A temperature monitoring method for an intelligent cable grounding box

Abstract

The present invention relates to the field of cable technology, and discloses a temperature monitoring method for an intelligent cable grounding box, comprising the following steps: obtaining material information of the cable grounding box; obtaining temperature and humidity data of the surrounding environment of the cable grounding box; determining the temperature monitoring accuracy requirements required for different types of grounding boxes according to the material properties of three types of cable grounding boxes; matching the temperature monitoring schemes that meet the temperature monitoring accuracy requirements of the three types of cable grounding boxes according to the environmental type and in combination with a preset temperature monitoring scheme database; obtaining historical temperature monitoring data of different types of cable grounding boxes; establishing a temperature prediction model for the cable grounding box in combination with the environmental temperature and humidity and the material of the grounding box; and continuously improving the temperature monitoring scheme and monitoring threshold through data accumulation and feedback according to the optimized temperature monitoring scheme. The present invention can realize accurate monitoring and early warning of different types of grounding boxes in various environments, and effectively improve the safe operation level of the cable grounding box.

Description

Technical Field

[0001] The invention relates to the technical field of cables, and in particular to a temperature monitoring method for an intelligent cable grounding box. Background Art

[0002] The core technical problem faced by cable grounding box temperature monitoring is how to develop an accurate temperature monitoring plan based on the heat dissipation characteristics and environmental conditions of grounding boxes of different materials. Due to the difference in materials, there are significant differences in the heat dissipation performance and temperature change laws of heavy metal, light metal and non-metal grounding boxes. At the same time, the temperature and humidity environmental conditions will further affect the temperature distribution of the grounding box. How to determine the appropriate monitoring frequency, monitoring point location and temperature threshold based on the material characteristics, environmental factors and cable load of the grounding box is the key to the design of the temperature monitoring plan. In addition, there are complex correlations between the historical temperature data of different types of grounding boxes and the environmental temperature and humidity data. How to discover these laws through data mining and use them to guide the optimization of the monitoring plan is also an important challenge. Ultimately, it is necessary to combine the static monitoring plan with the dynamic environmental changes and load changes to build an adaptive temperature monitoring model to achieve accurate identification and early warning of temperature anomalies of different types of grounding boxes. Summary of the invention

[0003] In order to solve the problems existing in the above-mentioned prior art, the purpose of the present invention is to provide a temperature monitoring method for an intelligent cable grounding box. The temperature monitoring method for an intelligent cable grounding box combines the material, environmental factors and historical data of the grounding box to construct an intelligent and personalized temperature monitoring solution, which can realize accurate monitoring and early warning of different types of grounding boxes in various environments, and effectively improve the safe operation level of the cable grounding box.

[0004] A temperature monitoring method for an intelligent cable grounding box according to the present invention comprises the following steps:

[0005] S1. Obtain material information of the cable grounding box, and classify the cable grounding box into three types: a heavy metal grounding box, a light metal grounding box, and a non-metal grounding box according to the material information;

[0006] S2. Acquire the temperature and humidity data of the surrounding environment of the cable grounding box to obtain the current environment type of the cable grounding box;

[0007] S3. Determine the temperature monitoring accuracy requirements for different types of grounding boxes according to the material properties of the three types of cable grounding boxes and in combination with a pre-established temperature monitoring accuracy requirement database;

[0008] S4. According to the environment type and in combination with a preset temperature monitoring solution database, respectively match the temperature monitoring solutions that meet the temperature monitoring accuracy requirements of the three types of cable grounding boxes;

[0009] S5. Obtain historical temperature monitoring data of different types of cable grounding boxes, obtain the daily temperature variation cycle of heat dissipation inside the cable grounding box and heating of the energized cable, and optimize the corresponding temperature monitoring scheme;

[0010] S6. Establish a temperature prediction model of the cable grounding box in combination with the ambient temperature and humidity and the material of the grounding box, and obtain temperature prediction results of the temperature prediction model under different materials and different ambient temperature and humidity inputs;

[0011] S7. Acquire the real-time temperature according to the optimized temperature monitoring scheme, and compare the real-time temperature with the temperature prediction result, continuously improve the temperature monitoring scheme and monitoring threshold through data accumulation and feedback, and form an adaptive grounding box temperature monitoring closed loop.

[0012] Preferably, the step S1 specifically includes:

[0013] The material information of the cable grounding box is obtained, including stainless steel, aluminum alloy and plastic;

[0014] Acquire the reflection spectrum data of the surface of the cable grounding box, perform curve difference calculation based on the reflection spectrum data and a pre-established standard spectrum database, obtain the material type and write it into the grounding box classification database;

[0015] According to the material type, the iron content and the nickel content in the material composition are determined to obtain a material composition measurement result;

[0016] Quantitatively measure the aluminum content according to the material composition measurement result to obtain a quantitative measurement result;

[0017] The IP protection grade value of the cable grounding box is obtained, and the cable grounding box is divided into the heavy metal grounding box, the light metal grounding box and the non-metal grounding box in combination with the material type, the material composition measurement result and the quantitative measurement result.

[0018] Preferably, the step S2 specifically includes:

[0019] Acquiring temperature and humidity data of the environment around the cable grounding box, and processing the temperature and humidity data through a digital bandpass filter and a Kalman filter to obtain temperature and humidity processed data;

[0020] According to the temperature and humidity processing data, the temperature and humidity values ​​within a preset time period in the past are extracted from the temperature and humidity database, and the temperature and humidity values ​​are plotted through a scatter plot to obtain a temperature curve and a humidity curve;

[0021] For the temperature curve and the humidity curve, the temperature and humidity change rate is calculated by using the slope, and the mutation point is marked to obtain the temperature and humidity values ​​of the mutation point;

[0022] The temperature and humidity values ​​and the mutation point temperature and humidity values ​​are classified into environmental categories by using the nearest neighbor clustering based on Euclidean distance, and the current environmental type of the cable grounding box is determined by combining with a pre-established environmental type database.

[0023] Preferably, the step S3 specifically includes:

[0024] According to the material properties of the heavy metal grounding box, the light metal grounding box and the non-metal grounding box, the thermal conductivity of the three types of grounding boxes is respectively obtained from the temperature monitoring accuracy database, and the thermal conductivity difference level of the three types of grounding boxes is calculated by percentage difference;

[0025] The monitoring frequency value, the monitoring point location coordinate value, the temperature sensor type value and the temperature acquisition accuracy value are obtained through the temperature monitoring accuracy database, and the monitoring parameter group is obtained according to the monitoring dimension grouping;

[0026] The monitoring parameter groups are classified by using a distance-based K-means clustering tool, and the temperature monitoring accuracy requirements for different types of grounding boxes are selected through monitoring point location coordinate planning, monitoring time interval division and sensor type.

[0027] Preferably, the step S4 specifically includes:

[0028] According to the environment type, in combination with a preset temperature monitoring solution database, respectively matching monitoring solutions that meet the temperature monitoring accuracy requirements of the heavy metal grounding box, the light metal grounding box, and the non-metal grounding box;

[0029] Get the temperature anomaly threshold, temperature warning threshold, temperature data storage format, and temperature data upload period in the solution, and upload the data to the log table according to the upload period.

[0030] Preferably, the step S5 specifically includes:

[0031] Obtain the historical temperature monitoring data of different types of cable grounding boxes, extract the internal temperature distribution and temperature change rate characteristics of the cable grounding box in the data, and obtain the daily temperature change cycle of the internal heat dissipation of the cable grounding box and the heating of the energized cable by analyzing the association rules between the historical data and the ambient temperature and humidity data, and optimize the corresponding temperature monitoring plan.

[0032] Preferably, the step S6 specifically includes:

[0033] A temperature prediction model of the cable grounding box combined with the ambient temperature and humidity and the material of the grounding box is established to obtain the temperature prediction results of the temperature prediction model under different materials and different ambient temperature and humidity inputs, and dynamically adjust the monitoring frequency and temperature anomaly threshold in the temperature monitoring scheme according to the temperature prediction results.

[0034] Preferably, the step S7 specifically includes:

[0035] Acquire the real-time temperature according to the optimized temperature monitoring scheme, and compare the real-time temperature with the temperature prediction result to obtain a temperature comparison result;

[0036] According to the temperature comparison result, combined with the monitoring frequency, monitoring point location, temperature sensor type and temperature acquisition accuracy parameters in the temperature monitoring scheme, and the temperature prediction model, the temperature monitoring device of the cable grounding box is deployed and debugged, and wireless upload is performed according to the data upload cycle. The temperature monitoring scheme and monitoring threshold are continuously improved through data accumulation and feedback to form an adaptive grounding box temperature monitoring closed loop.

[0037] The temperature monitoring method of an intelligent cable grounding box described in the present invention has the advantages that:

[0038] The temperature monitoring method of an intelligent cable grounding box of the present invention can accurately match the most suitable temperature monitoring scheme according to the material (heavy metal type, light metal type, non-metal type) of the cable grounding box and the temperature and humidity data of the surrounding environment; by analyzing the historical temperature monitoring data, the daily temperature change cycle law of the heat dissipation inside the cable grounding box and the heating of the energized cable can be understood, and then the temperature monitoring scheme is optimized; the established temperature prediction model can predict the temperature change of the grounding box according to the input different materials and environmental temperature and humidity data, and the maintenance personnel can take measures in advance according to the predicted temperature change to prevent equipment damage or safety problems that may be caused by excessive temperature; the optimized monitoring scheme and prediction model can ensure that the grounding box can maintain a stable operating state in various environments, thereby improving the overall reliability and stability of the system; through the intelligent monitoring and early warning system, not only the work efficiency is improved, but also the risk of human operation errors is reduced. The present invention combines the material, environmental factors and historical data of the grounding box to construct an intelligent and personalized temperature monitoring scheme, which can realize accurate monitoring and early warning of different types of grounding boxes in various environments, and effectively improve the safe operation level of the cable grounding box. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 The present invention discloses a flow chart of a temperature monitoring method for an intelligent cable grounding box. DETAILED DESCRIPTION

[0040] like Figure 1As shown, a temperature monitoring method for an intelligent cable grounding box according to the present invention comprises the following steps:

[0041] S1. Obtain material information of the cable grounding box, and classify the cable grounding box into three types: heavy metal grounding box, light metal grounding box and non-metal grounding box according to the material information;

[0042] S2. Obtain temperature and humidity data of the environment surrounding the cable grounding box to obtain the current environment type of the cable grounding box;

[0043] S3. According to the material properties of the three types of cable grounding boxes and the pre-established temperature monitoring accuracy requirement database, determine the temperature monitoring accuracy requirements for different types of grounding boxes;

[0044] S4. According to the environment type and in combination with the preset temperature monitoring solution database, temperature monitoring solutions that meet the temperature monitoring accuracy requirements of the three types of cable grounding boxes are matched respectively;

[0045] S5. Obtain historical temperature monitoring data of different types of cable grounding boxes, obtain the daily temperature variation cycle of heat dissipation inside the cable grounding box and heating of the energized cable, and optimize the corresponding temperature monitoring scheme;

[0046] S6. Establish a temperature prediction model for the cable grounding box in combination with the ambient temperature and humidity and the material of the grounding box, and obtain temperature prediction results of the temperature prediction model under different materials and different ambient temperature and humidity inputs;

[0047] S7. Obtain the real-time temperature according to the optimized temperature monitoring scheme, and compare the real-time temperature with the temperature prediction result. Through data accumulation and feedback, continuously improve the temperature monitoring scheme and monitoring threshold, and form an adaptive grounding box temperature monitoring closed loop.

[0048] Furthermore, in this embodiment, step S1 specifically includes:

[0049] Get the material information of the cable grounding box including stainless steel, aluminum alloy and plastic;

[0050] Acquire the reflection spectrum data of the surface of the cable grounding box, calculate the curve difference between the reflection spectrum data and the pre-established standard spectrum database, obtain the material type and write it into the grounding box classification database; acquire the reflection spectrum data of the surface of the cable grounding box through the material recognition sensor;

[0051] According to the material type, the iron content and nickel content in the material composition are determined to obtain the material composition measurement results; the iron content and nickel content in the material composition are determined by X-ray fluorescence spectrometer;

[0052] The aluminum content is quantitatively measured according to the material composition measurement results to obtain quantitative measurement results;

[0053] Obtain the IP protection grade value of the cable grounding box, and divide the cable grounding box into heavy metal grounding box, light metal grounding box and non-metal grounding box in combination with the material type, material composition measurement results and quantitative measurement results; obtain the IP protection grade value of the cable grounding box through a protection grade detector;

[0054] Specifically, a first preset threshold, a second preset threshold, a third preset threshold, a preset interval of iron content, and a preset IP value interval are set respectively;

[0055] If the iron content is higher than the first preset threshold and the nickel content is within the preset range, it is determined to be a heavy metal type grounding box and recorded in the grounding box classification database;

[0056] If the aluminum content is higher than the second preset threshold, it is determined to be a light metal grounding box and written into the grounding box classification database;

[0057] If the IP protection grade value is within the preset IP value range and the carbon content measured by the X-ray fluorescence spectrometer is higher than the third preset threshold, it is determined to be a non-metallic grounding box and written into the grounding box classification database;

[0058] Here is an example:

[0059] Set the first preset threshold value to 70%, the second preset threshold value to 50%, the third preset threshold value to 60%, the preset interval of iron content to [8%, 20%], and the preset IP value interval to [IP54, IP67];

[0060] The material identification sensor obtains the surface reflection spectrum data of the cable grounding box, and uses the Euclidean distance calculation tool to compare the standard stainless steel spectrum curve, standard aluminum alloy spectrum curve, and standard plastic spectrum curve in the spectrum database, calculates the spectrum curve difference, and determines the corresponding material type according to the curve difference;

[0061] The material type result is output to the grounding box classification database. Combined with the acquired material type of the grounding box, an X-ray fluorescence spectrometer is used to determine the iron and nickel content in the material composition. When the iron content is greater than 70% and the nickel content is between 8% and 20%, it is determined to be a heavy metal grounding box, and the measurement result is recorded in the grounding box classification database;

[0062] For the material composition measurement results, the element content percentage statistical tool is used to quantitatively measure the aluminum content. When the aluminum content is greater than 50%, it is judged as a light metal type grounding box, and the corresponding measurement results are written into the grounding box classification database;

[0063] The IP protection level value of the grounding box is obtained through a protection level detector. When the IP value measured by the protection level detector is within the range of IP54 to IP67, and the carbon content measured by the X-ray fluorescence spectrometer is higher than 60%, it is determined to be a non-metallic grounding box, and the corresponding measurement results are written into the grounding box classification database;

[0064] The material identification of the cable grounding box involves a variety of spectral detection principles. The reflection spectrum is measured using a near-infrared spectrometer with a wavelength range of 200-2500nm. The reflection spectrum curve of the stainless steel material has a characteristic absorption peak at a wavelength of 1064nm, the aluminum alloy has a typical reflection characteristic at a wavelength of 850nm, and the plastic material has a unique transmission absorption at a wavelength of 2100nm. The material is identified by the spectral curve morphology at these characteristic wavelengths.

[0065] The principle of X-ray fluorescence spectroscopy is to use the unique fluorescence wavelengths produced by different elements when excited by X-rays. The excitation wavelength of chromium in 304 stainless steel is 5.41keV, nickel has a characteristic peak at 7.47keV, and iron has the strongest characteristic peak at 6.40keV. The intensity ratio of these characteristic peaks is measured to determine the heavy metal type of grounding box. The typical composition ratio of 304 stainless steel is 18% chromium, 8% nickel, and 74% iron.

[0066] The determination of light metal grounding box adopts the aluminum content determination. The aluminum content in aluminum alloy 6063 is above 97%. The characteristic peak of aluminum in X-ray fluorescence spectrum is located at 1.49keV. The aluminum content is determined by measuring the peak intensity at this position. It is mainly used in outdoor distribution box shells and has the characteristics of light weight and good corrosion resistance.

[0067] Non-metallic grounding boxes are mostly made of engineering plastics. The common polycarbonate material has a carbon content of 63%. X-ray fluorescence spectrum analysis shows that the characteristic peak of carbon is at 0.28keV. The protection level standard stipulates that outdoor distribution boxes must reach IP65. The first digit 6 indicates the dustproof level, and the second digit 5 ​​indicates the waterproof level. The protection level detector is used to perform water spray and dust tests to determine the protection performance;

[0068] When the material type determination result is written into the classification database, the test number, test time, spectrum data, element content data, and protection level data are recorded. The database table structure includes a test information table to record basic information, a detection data table to store specific measurement values, and a classification result table to record the determination type. The three data tables are associated through the test number. When the measurement data is abnormal, an alarm is triggered to prompt that the measurement data exceeds the preset threshold range.

[0069] The material identification process of the grounding box is fully automated to avoid errors caused by manual judgment. The classification accuracy is improved through standardized testing methods. The test data is traceable throughout the process, providing data support for subsequent grounding box quality management, establishing complete material testing records, and realizing a complete testing chain from material identification, element analysis to protective performance testing.

[0070] Furthermore, in this embodiment, step S2 specifically includes:

[0071] The temperature and humidity data of the environment around the cable grounding box are obtained, and the temperature and humidity data are processed by a digital bandpass filter and a Kalman filter to obtain temperature and humidity processed data;

[0072] According to the temperature and humidity processing data, the temperature and humidity values ​​within the past preset time period are extracted from the temperature and humidity database, and the temperature and humidity values ​​are plotted through a scatter plot to obtain a temperature curve and a humidity curve;

[0073] For the temperature curve and humidity curve, the slope is used to calculate the temperature and humidity change rate, and the mutation point is marked to obtain the temperature and humidity values ​​at the mutation point;

[0074] The temperature and humidity values ​​and the temperature and humidity values ​​at mutation points are classified into environmental categories by using the nearest neighbor clustering based on Euclidean distance. Combined with the pre-established environmental type database, the current environmental type of the cable grounding box is determined.

[0075] Here is an example:

[0076] Temperature and humidity sensor collection points are arranged at intervals of 1 meter around the grounding box, and the temperature and humidity values ​​are obtained at a sampling frequency of every 10 seconds. A digital bandpass filter is used to filter out the noise with a frequency higher than 1 Hz in the original temperature and humidity data, and then the filtered temperature and humidity data are smoothed through a Kalman filter. The processed temperature and humidity values ​​and the corresponding sampling time points are recorded in the temperature and humidity database;

[0077] Extract the processed temperature and humidity values ​​in the past 24 hours from the temperature and humidity database, segment the temperature and humidity data points according to 10-minute time intervals, draw a scatter plot of the temperature and humidity data points, connect the adjacent temperature data points and humidity data points with straight lines to generate temperature curves and humidity curves, and the vertical coordinates of the two curves are in degrees Celsius and relative humidity percentage respectively;

[0078] For the temperature and humidity curve data, the slope calculation tool is used to count the slope of each curve segment. If the temperature change rate exceeds 5 degrees Celsius per hour or the humidity change rate exceeds 10% per hour, the environmental mutation point is marked on the curve graph, and the temperature and humidity values ​​of all mutation points are stored separately as environmental discrimination conditions;

[0079] According to the current temperature and humidity values ​​and mutation point records, the environment is classified using the nearest neighbor clustering tool based on Euclidean distance. According to the standard intervals in the environment type database, a temperature less than 10 degrees Celsius is considered a cold environment, a temperature between 10 and 35 degrees Celsius and a relative humidity of 20% to 80% is considered a normal temperature environment, a temperature greater than 35 degrees Celsius is considered a high temperature environment, and a relative humidity greater than 80% is considered a humid environment, the environment type of the grounding box is obtained;

[0080] The temperature and humidity monitoring process first involves the principle of sensor layout. When arranging sensors around the grounding box, the principle of equal spacing is adopted to avoid blind spots in measurement. The sensor installation height is flush with the middle of the grounding box. The sampling frequency is set to obtain data once every 10 seconds to meet the needs of environmental change detection without causing data redundancy. The sensor measurement accuracy is ±0.3 degrees Celsius for temperature and ±2% relative humidity for humidity.

[0081] The data processing uses a dual filtering mechanism. The cutoff frequency of the digital bandpass filter is set at 1 Hz to filter out high-frequency noise caused by equipment vibration and electromagnetic interference, and retain effective signals generated by environmental changes. The Kalman filter recursively estimates the temperature and humidity data. The filter gain coefficient is dynamically adjusted with the number of measurements. The gain coefficient decreases when the measurement noise is large, and increases when the measurement noise is small. The temperature and humidity curves are drawn using a segmented straight line connection method. The horizontal axis spans 24 hours, the vertical axis temperature ranges from 0 to 50 degrees Celsius, and the humidity ranges from 0 to 100%. A data point is taken every 10 minutes to draw a curve. The temperature curve and humidity curve are distinguished by different colors, and the specific values ​​are marked in the graph. The mutation points are highlighted with special marks;

[0082] The environment type is determined based on the temperature and humidity thresholds. It is triggered when the cold environment temperature is below 10 degrees Celsius. At this time, the metal shell of the grounding box is prone to condensation and the internal components work in a harsh environment. The normal temperature range is between 10 and 35 degrees Celsius and the relative humidity range is between 20% and 80%, which is suitable for normal operation of the equipment. The high temperature environment temperature exceeds 35 degrees Celsius. At this time, the heat dissipation of the equipment is limited. The relative humidity in a humid environment exceeds 80%, which is easy to cause the insulation performance to deteriorate.

[0083] The temperature and humidity data analysis uses the nearest neighbor clustering based on Euclidean distance to calculate the distance between the current measurement point and the standard environment type sample point. The type with the closest distance is the current environment type. Taking into account the impact of environmental mutation points, if a mutation point occurs within 24 hours, the mutation point weight is increased during the judgment to reflect the impact of drastic changes in the environment on the judgment results.

[0084] The environmental monitoring database contains a raw data table to record sensor data, a processed data table to store filtered data, a mutation point table to record abnormal change points, and an environmental type table to store judgment results. Each table is associated through a timestamp to achieve data traceability. The monitoring data retention period is set to 3 months, and overdue data is automatically archived to facilitate long-term trend analysis;

[0085] Temperature and humidity monitoring is fully automated, sensors are arranged in fixed positions to avoid human interference, data processing adopts standardized processes, environmental judgment standards are clear, measurement results are objective and reliable, and monitoring records are complete, realizing a complete monitoring chain from data acquisition, signal processing to environmental judgment.

[0086] Furthermore, in this embodiment, step S3 specifically includes:

[0087] According to the material characteristics of heavy metal grounding boxes, light metal grounding boxes and non-metal grounding boxes, the thermal conductivity of the three types of grounding boxes is obtained from the temperature monitoring accuracy database, and the difference level of thermal conductivity of the three types of grounding boxes is calculated by percentage difference;

[0088] The monitoring frequency value, monitoring point location coordinate value, temperature sensor type value and temperature acquisition accuracy value are obtained through the temperature monitoring accuracy database, and the monitoring parameter group is obtained according to the monitoring dimension grouping;

[0089] The monitoring parameter groups are classified by using the distance-based K-means clustering tool, and the temperature monitoring accuracy requirements for different types of grounding boxes are selected through the planning of monitoring point location coordinates, monitoring time interval division and sensor type;

[0090] If the monitoring point spacing in the monitoring layout plan is less than the specified threshold or the monitoring frequency exceeds the sensor response time limit, the monitoring layout plan is revised and recorded in the plan database after verification by the plan evaluation tool;

[0091] Here is an example:

[0092] The thermal conductivity of grounding boxes of different materials is read from the temperature monitoring accuracy database. The unit of thermal conductivity of heavy metal grounding boxes is watt per meter Kelvin, the unit of thermal conductivity of light metal grounding boxes is watt per meter Kelvin, and the unit of thermal conductivity of non-metal grounding boxes is watt per meter Kelvin. The three thermal conductivities are compared using the percentage difference calculation tool. When the difference of thermal conductivity of adjacent materials exceeds 50%, it is judged as a high difference level. When the difference is between 20% and 50%, it is judged as a medium difference level. When the difference is less than 20%, it is judged as a low difference level.

[0093] According to the difference level of thermal conductivity of materials, a parameter mapping tool based on a decision tree is used to construct the correspondence between material characteristics and monitoring parameters. The monitoring frequency value, monitoring point location coordinate value, temperature sensor type value, and temperature acquisition accuracy value corresponding to the high difference level are obtained from the monitoring accuracy database. The parameters are grouped according to the four dimensions of monitoring frequency, high, medium, and low, monitoring point location, sensor sensitivity, and acquisition accuracy, and the corresponding monitoring parameter group is generated.

[0094] For the generated monitoring parameter group, the distance-based K-means clustering tool is used to classify the monitoring parameters, a rectangular coordinate system is established with the center of the grounding box as the origin, the monitoring point locations are arranged and planned, the monitoring time intervals are divided according to the monitoring frequency intervals, thermocouples or thermistor sensors are selected according to the measurement temperature range, and the data acquisition resolution is selected according to the acquisition accuracy requirements, and three types of grounding box monitoring layout schemes are obtained;

[0095] Extract monitoring parameters from the generated monitoring layout plan. When the monitoring point spacing is less than the specified threshold or the monitoring frequency exceeds the sensor response time limit, revise the monitoring layout plan. Use the plan evaluation tool to test the revised monitoring layout plan. If it meets the temperature monitoring accuracy requirements, it will be recorded in the plan database as the standard monitoring plan for this type of grounding box.

[0096] The thermal conductivity of the grounding box material directly affects the temperature monitoring accuracy requirements. The thermal conductivity of stainless steel is 15 watts per meter Kelvin, the thermal conductivity of aluminum alloy is 200 watts per meter Kelvin, and the thermal conductivity of plastic is 0.2 watts per meter Kelvin. The thermal conductivity of the three materials is significantly different. When calculated by percentage difference, the difference between aluminum alloy and stainless steel is 92%, and the difference between stainless steel and plastic is 98%, both of which belong to the high difference level;

[0097] The parameter mapping process builds the relationship between materials and monitoring parameters based on the decision tree. Aluminum alloy with high thermal conductivity transfers heat quickly and evenly. The monitoring frequency can be set to a 5-minute interval. The monitoring points can be arranged at the four corners of the grounding box surface. A thermocouple sensor with a response time of 10 seconds is selected, and the acquisition accuracy is controlled at 0.5 degrees Celsius.

[0098] Stainless steel, a medium-conductivity material, has a moderate heat transfer rate. The monitoring frequency is set to a 2-minute interval. Nine monitoring points need to be evenly arranged on the surface. A thermocouple sensor with a response time of 5 seconds is selected, and the acquisition accuracy is controlled at 0.3 degrees Celsius.

[0099] Plastics with low thermal conductivity transfer heat slowly and unevenly. The monitoring frequency needs to be set to a 1-minute interval. 16 monitoring points should be arranged on the surface to form a dense monitoring network. Thermistor sensors with a response time of 3 seconds should be selected, and the acquisition accuracy should be controlled at 0.1 degrees Celsius to accurately reflect the temperature distribution.

[0100] The K-means clustering tool groups monitoring parameters and combines similar monitoring frequencies, measurement point densities, sensor types, and acquisition accuracy to facilitate subsequent plan generation. When generating monitoring layout plans, actual limiting factors should be considered. Too small a distance between monitoring points will cause interference, and the distance should be kept at least 5 cm. Too high a frequency of acquisition may exceed the sensor response capability, and the acquisition interval should be greater than the sensor response time.

[0101] The scheme evaluation adopts temperature field simulation verification to simulate the temperature distribution under typical working conditions and verify whether the monitoring layout scheme meets the temperature measurement accuracy requirements. The standard monitoring schemes of three types of grounding boxes are stored in the scheme database. The database table structure includes material type, thermal conductivity level, monitoring parameters, layout drawings, and verification result fields;

[0102] The temperature monitoring scheme design process adopts a hierarchical management mode. The monitoring level is determined according to the thermal conductivity of the material, and then the corresponding parameter combination is selected from the monitoring parameter library. The feasibility of the scheme is ensured through simulation verification.

[0103] The monitoring data is collected and stored in real time, supporting historical query and trend analysis. The distribution of monitoring points, sampling frequency, and sensor selection form a standardized monitoring plan to achieve accurate and reliable temperature monitoring.

[0104] Furthermore, in this embodiment, step S4 specifically includes:

[0105] According to the environment type, combined with the preset temperature monitoring solution database, match the temperature monitoring solutions that meet the temperature monitoring accuracy requirements of heavy metal grounding boxes, light metal grounding boxes and non-metal grounding boxes respectively;

[0106] Obtain the temperature anomaly threshold, temperature warning threshold, temperature data storage format and temperature data upload cycle in the solution, and upload the data to the log table according to the upload cycle;

[0107] Specifically, the environment type number in the temperature and humidity environment judgment table is obtained, and the cosine similarity calculation tool is used to perform feature matching between the environment type number and the applicable environment vector of the scheme in the monitoring scheme database. If the similarity is greater than a preset threshold, the corresponding monitoring scheme is obtained;

[0108] The temperature anomaly threshold is modified according to the monitoring scheme and the ambient temperature fluctuation amplitude. The temperature anomaly threshold decreases as the ambient relative humidity increases.

[0109] A data encoding tool is used to generate a fixed-length data packet for the data collected by the monitoring scheme, and the data packet has a measurement timestamp, a monitoring point number, a temperature value and a data status code;

[0110] Set the upload trigger condition according to the temperature fluctuation rule in the data packet. The upload trigger condition generates the upload time interval. Through the scheduled task, the data is uploaded to the log table according to the upload time interval.

[0111] Here is an example:

[0112] Obtain the current environment type number from the temperature and humidity environment judgment table, use the cosine similarity calculation tool to perform feature matching on each plan record in the monitoring plan database, calculate the similarity between the environmental feature vector and the applicable environment vector of the plan, and select the corresponding monitoring plan when the similarity is greater than 0.9. Modify the monitoring plan parameters according to the thermal conductivity of the grounding box material, and write the modified monitoring plan into the temperature monitoring record table;

[0113] Calculate the temperature fluctuation amplitude and humidity change rate according to the temperature and humidity environmental characteristic curve, select the corresponding temperature threshold in the preset threshold data table based on the environmental parameters, and use the temperature correction coefficient to adjust the abnormal threshold and warning threshold. When the ambient temperature fluctuation amplitude is greater than 5 degrees Celsius, the abnormal threshold increases the temperature fluctuation compensation value, and the warning threshold decreases with the increase of ambient relative humidity. Record the corrected temperature threshold in the monitoring parameter table;

[0114] The collected data is processed according to the data format specification of the monitoring plan. The data contains four fields: measurement timestamp, monitoring point number, temperature value, and data status code. A data encoding tool is used to generate a fixed-length data packet, and a cyclic redundancy check code is added to prevent transmission errors. The data packet is compressed and packaged and stored in the temperature data table in time sequence. The storage interval is automatically adjusted according to the frequency of temperature and humidity fluctuations.

[0115] Extract data upload configuration parameters from the monitoring plan, set upload trigger conditions according to the ambient temperature fluctuation rules, shorten the upload interval when the ambient temperature fluctuates violently, and extend the upload cycle when the temperature is stable. Execute data upload through scheduled tasks, record data transmission status in the upload log table, and automatically retransmit failed upload data;

[0116] Temperature and humidity environment judgment plays a key role in matching the monitoring plan. The low temperature cold environment temperature is below minus 10 degrees Celsius, the humid environment relative humidity exceeds 85%, the high temperature environment temperature exceeds 40 degrees Celsius, and the normal temperature environment temperature ranges from 10 to 35 degrees Celsius;

[0117] The environmental feature vector includes four dimensions: temperature mean, temperature fluctuation amplitude, humidity mean, and humidity change rate. The cosine value of the feature vector angle is calculated using cosine similarity. A cosine value of 1 indicates a complete match of the environmental features.

[0118] The suitability verification of the monitoring scheme is based on the thermal conductivity of the material. The heavy metal grounding box has high thermal conductivity and fast temperature response. The monitoring frequency is increased to once per minute in a high temperature environment. The light metal type has medium thermal conductivity and the monitoring frequency is set to once every 5 minutes in a normal temperature environment. The non-metal type has slow thermal conductivity and the monitoring frequency is reduced to once every 15 minutes in a low temperature environment to ensure that temperature changes can be captured in a timely manner.

[0119] Temperature threshold correction involves multiple environmental factors. In humid coastal environments, when the relative humidity is 85%, the warning threshold is 5 degrees Celsius lower than the standard value, and the abnormal threshold is 8 degrees Celsius lower to avoid the humid and hot environment accelerating equipment aging. In areas with large temperature differences between day and night, when the daily temperature fluctuation exceeds 20 degrees Celsius, the abnormal threshold increases the fluctuation compensation value by 5 degrees Celsius to adapt to the working conditions of drastic temperature fluctuations.

[0120] The data storage format uses a compact structure. The measurement timestamp occupies 4 bytes to store Unix time, the monitoring point number occupies 1 byte to represent 256 measurement points, the temperature value occupies 2 bytes to one decimal place, and the data status code occupies 1 byte to identify the data validity. A 2-byte cyclic redundancy check code is added to the data packet, and the check value is calculated by polynomial division. The receiver repeats the check process to find transmission errors;

[0121] The data upload strategy is adaptively adjusted as the environment changes. When the temperature is stable, the batch upload mode is adopted, and the data is uploaded once an hour. When the temperature fluctuates violently, it switches to the real-time upload mode and uploads the latest data once a minute.

[0122] After the upload fails, the retransmission is delayed according to the exponential backoff algorithm. The first delay is 1 minute, and the delay time is doubled thereafter. A maximum of 3 retries are allowed.

[0123] The monitoring data table adopts a partition storage structure, divides the data into partitions according to the time dimension, and generates new partitions every month to facilitate the archiving of historical data. When querying, the target partition is directly located to improve the retrieval speed;

[0124] The data compression uses a run-length encoding algorithm, and the temperature measurement points with the same value are combined and stored. The compression ratio is between 3:1 and 5:1, which reduces the storage space occupied;

[0125] The whole process of monitoring scheme is automated, from environmental feature extraction, scheme matching, parameter correction to data collection and storage, forming a closed-loop control;

[0126] The monitoring parameters are dynamically adjusted under different environmental conditions, the temperature threshold is corrected in real time as the environment changes, and the data collection and upload are adaptively adjusted according to the temperature fluctuation law to ensure that the monitoring data is accurate and reliable.

[0127] Furthermore, in this embodiment, step S5 specifically includes:

[0128] Obtain the historical temperature monitoring data of different types of cable grounding boxes, extract the internal temperature distribution and temperature change rate characteristics of the cable grounding boxes in the data, and obtain the daily temperature change cycle of the heat dissipation inside the cable grounding box and the heating of the energized cable by analyzing the association rules between the historical data and the ambient temperature and humidity data, and optimize the corresponding temperature monitoring plan;

[0129] Specifically, the radial basis kernel function is used to perform spatial interpolation operation on the temperature data of the monitoring points to obtain the temperature distribution thermodynamic map of the cable grounding box;

[0130] According to the temperature distribution thermodynamic diagram, the temperature gradient field value and the time temperature change rate value are obtained through central difference calculation;

[0131] A threshold is determined for the time-temperature change rate value. If the temperature change rate exceeds the specified threshold and the duration exceeds the specified interval, it is marked as a power-on heating interval. If the temperature change rate is lower than the specified threshold, it is marked as an ambient temperature influence interval.

[0132] For the temperature data of the power-on heating interval and the ambient temperature influence interval, long short-term memory network training is used to obtain the heat dissipation characteristic curves of heavy metal grounding boxes, light metal grounding boxes and non-metal grounding boxes;

[0133] Here is an example:

[0134] The temperature monitoring data of different types of grounding boxes were extracted from the temperature history database. The radial basis kernel function with a bandwidth parameter of 0.5 was used to perform spatial interpolation on the temperature distribution of the monitoring points to generate a thermal map of the internal temperature distribution of the grounding box. The spatial temperature gradient field and the time temperature change rate were calculated using the central difference calculation tool. The relationship curve between the temperature gradient value and the distance from the grounding box surface was recorded to obtain the temperature spatial distribution characteristics.

[0135] A cubic polynomial fitting model was established based on the temperature spatial distribution characteristics, and the historical temperature data was matched and analyzed with the ambient temperature and humidity data. Within a 24-hour time window, when the temperature change rate exceeded 2 degrees Celsius per hour and lasted for more than 4 hours, it was marked as a temperature change interval caused by power-on heating, and when the temperature change was less than 1 degree Celsius per hour, it was marked as an ambient temperature influence interval.

[0136] For the temperature data in the heating interval and the environmental impact interval, the long short-term memory network is used to extract the temperature time series characteristics. The network input layer sets three characteristic dimensions: temperature, time, and environmental parameters. The number of hidden layer neurons is set to 128. The heat dissipation characteristic curves of the heavy metal grounding box, the light metal grounding box, and the non-metal grounding box are obtained through training.

[0137] Based on the heat dissipation characteristic curve and daily temperature fluctuation law, the monitoring scheme is optimized. The density of monitoring points is increased in areas with drastic temperature changes, and the number of monitoring points is reduced in areas with stable temperature. The sampling frequency is dynamically adjusted with the temperature change rate, and the alarm threshold is set with differentiated values ​​according to the heat dissipation performance of the material. The optimized parameters are written into the temperature monitoring parameter table.

[0138] The temperature monitoring of the cable grounding box includes the dual characteristics of spatial distribution and time series change. The radial basis kernel function interpolation adopts the Gaussian kernel form. The bandwidth parameter 0.5 strikes a balance between ensuring smoothness and local characteristics. The farther away from the monitoring point, the more exponentially the influence weight decays. The generated heat map shows the temperature distribution form that spreads outward from the monitoring point.

[0139] When calculating the temperature gradient by central difference, the distance between two adjacent points is taken as 0.1 meters, and the spatial gradient field is obtained to reflect the diffusion direction and intensity of temperature in three-dimensional space;

[0140] The cubic polynomial fitting in historical data analysis can reflect the nonlinear change trend of temperature over time. The fitting curve includes the influence of ambient temperature and the superposition effect of equipment heating. Different influencing factors are distinguished by the temperature change rate. Under heating conditions, the temperature rise rate exceeds 2 degrees Celsius per hour. Under environmental influences, the temperature rise is slow, and the temperature difference between day and night is within 15 degrees Celsius.

[0141] The heavy metal type grounding box has good thermal conductivity, high temperature field uniformity, and a spatial gradient of less than 0.5 degrees Celsius per centimeter. The light metal type is second, and the non-metal type has uneven temperature distribution, with a local temperature difference of up to 5 degrees Celsius.

[0142] The LSTM network training uses a week of temperature data. The input features include 24-hour temperature series, timestamp encoding, and ambient temperature and humidity parameters. The network captures long-term dependencies through a gating mechanism and extracts periodic temperature change patterns.

[0143] The heavy metal grounding box has fast heat dissipation and a steep temperature rise curve, reaching a steady state in half an hour, while the light metal type needs 1 hour, and the non-metal type has slow heat dissipation and the temperature rise lasts for more than 2 hours;

[0144] The monitoring scheme is optimized to adjust parameters according to the characteristics of different materials. For heavy metal grounding boxes, measuring points are arranged at the cable joints, with a spacing of 0.2 meters and a sampling interval of 5 minutes. For light metal grounding boxes, measuring points are arranged at the four corners and center of the box, with a spacing of 0.3 meters and a sampling interval of 10 minutes. For non-metallic grounding boxes, the density of measuring points is increased, with a spacing of 0.15 meters and a sampling interval of 2 minutes.

[0145] The temperature alarm threshold setting varies with the heat dissipation performance of the material. The heavy metal type is set to 10 degrees Celsius above the rated temperature, the light metal type is set to 8 degrees Celsius above the rated temperature, and the non-metal type is set to 5 degrees Celsius above the rated temperature.

[0146] The database records and saves information such as the original temperature value, the processed temperature distribution diagram, the change trend curve, the monitoring parameter configuration, etc., and establishes indexes according to the time and space dimensions to support fast retrieval and statistical analysis;

[0147] The entire monitoring process forms a closed loop, from data collection, feature extraction to law analysis, and ultimately guides the optimization of the monitoring plan to achieve accurate and reliable temperature monitoring;

[0148] The measurement point layout and sampling frequency are dynamically adjusted according to the temperature change law to avoid wasting monitoring resources and not missing key data.

[0149] Furthermore, in this embodiment, step S6 specifically includes:

[0150] Establish a temperature prediction model for the cable grounding box combined with the ambient temperature and humidity and the material of the grounding box, obtain the temperature prediction results of the temperature prediction model under different materials and different ambient temperature and humidity inputs, and dynamically adjust the monitoring frequency and temperature anomaly threshold in the temperature monitoring scheme according to the temperature prediction results;

[0151] Specifically, obtain the ambient temperature data, ambient humidity data and material type data in the temperature monitoring database, obtain the standardized feature value through the z-score standardization tool, and obtain the material coding value through one-hot encoding;

[0152] A gradient boosting tree prediction model is established based on the standardized feature values ​​and material coding values. The gradient boosting tree prediction model includes a temperature and humidity product term, a temperature and material thermal conductivity product term, and a humidity and material thermal conductivity product term.

[0153] The environmental parameter data in the real-time monitoring database is input into the gradient boosting tree prediction model to determine whether the predicted temperature value exceeds the preset multiple standard deviation range. If it exceeds the range of two times the standard deviation, the abnormal point data is recorded;

[0154] The monitoring interval is determined by predicting the temperature change rate. If the temperature rise rate exceeds the specified threshold, the monitoring interval is set to five minutes. If the temperature fluctuation is below the specified threshold, the monitoring interval is set to thirty minutes.

[0155] Here is an example:

[0156] The ambient temperature, ambient humidity, and material type are extracted from the temperature monitoring database as input features. The temperature and humidity data are normalized using the z-score standardization tool. The material type is converted using one-hot encoding. The material grades are divided into different thermal conductivity intervals. The standardized feature data and the measured temperature values ​​form a training data set, and the data statistical parameters are recorded in the feature processing parameter table.

[0157] For the standardized data, a temperature prediction model was established using a gradient boosting tree. The input features included the product of temperature and humidity, the product of temperature and material thermal conductivity, and the product of humidity and material thermal conductivity. The feature importance score was calculated based on information gain. The prediction model was trained separately at three prediction time scales: short-term, medium-term, and long-term, and the temperature predictor for the corresponding time scale was obtained.

[0158] Obtain current environmental parameters and material parameters from the real-time monitoring database, input different time scale predictors to calculate future temperature values, and use the temperature change trend verification tool to determine the deviation between the predicted temperature and the historical trend. When the predicted temperature confidence interval exceeds the range of two standard deviations, or the temperature change trend is reversed, mark the predicted result as an abnormal point, and record the comparison data between the predicted result and the measured result;

[0159] The monitoring plan was adjusted according to the predicted temperature change trend. When the temperature rise rate was greater than 2 degrees Celsius, the monitoring frequency was increased to 5 minutes per hour. When the temperature fluctuation was less than 1 degree Celsius, the monitoring frequency was reduced to 30 minutes per hour. Differentiated temperature thresholds were set based on the thermal conductivity of the material. The temperature thresholds of heavy metal grounding boxes and light metal grounding boxes were 5 degrees Celsius higher than the standard value, and the temperature thresholds of non-metal grounding boxes were 5 degrees Celsius lower than the standard value. The adjusted monitoring parameters were recorded.

[0160] The temperature prediction of the cable grounding box includes feature preprocessing and model construction. Feature preprocessing uses z-score standardization to convert temperature and humidity data into a standard normal distribution with a mean of 0 and a standard deviation of 1 to eliminate the dimension effect.

[0161] The material type uses One-Hot Encoding to convert heavy metal, light metal, and non-metal types into three-dimensional vectors, and the thermal conductivity is divided into levels of 200 watts per meter Kelvin, 50 watts per meter Kelvin, and 0.5 watts per meter Kelvin;

[0162] The construction of feature cross terms takes into account the interaction between temperature, humidity and material. The product of temperature and humidity reflects the comprehensive environmental impact, the product of temperature and thermal conductivity reflects the heat dissipation capacity, and the product of humidity and thermal conductivity represents the moisture-proof performance. The gradient boosting tree model uses a decision tree with a depth of 6 as the base learner, the number of iterations is set to 100, the learning rate is set to 0.1, and the optimal parameters are determined through cross-validation.

[0163] The prediction time scale division follows the law of temperature change. The short-term prediction range is within 1 hour to capture sudden temperature rise. The medium-term prediction range is 24 hours to reflect the diurnal temperature cycle. The long-term prediction range is one week to reflect the seasonal change trend. The confidence interval of the prediction result adopts two standard deviations, and the confidence level reaches 95%.

[0164] The temperature forecast anomaly judgment combines statistical characteristics and physical laws. When the predicted temperature exceeds twice the standard deviation of the historical distribution, or violates the law of heat conduction and an unreasonable temperature jump occurs, the forecast result will be marked as abnormal. The predicted result is compared with the measured result and the root mean square error is used for evaluation. The short-term forecast error is controlled within 0.5 degrees Celsius, the medium-term forecast error is less than 2 degrees Celsius, and the long-term forecast error is less than 5 degrees Celsius.

[0165] The monitoring scheme is dynamically adjusted based on the predicted temperature change rate. Under severe temperature rise conditions, 5-minute sampling intervals can capture temperature anomalies in a timely manner, and under stable conditions, 30-minute sampling intervals can reduce data redundancy.

[0166] The thermal conductivity of the material affects the temperature threshold setting. Heavy metal and light metal types dissipate heat quickly and have large temperature fluctuations, so the threshold is increased by 5 degrees Celsius. Non-metal types dissipate heat slowly and have obvious temperature accumulation, so the threshold is decreased by 5 degrees Celsius.

[0167] The prediction results are stored in a time series database, which records the prediction time point, predicted value, measured value, upper and lower limits of the confidence interval, and establishes a comparison between the prediction sequence and the measured sequence to support prediction accuracy evaluation and model optimization.

[0168] The entire prediction process forms a complete closed loop, from data preprocessing, feature construction, model training to application of prediction results, to achieve accurate prediction and effective monitoring of grounding box temperature changes.

[0169] Furthermore, in this embodiment, step S7 specifically includes:

[0170] The real-time temperature is obtained according to the optimized temperature monitoring scheme, and the real-time temperature is compared with the temperature prediction result to obtain the temperature comparison result;

[0171] According to the temperature comparison results, combined with the monitoring frequency, monitoring point location, temperature sensor type and temperature acquisition accuracy parameters in the temperature monitoring plan, as well as the temperature prediction model, the temperature monitoring device of the cable grounding box is deployed and debugged, and wireless data upload is performed according to the data upload cycle. The temperature monitoring plan and monitoring threshold are continuously improved through data accumulation and feedback, forming an adaptive grounding box temperature monitoring closed loop;

[0172] Specifically, the monitoring point layout coordinates and sensor type parameters in the temperature monitoring plan are obtained, signal repeaters are arranged at the monitoring points according to the signal strength test values, and the acquisition parameters are set using the sensor range calibration values;

[0173] Acquire the temperature data of the monitoring points according to the acquisition parameters, remove abnormal values ​​through the data verification tool, and use the run length encoding compression tool to generate compressed data;

[0174] For compressed data, the online Bayesian algorithm is used to update the temperature prediction model parameters, and the monitoring frequency correction coefficient is determined according to the deviation between the measured data and the predicted data;

[0175] For the prediction model, the prediction root mean square error is calculated through prediction accuracy evaluation. If the prediction root mean square error exceeds the preset threshold, the prediction model is retrained using a sliding time window and updated to the monitoring device through the model deployment tool;

[0176] Here is an example:

[0177] Read the monitoring point layout coordinates, sensor type, and sampling frequency parameters from the temperature monitoring plan, use a signal strength tester to test the wireless transmission channel, deploy a signal repeater when the signal strength is lower than minus 70 decibels, perform range calibration and accuracy inspection on the temperature sensor, set the acquisition parameters according to the standard of 0.1 degrees Celsius for heavy metal grounding boxes, 0.2 degrees Celsius for light metal types, and 0.5 degrees Celsius for non-metal types, and record the deployment parameters of the monitoring device;

[0178] Collect raw temperature data from deployed monitoring devices, use data verification tools to check sensor data point by point, remove values ​​that exceed the upper and lower limits of the temperature range and single-point mutations exceeding 10 degrees Celsius, use run-length encoding compression tools to losslessly compress temperature data, with a compression ratio of 5 to 1, upload data in cycles of 2 minutes for heavy metals, 5 minutes for light metals, and 10 minutes for non-metals, and simultaneously record equipment self-test status data;

[0179] For the accumulated temperature data in the monitoring database, the online Bayesian algorithm is used to update the temperature prediction model parameters. The learning rate is set to 0.01. The monitoring frequency correction coefficient is calculated based on the deviation between the measured data and the predicted data. When the temperature fluctuation amplitude exceeds 2 times the preset threshold, the sampling frequency is increased. When the fluctuation amplitude is less than 0.5 times the preset threshold, the sampling frequency is reduced. The updated parameters are sent to the monitoring device through the remote parameter configuration tool.

[0180] The updated model is verified through the prediction accuracy evaluation tool. When the prediction root mean square error exceeds 1 degree Celsius during the continuous monitoring period, the model retraining is triggered. The new data of the last month is added to the training set using a sliding time window, the prediction model parameters and monitoring thresholds are updated, the update time and version number are recorded in the version database, and the new version model is synchronized to the monitoring device through the model deployment tool;

[0181] Wireless signal testing is crucial during the deployment of monitoring devices. Complex on-site environments lead to signal attenuation. Signal strength in open areas attenuates with the square of the distance. Building shielding causes additional loss. During testing, a spectrum analyzer is used to scan signal strength at monitoring points and record changes in signal strength at different times. -70 decibels is the lowest working level for wireless data transmission. If the transmission bit error rate is lower than this value, it will rise sharply, and repeaters need to be added for signal compensation.

[0182] The sensor calibration adopts the piecewise linear calibration method. Calibration points are set every 10 degrees in the range of 0 to 100 degrees Celsius. The deviation between the sensor output value and the standard value is recorded to generate a calibration curve. The temperature of the metal grounding box changes quickly. The platinum resistance temperature sensor is used to achieve an accuracy of 0.1 degrees. The light metal type uses a thermocouple sensor with an accuracy of 0.2 degrees. The non-metal type uses a thermistor sensor with an accuracy of 0.5 degrees to meet the temperature measurement requirements.

[0183] The data compression adopts the run-length encoding algorithm. The temperature value is converted to an integer after retaining the decimal places according to the accuracy requirements. The continuous repeated values ​​are replaced by the count value. In actual monitoring, the temperature changes slowly, the adjacent data points change little, and the compression effect is significant.

[0184] The temperature response of heavy metal grounding boxes is fast, and the upload time is 2 minutes; that of light metal boxes is 5 minutes; that of non-metal boxes is 10 minutes. Different materials use differentiated sampling strategies, and the Bayesian algorithm is used to guide parameter updates through prior information. The learning rate of 0.01 ensures that the model converges stably and can track temperature changes in a timely manner.

[0185] If the temperature prediction deviation exceeds 2 times the threshold, it means that the environment fluctuates violently and the sampling frequency needs to be increased. If the deviation is less than 0.5 times the threshold, it means that the temperature is stable and the sampling frequency can be reduced to save resources. The remote parameter configuration adopts encrypted transmission to ensure the safe update of the monitoring device parameters.

[0186] The prediction model is evaluated using the root mean square error indicator. If the error of the continuous monitoring period exceeds 1 degree Celsius, the model retraining is triggered, indicating that the existing model cannot accurately describe the temperature change pattern;

[0187] A one-month time window is used to retain sufficient training samples and reflect the latest change trends. Version management records model update time, trigger reasons, and performance indicators to achieve model traceability. The monitoring device obtains the latest model parameters through a secure tunnel.

[0188] The monitoring device is adaptively adjusted to form a complete closed loop, with each link working closely together from data collection, compression and uploading to model prediction and parameter optimization;

[0189] Different material grounding boxes adopt differentiated monitoring strategies. High-frequency sampling of heavy metal and light metal types captures rapid changes, and low-frequency sampling of non-metallic types reduces data redundancy, achieving accurate and efficient temperature monitoring. The model is continuously optimized to ensure prediction accuracy and adapt to environmental and load changes.

[0190] In the description of the present invention, it is necessary to understand that the directions or positional relationships indicated by directional words such as "front, back, up, down, left, right", "lateral, vertical, horizontal" and "top, bottom" are usually based on the directions or positional relationships shown in the drawings. They are only for the convenience of describing the present invention and simplifying the description. Unless otherwise specified, these directional words do not indicate or imply that the device or element referred to must have a specific direction or be constructed and operated in a specific direction. Therefore, they cannot be understood as limiting the scope of protection of the present invention.

[0191] For those skilled in the art, various other corresponding changes and deformations can be made according to the technical solutions and concepts described above, and all of these changes and deformations should fall within the protection scope of the claims of the present invention.

Claims

1. A temperature monitoring method for an intelligent cable grounding box, characterized in that: The following steps are involved: S1. Obtain material information of the cable grounding box, and classify the cable grounding box into three types: a heavy metal grounding box, a light metal grounding box, and a non-metal grounding box according to the material information; S2. Acquire the temperature and humidity data of the surrounding environment of the cable grounding box to obtain the current environment type of the cable grounding box; S3. Determine the temperature monitoring accuracy requirements for different types of grounding boxes according to the material properties of the three types of cable grounding boxes and in combination with a pre-established temperature monitoring accuracy requirement database; S4. According to the environment type and in combination with a preset temperature monitoring solution database, respectively match the temperature monitoring solutions that meet the temperature monitoring accuracy requirements of the three types of cable grounding boxes; S5. Obtain historical temperature monitoring data of different types of cable grounding boxes, obtain the daily temperature variation cycle of heat dissipation inside the cable grounding box and heating of the energized cable, and optimize the corresponding temperature monitoring scheme; S6. Establish a temperature prediction model of the cable grounding box in combination with the ambient temperature and humidity and the material of the grounding box, and obtain temperature prediction results of the temperature prediction model under different materials and different ambient temperature and humidity inputs; S7. Acquire the real-time temperature according to the optimized temperature monitoring scheme, and compare the real-time temperature with the temperature prediction result, continuously improve the temperature monitoring scheme and monitoring threshold through data accumulation and feedback, and form an adaptive grounding box temperature monitoring closed loop; The step S1 specifically includes: The material information of the cable grounding box is obtained, including stainless steel, aluminum alloy and plastic; Acquire the reflection spectrum data of the surface of the cable grounding box, perform curve difference calculation based on the reflection spectrum data and a pre-established standard spectrum database, obtain the material type and write it into the grounding box classification database; According to the material type, the iron content and the nickel content in the material composition are determined to obtain a material composition measurement result; Quantitatively measure the aluminum content according to the material composition measurement result to obtain a quantitative measurement result; The IP protection grade value of the cable grounding box is obtained, and the cable grounding box is divided into the heavy metal grounding box, the light metal grounding box and the non-metal grounding box in combination with the material type, the material composition measurement result and the quantitative measurement result.

2. The temperature monitoring method of the intelligent cable grounding box according to claim 1 is characterized in that: The step S2 specifically includes: Acquiring temperature and humidity data of the environment around the cable grounding box, and processing the temperature and humidity data through a digital bandpass filter and a Kalman filter to obtain temperature and humidity processed data; According to the temperature and humidity processing data, the temperature and humidity values ​​within a preset time period in the past are extracted from the temperature and humidity database, and the temperature and humidity values ​​are plotted through a scatter plot to obtain a temperature curve and a humidity curve; For the temperature curve and the humidity curve, the temperature and humidity change rate is calculated by using the slope, and the mutation point is marked to obtain the temperature and humidity values ​​of the mutation point; The temperature and humidity values ​​and the mutation point temperature and humidity values ​​are classified into environmental categories by using the nearest neighbor clustering based on Euclidean distance, and the current environmental type of the cable grounding box is determined by combining with a pre-established environmental type database.

3. The temperature monitoring method of the intelligent cable grounding box according to claim 1 is characterized in that: The step S3 specifically includes: According to the material properties of the heavy metal grounding box, the light metal grounding box and the non-metal grounding box, the thermal conductivity of the three types of grounding boxes is respectively obtained from the temperature monitoring accuracy database, and the thermal conductivity difference level of the three types of grounding boxes is calculated by percentage difference; The monitoring frequency value, the monitoring point location coordinate value, the temperature sensor type value and the temperature acquisition accuracy value are obtained through the temperature monitoring accuracy database, and the monitoring parameter group is obtained according to the monitoring dimension grouping; The monitoring parameter groups are classified by using a distance-based K-means clustering tool, and the temperature monitoring accuracy requirements for different types of grounding boxes are selected through monitoring point location coordinate planning, monitoring time interval division and sensor type.

4. The temperature monitoring method of the intelligent cable grounding box according to claim 1 is characterized in that: The step S4 specifically includes: According to the environment type, in combination with a preset temperature monitoring solution database, respectively matching temperature monitoring solutions that meet the temperature monitoring accuracy requirements of the heavy metal grounding box, the light metal grounding box, and the non-metal grounding box; Get the temperature anomaly threshold, temperature warning threshold, temperature data storage format, and temperature data upload period in the solution, and upload the data to the log table according to the upload period.

5. The temperature monitoring method of the intelligent cable grounding box according to claim 1 is characterized in that: The step S5 specifically includes: Obtain the historical temperature monitoring data of different types of cable grounding boxes, extract the internal temperature distribution and temperature change rate characteristics of the cable grounding box in the data, and obtain the daily temperature change cycle of the internal heat dissipation of the cable grounding box and the heating of the energized cable by analyzing the association rules between the historical data and the ambient temperature and humidity data, and optimize the corresponding temperature monitoring plan.

6. The temperature monitoring method of the intelligent cable grounding box according to claim 1 is characterized in that: The step S6 specifically includes: A temperature prediction model of the cable grounding box combined with the ambient temperature and humidity and the material of the grounding box is established to obtain the temperature prediction results of the temperature prediction model under different materials and different ambient temperature and humidity inputs, and dynamically adjust the monitoring frequency and temperature anomaly threshold in the temperature monitoring scheme according to the temperature prediction results.

7. The temperature monitoring method of the intelligent cable grounding box according to claim 1 is characterized in that: The step S7 specifically includes: Acquire the real-time temperature according to the optimized temperature monitoring scheme, and compare the real-time temperature with the temperature prediction result to obtain a temperature comparison result; According to the temperature comparison result, combined with the monitoring frequency, monitoring point location, temperature sensor type and temperature acquisition accuracy parameters in the temperature monitoring scheme, and the temperature prediction model, the temperature monitoring device of the cable grounding box is deployed and debugged, and wireless upload is performed according to the data upload cycle. The temperature monitoring scheme and monitoring threshold are continuously improved through data accumulation and feedback to form an adaptive grounding box temperature monitoring closed loop.

Images