature, and dynamically compensate the formula for adjusting the heat dissipation power.

[0094] In a specific embodiment, the method of minimizing and improving heat dissipation efficiency is:

[0095] The temperature prediction model is established using the autoregressive integrated moving average model ARIMA;

[0096] First, the historical temperature data is tested for stationarity and then stabilized by the difference method; the autoregressive order p, difference order d, and moving average order q of the model are determined, and the optimal model parameters are determined by minimizing the information criterion;

[0097] Use the established temperature prediction model to predict the temperature T in the next hour pred ;

[0098] Based on the predicted temperature T pred Adjust heat dissipation efficiency;

[0099] Specifically, if T pred >T safe, increase the heat dissipation efficiency h according to the preset increase ratio; let the current heat dissipation efficiency be h0, the adjusted heat dissipation efficiency Where ρ is the preset adjustment coefficient, which is set to 0.5;

[0100] The adjusted heat dissipation efficiency h new Substitute into the heat balance equation, combine with the predicted environmental parameters, and use the finite difference method to predict the predicted heat dissipation temperature T of the power equipment after adjusting the heat dissipation efficiency. post-cool ;

[0101] At the predicted time point, obtain the actual temperature T actual , calculate and adjust the predicted heat dissipation temperature T after the heat dissipation power post-cool Deviation ΔT=T actual -T post-cool ;

[0102] For any key node in the power equipment, calculate the deviation ΔPC between the predicted temperature and the actual temperature i =T i,actual -T i,prcd , where T i,actual is the actual temperature of the i-th key node, T i,prcd is the predicted temperature of the i-th key node;

[0103] Analyze the distribution of deviations and calculate the mean of the deviations Where m is the number of key nodes, and the standard deviation σΔPC i Plot a histogram of the deviations;

[0104] Dynamically compensate the formula for adjusting the heat dissipation power according to the distribution of deviations;

[0105] Specifically, if the deviation from the mean This indicates that the predicted temperature is generally lower than the actual temperature, and the cooling power adjustment needs to be increased. Ensure that the deviation between the predicted temperature after cooling adjustment and the actual temperature and the proportion of power equipment meet the preset standards. For example, the proportion of key nodes with an absolute deviation greater than 1°C is less than 10%.

[0106] The technical solution of an embodiment of the present invention is: using a time series model to predict the temperature of power equipment in the next 1 hour, and taking measures to adjust the heat dissipation efficiency in advance before the temperature actually changes, which helps to avoid damage to power equipment caused by excessive temperature and extend the service life of power equipment. It can also reduce the light decay problem caused by abnormal temperature. After predicting the temperature of the power equipment after adjusting the heat dissipation efficiency, it is compared with the actual temperature, and the formula for adjusting the heat dissipation power is dynamically compensated according to the deviation, so as to adapt to the changes in the working environment and the state of the power equipment in real time. When establishing the model, the historical temperature data is first tested for stationarity and stabilized by the difference method. The optimal model parameters are determined by minimizing the information criterion, and the information in the historical data is fully utilized to improve the prediction accuracy of the model.

[0107] Example 3

[0108] like Figure 2 As shown, an embodiment of the present invention provides an artificial intelligence-based power equipment maintenance management system, which specifically includes the following modules:

[0109] Data acquisition and synchronization module: collects multi-source data through sensors and maintains time synchronization;

[0110] Data processing and noise reduction module: Based on the improved Kalman filter, it decomposes the sensor signal into layers, performs adaptive noise reduction, and extracts dynamic features;

[0111] Thermal field modeling and analysis module: Generates a three-dimensional thermal field model to display the temperature gradient between power equipment and identify high-temperature concentration areas;

[0112] Power equipment heat dissipation control module: This module builds a model based on environmental parameters to correct the heat transfer coefficient and convection heat dissipation efficiency in real time. It also analyzes temperature changes in historical data, analyzes temperature anomalies based on the average temperature change and maximum temperature within a period, and uses temperature thresholds to analyze temperature anomalies and adjust the heat dissipation power in a timely manner.

[0113] Heat dissipation feedback compensation module: predicts the temperature of power equipment based on the temperature change trend of power equipment, adjusts the future heat dissipation efficiency, minimizes the heat dissipation efficiency, and predicts the temperature of power equipment after adjusting the heat dissipation efficiency. Analyze and compare the deviation between the actual temperature and the adjusted predicted temperature, and dynamically compensate the formula for adjusting the heat dissipation power to ensure that the degree of deviation and the deviation ratio meet the preset standards.

[0114] An embodiment of the present invention is described in detail above, but the content described is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention; the above formulas are all dimensionless and numerical calculations, and the formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field based on actual conditions and historical experience, and can be adjusted according to actual conditions; the above description is only a preferred embodiment of the present invention and is not used to limit the present invention. All equal changes and improvements made according to the scope of application of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A power equipment maintenance management method based on artificial intelligence, characterized in that: The following steps are involved: Through multi-source data collection by sensors, sensor noise is eliminated, and a three-dimensional thermal field model is generated to display the temperature gradient between power equipment and identify high-temperature accumulation areas; Combined with environmental parameters, the model is built to correct the thermal conductivity coefficient and convection heat dissipation efficiency in real time. The temperature changes in historical data are analyzed. Based on the average temperature change and the maximum temperature within the cycle, the temperature anomaly is analyzed through the temperature threshold and the heat dissipation power is adjusted in time. Based on the temperature change trend of power equipment, the temperature of power equipment is predicted, the future heat dissipation efficiency is adjusted, the heat dissipation efficiency is minimized, and the temperature of the power equipment after the heat dissipation efficiency is adjusted is predicted. The actual temperature and the adjusted predicted temperature are analyzed and compared for deviation, and the formula for adjusting the heat dissipation power is dynamically compensated.

2. The artificial intelligence-based power equipment maintenance management method according to claim 1, characterized in that: The method for eliminating sensor noise is: Decompose and layer the sensor signals at different scales and locations; In the state space model, the temperature value and the temperature change rate are used as state variables, so that the state space model can capture the dynamic change trend of temperature; Residual feedback drives the dynamic update of noise covariance, achieving a dynamic balance between model predictions and actual observations.

3. The artificial intelligence-based power equipment maintenance management method according to claim 1, characterized in that: The method for displaying the temperature gradient between power equipment is: A mathematical model is constructed using the semivariogram to describe the spatial correlation of temperature. The weight coefficient is solved in three-dimensional space based on unbiasedness. The temperature values ​​of surrounding known points are weighted and summed to accurately estimate the temperature of any point in a sensor-sparse area. The central difference method is used to discretize and calculate the temperature gradient. Based on the x-axis direction, the temperature gradient in this direction is calculated through the temperature values ​​of adjacent coordinate points and the grid spacing. The temperature change rate in all directions of the space is monitored. Areas with large temperature gradients can reflect concentrated heat transfer paths or areas with poor heat dissipation.

4. The artificial intelligence-based power equipment maintenance management method according to claim 1, characterized in that: The method for identifying high temperature accumulation areas is: Visualize temperature distribution through color areas, draw isothermal surfaces to identify critical high-temperature areas, and based on the isothermal surfaces, find that the temperature near the components of power equipment is close to the critical value; Using gradient vector fields to overlay arrow diagrams, the direction and intensity of heat flow can be intuitively displayed through arrow length and direction, and the heat transfer paths and intensity differences in different parts can be analyzed. The Z-Score standardization method is used to calculate the deviation between the grid point temperature and the neighborhood mean. When the deviation is ≥2.5, it is marked as a high-temperature cluster point to determine the temperature anomaly area.

5. The artificial intelligence-based power equipment maintenance management method according to claim 1, characterized in that: The method for analyzing temperature changes in historical data is: The posterior probability P(k,h|D) is calculated by the formula; Where k is the thermal conductivity, h is the convective heat dissipation efficiency, and D is the collected environmental parameters and temperature data; The likelihood function P(D|k,h) is calculated through the heat balance equation and measurement data. After discretizing the heat balance equation, the temperature change under given k and h is calculated using numerical methods and compared with the actual measured temperature data to obtain the likelihood function value.

6. The artificial intelligence-based power equipment maintenance management method according to claim 5, characterized in that: The method for real-time correction of thermal conductivity and convection heat dissipation efficiency is: Establish a heat balance equation. For power equipment, incorporate environmental parameters into the model. Humidity affects the thermal conductivity of air, which in turn affects the heat transfer coefficient k. Wind speed directly affects the convective heat dissipation efficiency h, and a functional relationship is established. The updated values ​​of k and h are obtained by sampling from the posterior distribution through the Markov Chain Monte Carlo method.

7. The artificial intelligence-based power equipment maintenance management method according to claim 1, characterized in that: The method for adjusting the heat dissipation power is: By formula Calculate the average temperature within a unit period Where n is the number of temperature measurements in a cycle, T i is the temperature value measured for the i-th time; at the same time, the highest temperature T in the cycle is determined max =maxT i ; By comparing the average temperature and the maximum temperature T max The temperature threshold is used to determine the abnormal temperature. If or T max >T critical , it is considered that temperature anomaly occurs; Among them, T warning is the warning temperature threshold, T critical is the dangerous temperature threshold; Adjust the heat dissipation power P according to the abnormal temperature cool , using the proportional-integral-differential control algorithm method for adjustment; When the temperature rises abnormally, the heat dissipation power is increased to reduce the temperature.

8. The artificial intelligence-based power equipment maintenance management method according to claim 1, characterized in that: The method for minimizing and improving heat dissipation efficiency is: The temperature prediction model is established using the autoregressive integrated moving average model ARIMA; First, the historical temperature data is tested for stationarity and then stabilized by the difference method; the autoregressive order p, difference order d, and moving average order q of the model are determined, and the optimal model parameters are determined by minimizing the information criterion; Use the established temperature prediction model to predict the temperature T in the next hour pred ; If T pred >T safe , increase the heat dissipation efficiency h according to the preset increase ratio; let the current heat dissipation efficiency be h0, the adjusted heat dissipation efficiency Where ρ is the preset adjustment coefficient, T safe To approach the safety temperature threshold.

9. The artificial intelligence-based power equipment maintenance management method according to claim 1, characterized in that: The method for dynamically compensating the formula for adjusting the heat dissipation power is: Substitute the adjusted heat dissipation efficiency into the heat balance equation, combine it with the predicted environmental parameters, and use the finite difference method to predict the predicted heat dissipation temperature of the power equipment after adjusting the heat dissipation efficiency; At the predicted time point, the actual temperature is obtained and the deviation from the predicted heat dissipation temperature after adjusting the heat dissipation power is calculated; Analyze the distribution of deviations, calculate the mean and standard deviation of the deviations, and draw a histogram of the deviations; dynamically compensate the formula for adjusting the heat dissipation power based on the distribution of the deviations; If the mean deviation is greater than 0, it means that the predicted temperature is generally lower than the actual temperature, and the cooling power adjustment needs to be increased.

10. An artificial intelligence-based power equipment maintenance management system, the system is used to execute the artificial intelligence-based power equipment maintenance management method according to any one of claims 1 to 9, characterized in that: include: Data acquisition and synchronization module: collects multi-source data through sensors and maintains time synchronization; Data processing and noise reduction module: Based on the improved Kalman filter, it decomposes the sensor signal into layers, performs adaptive noise reduction, and extracts dynamic features; Thermal field modeling and analysis module: Generates a three-dimensional thermal field model to display the temperature gradient between power equipment and identify high-temperature concentration areas; Power equipment heat dissipation control module: This module builds a model based on environmental parameters to correct the heat transfer coefficient and convection heat dissipation efficiency in real time. It also analyzes temperature changes in historical data, analyzes temperature anomalies based on the average temperature change and maximum temperature within a period, and uses temperature thresholds to analyze temperature anomalies and adjust the heat dissipation power in a timely manner. Heat dissipation feedback compensation module: predicts the temperature of power equipment, minimizes heat dissipation efficiency, and predicts the temperature of power equipment after adjusting the heat dissipation efficiency. Analyzes and compares the deviation between the actual temperature and the adjusted predicted temperature, and dynamically compensates the formula for adjusting the heat dissipation power.

Images