An intrinsically safe monitoring system and method for branch cables in combiner boxes

Abstract

This application discloses an intrinsically safe monitoring system and method for branch cables in a combiner box, comprising: a temperature sensor network with multiple temperature sensors, each used to detect cable temperature data of at least one branch cable in the combiner box; a distributed data acquisition device with multiple data acquisition devices for collecting cable temperature data of each branch cable in a preset area; a data processing module for preprocessing the cable temperature data; a temperature prediction module for generating corresponding temperature prediction results; an alarm module for generating corresponding alarm information; and a navigation module for generating corresponding navigation information. By using temperature sensors to detect potential faults in the early stages of heating, the timeliness of fault detection is improved.

Description

Technical Field

[0001] This application relates to the Internet field, specifically to an intrinsically safe monitoring system and method for branch cables in combiner boxes. Background Technology

[0002] Combiner box branch cables are cables used in power systems such as photovoltaic power generation systems to connect the combiner box to various photovoltaic modules or other power generation units. Most faults during use (such as poor contact at crimping points, insulation aging, overload, short circuit, etc.) are often accompanied by heat generation. Severe overheating can even cause serious accidents such as fires.

[0003] However, in practical applications, combiner boxes often contain a large number of branch cables, especially some large power plants that contain multiple combiner boxes, which further increases the number of branch cables and makes supervision more difficult.

[0004] Traditional monitoring methods typically involve using cameras to capture images of the junction box and analyzing any abnormal phenomena such as sparks or burning to determine if the junction box is malfunctioning.

[0005] However, sparks and combustion usually only occur after the junction box has been continuously heated for a certain period of time. By this time, the junction box has already malfunctioned and has been malfunctioning for a long time. This makes this detection method lagging and difficult to detect junction box malfunctions in a timely manner. Summary of the Invention

[0006] To address the aforementioned issues, this application proposes an intrinsically safe monitoring system for branch cables in combiner boxes, comprising: a temperature sensor network, a distributed data acquisition device, and a monitoring center;

[0007] The temperature sensor network is equipped with multiple temperature sensors, which are used to detect the cable temperature data of each branch cable in the combiner box at least.

[0008] The distributed data acquisition device is equipped with multiple data acquisition devices, which are connected to the temperature sensor network and are used to collect cable temperature data of each branch cable in the preset area.

[0009] The monitoring center includes a data processing module, a temperature prediction module, an alarm module, and a navigation module;

[0010] The data processing module receives the cable temperature data fed back by the distributed data acquisition device and preprocesses the cable temperature data.

[0011] The temperature prediction module makes predictions based on the cable temperature data and generates corresponding temperature prediction results.

[0012] The alarm module determines anomalies based on the cable temperature data and the temperature prediction results, and generates corresponding alarm information.

[0013] The navigation module generates corresponding navigation information based on the abnormal target location information contained in the alarm information and the location information of the personnel receiving the alarm information.

[0014] In one example, the temperature sensor network is provided with multiple sub-networks, each sub-network corresponding to a single preset area;

[0015] Each sub-network contains at least one first temperature sensor and multiple second temperature sensors;

[0016] The distance between the first temperature sensor and the junction box is lower than a first preset distance, which is used to detect the ambient temperature data corresponding to the junction box;

[0017] For the same branch cable, the second temperature sensor is set at different positions on the branch cable to detect the cable temperature data corresponding to the combiner box.

[0018] In one example, the temperature sensor network determines the connection location between each branch cable and the junction box for each branch cable.

[0019] A second temperature sensor is set within a second preset distance of the connection position to collect the temperature data of the first cable, and at least one second temperature sensor is set outside a third preset distance of the connection position to collect the temperature data of the second cable at its own location.

[0020] The data processing module preprocesses and categorizes the cable temperature data in each preset area;

[0021] Based on the classification results, for a single branch cable, determine the number of branches of the branch cable currently being detected simultaneously by the second temperature sensor corresponding to the temperature data of the first cable, and determine the number of historical anomalies at the location corresponding to each second cable temperature data.

[0022] When the alarm module makes an anomaly determination based on the temperature data, it adjusts the first anomaly determination threshold of the first cable temperature data according to the number of branches, and adjusts the second anomaly determination threshold of the second cable temperature data according to the number of historical anomalies.

[0023] The more branches there are, the higher the first anomaly determination threshold becomes; the more historical anomalies there are, the lower the second anomaly determination threshold becomes.

[0024] In one example, the temperature prediction module generates a corresponding temperature prediction result for each cable temperature data.

[0025] The alarm module performs anomaly determination on each cable temperature data according to the temperature prediction result;

[0026] Identify the third temperature sensor corresponding to the abnormal cable temperature data, and identify other temperature sensors within the influence range of the third temperature sensor.

[0027] The abnormal target location information corresponding to the third temperature sensor is used as the primary alarm target, and the abnormal target location information corresponding to other temperature sensors within the influence range is used as the secondary alarm target, and corresponding alarm information is generated.

[0028] Specifically, for the second temperature sensor installed within the second preset distance of the connection location, its influence range includes the junction box where it is located and one or more corresponding branch cables; for the second temperature sensor installed outside the third preset distance of the connection location, its influence range includes one or more corresponding branch cables.

[0029] In one example, the temperature prediction module generates a temperature prediction result corresponding to the second temperature sensor by using a pre-trained regression model based on the cable temperature data and the ambient temperature data.

[0030] In one example, the training process of the regression model includes:

[0031] Based on the ambient temperature data, a corresponding time period is generated;

[0032] Within the time period, the corresponding cable temperature data is collected by the second temperature sensor to form a corresponding time series, and the time series is processed to improve its stability.

[0033] Autocorrelation analysis and partial autocorrelation analysis are performed on the time series, and a regression model with a matching architecture is selected based on the analysis results; wherein, the architecture of the regression model includes: autoregressive model, moving average model, autoregressive moving average model, and autoregressive integral moving average model;

[0034] Based on the selected regression model, the model parameters are estimated according to the analysis results, and the model is fitted based on the estimated model parameters.

[0035] The fitting results are evaluated, and the regression model is optimized based on the evaluation results.

[0036] In one example, the temperature prediction module identifies the upstream device connected to the branch cable and determines the device type and status of the upstream device.

[0037] The temperature prediction result of the first cable temperature data is corrected according to the equipment type and the equipment status;

[0038] The correction process includes:

[0039] Based on the equipment type, select power generation equipment from the upstream equipment, and determine the number of power generation equipment and the power generation type of the power generation equipment;

[0040] If the number of devices is a single device, then based on the first preset coefficient corresponding to the power generation type and the difference between the device status and the preset normal status, the first correction coefficient corresponding to the power generation device is determined, and the temperature prediction result of the first cable temperature data is corrected by the first correction coefficient.

[0041] If the number of devices is at least two, then for each power generation device, a corresponding first correction coefficient is determined;

[0042] If there are multiple power generation types among all power generation equipment, then for each power generation equipment, the first correction coefficient of the power generation equipment is adjusted according to the number of power generation types to obtain the second correction coefficient, and the temperature prediction result of the first cable temperature data is corrected by the second correction coefficient.

[0043] The second correction coefficient has a higher degree of correction than the first correction coefficient.

[0044] In one example, the navigation module abstracts the map of the current region into a spatial map;

[0045] Generate corresponding nodes and edges in the spatial graph; wherein, the nodes include primary nodes and secondary nodes, which correspond to the primary alarm target and the secondary alarm target, respectively, and the edges correspond to roads in the current region;

[0046] Based on the personnel location information, corresponding personnel nodes are generated in the spatial diagram;

[0047] For each node, calculate the shortest spatial distance between that node and other nodes in the spatial graph;

[0048] Using the personnel nodes as the starting point and the main nodes as the ending point, a first navigation path is generated based on the corresponding shortest spatial distance.

[0049] Using the personnel nodes as the starting point and the main nodes as the ending point, and by arranging and combining the shortest spatial distances between the nodes, one or more secondary nodes are used as path nodes to generate their respective corresponding second navigation paths.

[0050] Among all the second navigation paths, the second navigation path whose spatial distance is greater than the spatial distance corresponding to the first navigation path by a fourth preset distance is deleted, and among the remaining second navigation paths, the second navigation path with the most path nodes is selected as the third navigation path;

[0051] Based on the first navigation path and the third navigation path, corresponding navigation information is generated.

[0052] In one example, the system also includes: a cloud platform;

[0053] The cloud platform is used to back up and record at least one of the following: cable temperature data, temperature prediction results, anomaly determination results, alarm information, and navigation information.

[0054] On the other hand, this application also proposes an intrinsic safety monitoring method for combiner box branch cables, which performs safety monitoring through an intrinsic safety monitoring system for combiner box branch cables as described in any of the above examples, the method comprising:

[0055] The temperature data of each branch cable in the combiner box is detected by multiple temperature sensors in the temperature sensor network.

[0056] The cable temperature data is collected by multiple data acquisition devices in the distributed data acquisition system and transmitted to the monitoring center.

[0057] The cable temperature data is preprocessed by the data processing module in the monitoring center, and then predicted by the temperature prediction module in the monitoring center based on the cable temperature data to generate the corresponding temperature prediction result.

[0058] The alarm module in the monitoring center determines anomalies based on the cable temperature data and the temperature prediction results, and generates corresponding alarm information.

[0059] The navigation module in the monitoring center generates corresponding navigation information based on the abnormal target location information contained in the alarm information and the location information of the personnel receiving the alarm information.

[0060] The intrinsic safety monitoring system for junction box branch cables proposed in this application can bring the following benefits:

[0061] Beneficial effects:

[0062] 1. Compared to detecting sparks or combustion to diagnose faults, this application uses a temperature sensor network to directly monitor the temperature of the branch cables in the combiner box in real time. The temperature sensors detect potential faults in the early stages of heating, improving the timeliness of fault detection. Furthermore, the temperature prediction module predicts the temperature change trend of the branch cables, further enabling the early detection of possible abnormal temperature increases, providing early warnings of potential faults, and notifying maintenance personnel in advance, thus reducing the probability of fault occurrence.

[0063] 2. The temperature sensor network is distributed across each branch cable, enabling precise identification of which branch cable is experiencing temperature anomalies, thus providing more accurate branch cable location. In the complex environment of large power plants with numerous branch cables and combiner boxes, this allows for rapid location of the faulty cable, reducing the scope and time of troubleshooting and improving maintenance efficiency.

[0064] 3. The navigation module optimizes the maintenance process. Once the location of the faulty cable is determined, the navigation module provides maintenance personnel with precise navigation information, helping them quickly reach the fault site. In large power plants or complex wiring environments, this effectively saves maintenance personnel time in locating the fault and improves the efficiency of the entire maintenance work.

[0065] 4. The distributed data acquisition device can effectively collect data from various temperature sensors and transmit it to the monitoring center. At the monitoring center, the data processing module can uniformly store, organize, and analyze this data, thereby enabling remote monitoring and management of the combiner box. Attached Figure Description

[0066] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0067] Figure 1 This is a schematic diagram of the architecture of the intrinsic safety monitoring system for branch cables in a combiner box, as described in this application.

[0068] Figure 2 This is a schematic diagram illustrating the training process of the regression model in an embodiment of this application;

[0069] Figure 3 This is a schematic diagram illustrating the generation of navigation paths in an embodiment of this application;

[0070] Figure 4 This is a flowchart illustrating the intrinsic safety monitoring method for branch cables in a combiner box, as described in this application. Detailed Implementation

[0071] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0072] The technical solutions provided by the various embodiments of this application are described in detail below with reference to the accompanying drawings.

[0073] like Figure 1 As shown in the figure, this application provides an intrinsically safe monitoring system for branch cables in a combiner box: a temperature sensor network, a distributed data acquisition device, and a monitoring center.

[0074] The temperature sensor network contains multiple temperature sensors. Depending on their location, these sensors can be used to detect different temperature data at different locations within the combiner box and branch cables, as well as in the surrounding environment. Of course, the temperature sensors are primarily used to detect the cable temperature data of each branch cable within the combiner box.

[0075] The temperature sensor network utilizes industrial-grade digital temperature sensors (such as DS18B20 or PT100). The sensor housings are made of high-temperature resistant and flame-retardant materials, achieving a protection rating of IP67 or higher. They are fixed to the measured part using snap-on clamps or thermally conductive tape. Each temperature sensor has a unique built-in ID code, following a three-level structure of "preset area number - device type - serial number." For example, A01-JH-001 represents sensor number 1 for environmental monitoring in area A01.

[0076] Specifically, in some large power plants, there are multiple combiner boxes, each containing multiple branch cables. Each branch cable also has a certain length. Therefore, multiple sub-networks are set up in the temperature sensor network. Each sub-network corresponds to a single preset area (for example, the area where each combiner box is located, or the area where multiple adjacent combiner boxes are located, or further division principles can be set, including: branch cables of the same electrical circuit are grouped into the same sub-network, adjacent combiner boxes with a distance of less than 5 meters are divided into the same sub-network, and each sub-network covers an area of ​​less than 100 square meters, etc.). Each sub-network is equipped with a Zigbee self-organizing network module. The network topology adopts a star + mesh hybrid structure. By meshing the temperature sensor network, it is easier to manage the temperature sensor network.

[0077] Each sub-network contains at least one first temperature sensor and multiple second temperature sensors. It should be noted that both the first and second temperature sensors are temperature sensors; the different descriptions are used here for ease of explanation.

[0078] The first temperature sensor is positioned at a distance less than a preset distance from the junction box (e.g., 30cm ± 5cm) to detect the ambient temperature data corresponding to the junction box. The sensor should be installed away from the heat dissipation channels of the junction box and at least 10cm from the nearest heat dissipation hole to ensure accurate detection. If a single preset area contains multiple junction boxes, the ambient temperature data detected by the first temperature sensor can be directly used as the ambient temperature data for each of these junction boxes.

[0079] Generally speaking, the temperature of the branch cables at the combiner box tends to reach higher temperatures. Therefore, it is only necessary to install a primary temperature sensor at the combiner box to detect the ambient temperature data. For other locations of the branch cables, the ambient temperature data can be directly used as their own ambient temperature data.

[0080] In addition, each branch cable has a certain length. Therefore, for the same branch cable, the second temperature sensor is set at different locations on the branch cable to detect the cable temperature data corresponding to the combiner box. The locations can be set at fixed intervals (for example, a fixed interval of about 10% of the cable length, with a minimum interval of more than 1 meter), or they can be set according to actual conditions (for example, in locations prone to failure).

[0081] Of course, if the distance between multiple branch cables is very close (for example, when the cable spacing is less than 15cm), then a second temperature sensor can be used to simultaneously detect the cable temperature data of multiple branch cables at their corresponding locations. In this case, a multi-probe temperature sensor can be used, with each probe corresponding to one cable, and the probe spacing can be adjusted.

[0082] Furthermore, when setting up the second temperature sensor, since the connection point between the branch cable and the combiner box is usually a faulty location and its temperature is likely to be higher, the connection point between the branch cable and the combiner box is determined for each branch cable in the temperature sensor network.

[0083] A second temperature sensor is installed within a second preset distance from the connection location (determined by the volume of the combiner box, for example, 20cm ± 5cm). (Contact temperature measurement can be used here, with the sensor probe directly contacting the copper conductor of the cable for more accurate temperature acquisition.) This sensor is used to collect the temperature data of the first cable. At least one second temperature sensor is installed beyond a third preset distance from the connection location (determined by the length of the branch cable, for example, at least 1 meter). (The sensor's location can be based on historically identified fault-prone locations or at fixed intervals.) This sensor is used to collect the temperature data of the second cable at its own location. (Non-contact infrared temperature measurement can be used here to detect the cable sheath temperature, facilitating position adjustments.)

[0084] The distributed data acquisition device is equipped with multiple data acquisition devices, which are connected to a temperature sensor network to collect cable temperature data of each branch cable in the preset area.

[0085] The data acquisition device features an adjustable sampling frequency of 1Hz to 1kHz, 16 differential input channels, a 24-bit ADC resolution, and communication interfaces including RS485, Ethernet, and 4G wireless. It operates at temperatures ranging from -40℃ to 85℃ and may also include a signal conditioning circuit. This circuit includes modules such as a low-pass filter (cutoff frequency 100Hz), a programmable amplifier (adjustable gain from 1 to 1000 times), and cold junction compensation (for thermocouples).

[0086] Typically, each preset area corresponds to one data acquisition device, used to collect cable temperature data from the temperature sensor in that area. However, in some larger preset areas, multiple data acquisition devices can be set up.

[0087] Each data acquisition device has multiple data acquisition channels and uses time-division multiplexing technology. Each channel supports polling and acquiring data from up to 256 sensors. The polling period is configurable. In this way, each data acquisition channel can be used to acquire data from a single temperature sensor within the same time period.

[0088] The monitoring center communicates with the data acquisition device to obtain the corresponding cable temperature data and perform the corresponding data processing.

[0089] Specifically, the monitoring center includes a data processing module, a temperature prediction module, an alarm module, and a navigation module.

[0090] The data processing module receives cable temperature data from the distributed data acquisition device and preprocesses the cable temperature data.

[0091] Specifically, the data processing module preprocesses and categorizes the cable temperature data in each preset area. Preprocessing can include conventional preprocessing (such as data cleaning and format conversion) and custom preprocessing (setting corresponding processing rules based on the actual scenario). Categorization refers to determining the preset area to which each temperature sensor belongs and whether the collected data is from the first or second cable. Data cleaning can include: outlier removal (using the 3σ criterion, marking three consecutive sampling points exceeding the mean ± three times the standard deviation as outliers), missing value compensation (using linear interpolation to compensate for fewer than five consecutive missing points), and noise filtering (using moving average filtering, adjustable in window width from 5 to 15 points).

[0092] The temperature prediction module makes predictions based on cable temperature data and generates corresponding temperature prediction results.

[0093] The temperature prediction module uses a pre-trained regression model to predict the temperature of the cable and the ambient temperature, generating the temperature prediction result corresponding to the second temperature sensor, so as to provide early warning of cable temperature.

[0094] Specifically, such as Figure 2 As shown, the training process of the regression model includes:

[0095] After the raw data is input, the corresponding time period is generated based on the ambient temperature data. This involves first performing seasonal analysis by plotting line graphs of the ambient temperature data or using seasonal decomposition methods to observe whether seasonal patterns exist. If a seasonal pattern exists, the corresponding time period is determined (usually using the seasonal cycle as the standard seasonal cycle; for example, seasonality with a yearly cycle has a 12-month cycle, seasonality with a weekly cycle has a 7-day cycle, etc.).

[0096] Within a time period, the corresponding cable temperature data is collected by a second temperature sensor to form a corresponding time series. The time series can collect and record corresponding temperature values ​​at preset intervals (typically ranging from a few seconds to tens of seconds). The series can be stored in the form of an arithmetic sequence or a line graph.

[0097] Stationarity processing is performed on the time series data. Generally, regression models require the data to be stationary. Therefore, the stationarity of the time series data is first tested, for example, using a unit root test (such as the ADF test). This test determines whether a unit root exists in the time series data. If a unit root exists, it indicates that the data is non-stationary.

[0098] For non-stationary time series, differencing can be used to make them stationary. First-order differencing typically eliminates the trend in the data by calculating the difference between adjacent time points. If the data is still not stationary after first-order differencing, second-order or higher-order differencing can be performed. Simultaneously, the autocorrelation function (ACF) and partial autocorrelation function (PACF) of the differrated data can be observed to determine whether the data's stationarity meets the requirements.

[0099] Then, autocorrelation and partial autocorrelation analyses can be performed on the time series, and a regression model with a matching architecture can be selected based on the analysis results. The architectures of the regression models include: autoregressive model, moving average model, and autoregressive moving average model.

[0100] Specifically, the autocorrelation function (ACF) reflects the correlation between a time series and its lagged values, while the partial autocorrelation function (PACF), after controlling for the influence of intermediate terms, measures the direct correlation between a time series and its lagged values.

[0101] Therefore, the regression model for the corresponding architecture can be selected by observing the graphs of ACF and PACF.

[0102] If the autocorrelation function (ACF) exhibits tailing, while the partial autocorrelation function (PACF) truncates after a certain lag order (e.g., p), an autoregressive model, AR(p), can be chosen. The AR(p) model assumes that the current value is a linear combination of the past p values ​​plus a white noise term. For example, the AR(1) model has the following form: Where is y t The temperature value at time t (which is the temperature value corresponding to the cable temperature data). It is the autoregressive coefficient, ∈ t It's white noise. Autoregressive models are suitable for describing situations where the current temperature value is mainly influenced by its past values. White noise is a stochastic process; a white noise sequence has zero mean, constant variance, and the values ​​in the sequence are independent of each other. If the model's residuals are white noise, it means the model has successfully extracted all useful information from the data, with no remaining predictable patterns. If the residuals are not white noise, for example, if the residuals exhibit autocorrelation, it indicates that the model may not have fully captured the dynamic patterns in the data, and there is still some information that can be used to further improve the model.

[0103] If the ACF truncates after a certain lag order (e.g., q), while the PACF exhibits a trailing pattern, a moving average model, MA(q), can be chosen. The MA(q) model assumes that the current value is a linear combination of the past q white noise terms plus a constant and the current white noise term. For example, the MA(1) model can be written as y t =μ+θ1∈t-1 +∈ t , where μ is a constant and θ1 is the moving average coefficient. The moving average model is suitable for situations where temperature changes are mainly affected by past random shocks (white noise).

[0104] If both ACF and PACF exhibit tailing, an autoregressive moving average (ARMA) model can be chosen for fitting. The ARMA model combines the characteristics of autoregression and moving average. The ARMA(p,q) model can be written as... It is suitable for complex situations where temperature changes are affected by both their own past values ​​and past random shocks.

[0105] If the original temperature data is non-stationary but becomes stationary after differencing, and the ACF and PACF of the differencing data exhibit the characteristics of the AR, MA, or ARMA models mentioned above, then the Autoregressive Integral Moving Average (ARIMA) model can be used. In the ARIMA(p,d,q) model, 'd' represents the order of differencing, used to handle non-stationary data so that the data conforms to the assumptions of the AR, MA, or ARMA models after differencing. For example, if the original data conforms to the characteristics of ARMA(p,q) after first-order differencing, then the corresponding ARIMA model is ARIMA(p,1,q). Generally, the ARIMA(p,d,q) model can be written as... Where B is the shift operator, defined by By t =y t-1 B 2 y t =y t-2 And so on. in, is the autoregressive coefficient, p is the autoregressive order, representing the autoregressive relationship between the current value and the past p values. (1-B) d It is the difference operator, where d is the order of the difference. θ(B) = 1 + θ1B + θ2B 2 +…+θ q B q It is a polynomial of the moving average part, θ1…θ q is the moving average coefficient, and q is the moving average order, representing the moving average relationship between the current value and the past q white noise terms. ∈ t It is a white noise sequence.

[0106] After selecting the appropriate regression model, the model parameters are estimated based on the analysis results, and then the model is fitted based on the estimated parameters.

[0107] Taking the autoregressive moving average model as an example, we can find the maximum lag order where the PACF and ACF values ​​are significantly non-zero based on the PACF and ACF plots, and initially determine the values ​​of p and q. Then, by adjusting the values ​​of p and q and trying different values, we can finally determine the magnitude of p and q by combining the model's fit and statistical indicators. The value of d can be determined through the previous differencing operation. After several orders of differencing, the data reaches a stationary state, and the value of d is taken as that number.

[0108] For regression models, after determining the initial values ​​of the model parameters, the model is fitted using methods such as least squares or maximum likelihood estimation. During the fitting process, the regression model estimates the parameter values ​​based on the given time series data, so that the regression model can best describe the changing patterns of the data.

[0109] The fitting results are evaluated, and the regression model is optimized based on the evaluation results.

[0110] Evaluation metrics may include: Mean Squared Error (MSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE). MSE measures the average of the squared errors between the predicted and actual values, MAE is the average of the absolute errors between the predicted and actual values, and MAPE is the average of the absolute percentage errors. The smaller these metric values ​​are, the better the model's predictive performance.

[0111] You can also analyze the residuals of the model to check if they conform to the characteristics of white noise, that is, whether the residuals have a mean of zero, a constant variance, and no autocorrelation. If the residuals do not meet the characteristics of white noise, it indicates that the model may have missed information or overfitting issues, and further model adjustments are needed.

[0112] When optimizing a model, if the model's prediction performance is unsatisfactory, you can try adjusting the values ​​of model parameters such as p, d, and q, refitting the model, or consider performing other transformations on the data (such as logarithmic transformation, Box-Cox transformation, etc.) to improve the data characteristics and enhance the model's fit. Alternatively, you can change the model architecture and retrain to obtain a regression model.

[0113] Taking the SARIMA model in actual implementation as an example, a periodic parameter s = 24 (corresponding to a 24-hour daily cycle) can be added, implemented using the Python statsmodels library. The model update strategy adopts a sliding window mechanism, retraining the model every 1000 new data points.

[0114] The alarm module determines anomalies based on cable temperature data and temperature prediction results, and generates corresponding alarm information. Specifically, if the cable temperature data or temperature prediction results exceed preset normal values, an anomaly is considered to exist, and an alarm should be generated. The alarm information can include the alarm target (which maintenance personnel should trigger the alarm, e.g., based on the day's shift, the person responsible for each combiner box, or the current distance between personnel), the alarm method (the selected method for triggering the alarm, e.g., social media message, SMS, telephone, email), and the alarm level (e.g., using a three-level alarm mechanism: Level 1 (yellow): temperature exceeds the threshold by 10%; Level 2 (orange): temperature exceeds the threshold by 20% or is predicted to exceed the threshold within one hour; Level 3 (red): temperature exceeds the threshold by 30% or is predicted to exceed the threshold within 30 minutes).

[0115] Furthermore, after the temperature prediction module generates a corresponding temperature prediction result for each cable temperature data, the alarm module performs anomaly judgment on each cable temperature data according to the temperature prediction result.

[0116] At this point, identify the third temperature sensor corresponding to the abnormal cable temperature data (that is, the temperature sensor corresponding to the abnormal cable temperature data), and identify other temperature sensors within the influence range of the third temperature sensor.

[0117] The scope of influence is preset. For example, for a second temperature sensor located within a second preset distance from the connection point, since it monitors the connection point, any impact will typically affect the entire branch cable. Therefore, the scope of influence includes the junction box and one or more corresponding branch cables (the number of corresponding branch cables depends on how many branch cables the second temperature sensor is monitoring simultaneously). For a second temperature sensor located beyond a third preset distance from the connection point, its scope of influence is more limited, affecting only its corresponding branch cable. In other words, its scope of influence includes one or more corresponding branch cables (again, depending on how many branch cables the second temperature sensor is monitoring simultaneously).

[0118] At this point, the abnormal target location information corresponding to the third temperature sensor can be considered as the location where an anomaly is likely to occur. This location is taken as the primary alarm target, and the abnormal target location information corresponding to other temperature sensors within the influence range is considered as the location that may be affected. This location is taken as the secondary alarm target, and corresponding alarm information is generated.

[0119] The navigation module generates corresponding navigation information based on the abnormal target location information contained in the alarm information and the location information of the personnel receiving the alarm information.

[0120] Specifically, when the location information of the personnel is outside the park where the junction box is located, the navigation information can be divided into two parts. The part outside the park can be navigated by integrating the corresponding third-party program, while the part inside the park is navigated by itself because the third-party map is often not comprehensive and detailed enough.

[0121] First, the map of the current area is abstracted into a spatial map. Generally speaking, a two-dimensional spatial map is sufficient for most parks, while for parks with complex elevation information (such as those with complex terrain undulations, or those where the junction box is built on a high-rise building), a corresponding three-dimensional spatial map is required.

[0122] Next, corresponding nodes and edges are generated in the spatial graph. Nodes include primary nodes and secondary nodes, corresponding to primary and secondary alarm targets respectively, while edges correspond to roads in the current region. Of course, if the alarm information only contains primary alarm targets, secondary nodes do not need to be generated.

[0123] Simultaneously, corresponding personnel nodes are generated in the spatial map based on the personnel location information. If the maintenance personnel enters from outside the park, the location of the entrance can be used as the personnel location information. If the maintenance personnel is already inside the park, their current location information can be obtained from the positioning device of their smart terminal and used as the personnel location information.

[0124] For each node, calculate the shortest spatial distance between that node and other nodes in the spatial graph. The shortest spatial distance is not a straight-line distance, but rather a distance that can be reached by connecting nodes with edges. We can take each node as the starting point and determine the shortest spatial distance by traversing all paths from that node to other nodes.

[0125] Then, using the personnel nodes as the starting point and the main nodes as the ending point, a first navigation path is generated based on the corresponding shortest spatial distance. The first navigation path includes the corresponding shortest spatial distance and the corresponding navigation route, and can also include recommended modes of transportation (including walking, cycling, driving, etc.) based on the terrain.

[0126] In addition to the first navigation path, in order to allow maintenance personnel to reach as many secondary nodes as possible on their way to the main node without wasting time, the personnel node can be used as the starting point and the main node as the destination. By arranging and combining the shortest spatial distances between the nodes, one or more secondary nodes can be used as waypoints to generate their respective second navigation paths.

[0127] For example, such as Figure 3As shown, let the current personnel node be node A, the primary node be node B, and there be two secondary nodes, C and D. Multiple second navigation paths are generated through permutations and combinations, where the distance between any two nodes in the second navigation path is the shortest spatial distance between those two nodes. Therefore, the second navigation paths include: ACB, ADB, ACDB, and ADCB.

[0128] In all secondary navigation paths, to prevent excessive distances from delaying maintenance personnel from reaching key nodes, those secondary navigation paths with spatial distances exceeding a fourth preset distance compared to the corresponding distance of the primary navigation path were deleted. Figure 3 For example, after calculation, if the distances of the routes ADCB and ACDB do not meet the requirements, they will be deleted.

[0129] Meanwhile, in the remaining second navigation path (at this time, Figure 3 In the examples remaining (ACB, ADB), to allow maintenance personnel to see as many secondary nodes as possible, facilitating quick assessment of primary nodes and enabling proactive maintenance of secondary nodes, the second navigation path with the most nodes can be selected as the third navigation path. Figure 3 In the example, since all the remaining second navigation paths have the same number of path nodes, the shortest path ACB can be chosen as the third navigation path.

[0130] According to the first navigation path (in) Figure 3 (represented by solid lines), third navigation path (in) Figure 3 (Represented by dashed lines), generating corresponding navigation information. The navigation information includes both navigation paths, allowing maintenance personnel to choose independently and determine, based on different actual situations, whether to go directly to the main node or first to the secondary node.

[0131] The intrinsic safety monitoring system for junction box branch cables proposed in this application can bring the following benefits:

[0132] Beneficial effects:

[0133] 1. Compared to detecting sparks or combustion to diagnose faults, this application uses a temperature sensor network to directly monitor the temperature of the branch cables in the combiner box in real time. The temperature sensors detect potential faults in the early stages of heating, improving the timeliness of fault detection. Furthermore, the temperature prediction module predicts the temperature change trend of the branch cables, further enabling the early detection of possible abnormal temperature increases, providing early warnings of potential faults, and notifying maintenance personnel in advance, thus reducing the probability of fault occurrence.

[0134] 2. The temperature sensor network is distributed across each branch cable, enabling precise identification of which branch cable is experiencing temperature anomalies, thus providing more accurate branch cable location. In the complex environment of large power plants with numerous branch cables and combiner boxes, this allows for rapid location of the faulty cable, reducing the scope and time of troubleshooting and improving maintenance efficiency.

[0135] 3. The navigation module optimizes the maintenance process. Once the location of the faulty cable is determined, the navigation module provides maintenance personnel with precise navigation information, helping them quickly reach the fault site. In large power plants or complex wiring environments, this effectively saves maintenance personnel time in locating the fault and improves the efficiency of the entire maintenance work.

[0136] 4. The distributed data acquisition device can effectively collect data from various temperature sensors and transmit it to the monitoring center. At the monitoring center, the data processing module can uniformly store, organize, and analyze this data, thereby enabling remote monitoring and management of the combiner box.

[0137] In one embodiment, the data processing module can also determine, based on the classification results, the number of branches of the branch cable currently being simultaneously detected by the second temperature sensor corresponding to the first cable temperature data for a single branch cable. For example, at the connection point in the combiner box, since the distance between multiple branch cables at the connection point is very close, a second temperature sensor is used to simultaneously detect the first cable temperature of four branch cables. In this case, the number of branches of the branch cable corresponding to the first cable temperature is four. Simultaneously, the module determines the historical anomaly count corresponding to each second cable temperature data location. This data can be obtained from the corresponding historical database; the historical anomaly counts (e.g., temperature anomalies) near the location corresponding to the second cable temperature data location can all be used as the corresponding historical anomaly count.

[0138] At this time, when the alarm module makes anomaly judgments based on the temperature data, it can adjust the first anomaly judgment threshold of the first cable temperature data according to the number of branches, and adjust the second anomaly judgment threshold of the second cable temperature data according to the number of historical anomalies.

[0139] The more branches there are, the more upstream devices they connect to. Even under normal operating conditions, the temperature at these connection points will be higher, thus requiring a higher first anomaly threshold and a more stringent judgment standard. Conversely, a higher number of historical anomalies indicates a greater probability of anomalies occurring at subsequent locations along the branch cable. Therefore, a lower second anomaly threshold is needed to more rigorously detect any anomalies at those locations.

[0140] In one embodiment, if the temperature prediction module can also consider the impact of upstream equipment when making temperature predictions, the accuracy of the temperature prediction results can be further improved.

[0141] Specifically, the temperature prediction module identifies the upstream devices connected to the branch cables and determines the device type and status. Each branch cable connects to one upstream device, and all branch cables in a combiner box may connect to the same or different upstream devices. Device types primarily include power generation equipment and energy storage equipment. For power generation equipment, more specific device types can be defined, such as wind power generation equipment and hydropower generation equipment. For power generation equipment, the status can be categorized based on load conditions, such as high load, normal load, and low load.

[0142] At this point, based on the equipment type and status, the temperature prediction result of the first cable temperature data (which corresponds to the cable temperature data collected by the second temperature sensor set within a second preset distance of the connection location) is corrected, thereby making the temperature prediction result more accurate. Only the temperature prediction result of the first cable temperature data is corrected here because cable temperature data near the connection location is more easily affected by upstream equipment.

[0143] The correction process includes:

[0144] First, select power generation equipment from upstream equipment based on the equipment type, and determine the number of power generation equipment and the type of power generation.

[0145] If the number of devices is single, it indicates that the current combiner box is connected to the same power generation device. In this case, based on the first preset coefficient corresponding to the power generation type and the difference between the device status and the preset normal state, a first correction coefficient corresponding to the power generation device is determined. For example, the ratio between the device status and the preset normal state is calculated; this ratio represents the difference between the power generation device and the normal state. Then, the first preset coefficient is used to reasonably reduce or increase this difference, thereby generating the first correction coefficient. The temperature prediction result of the first cable temperature data is corrected using the first correction coefficient. Generally speaking, the greater the difference between the device status and the preset normal state, the greater the upward correction of the temperature prediction result. If the device status has not yet reached the preset normal state, no correction is required.

[0146] If the number of devices is at least two, then for each power generation device, determine its corresponding first correction factor.

[0147] If there are multiple power generation types in all power generation equipment, there may be incompatibility between the power generation results of the multiple power generation types, which may cause more serious heat generation. Therefore, for each power generation equipment, the first correction coefficient of the power generation equipment is adjusted according to the number of power generation types (generally speaking, the more power generation types there are, the higher the degree of adjustment of the first correction coefficient) to obtain a second correction coefficient. At this time, the correction degree of the second correction coefficient obtained after adjustment is higher than that of the first correction coefficient, thus corresponding to the possible incompatibility and the situation that will aggravate heat generation.

[0148] The temperature prediction result of the first cable temperature data is corrected using a second correction coefficient. Since there are multiple power generation devices, each device has its own second correction coefficient. Assuming that the branch cable corresponding to each power generation device uses first cable temperature data collected by an independent temperature sensor, then each first cable temperature data point can be corrected independently. However, if the first cable temperature data for multiple branch cables is collected by the same temperature sensor, resulting in a uniform first cable temperature value used for all branch cables, then the largest second correction coefficient can be selected from all power generation devices to correct this uniform first cable temperature data.

[0149] In one embodiment, such as Figure 1 As shown, the system also includes a cloud platform. The cloud platform is used to back up and record at least one of the following: cable temperature data, temperature prediction results, anomaly detection results, alarm information, and navigation information, thereby facilitating traceability.

[0150] like Figure 4 As shown in the embodiments of this application, an intrinsic safety monitoring method for branch cables in a combiner box is also provided. Safety monitoring is performed using the intrinsic safety monitoring system for branch cables in a combiner box as described in any of the above embodiments. The method includes:

[0151] S401: Detects cable temperature data of each branch cable in the combiner box through multiple temperature sensors in the temperature sensor network.

[0152] S402: The cable temperature data is collected by multiple data acquisition devices in the distributed data acquisition device and transmitted to the monitoring center.

[0153] S403: The data processing module in the monitoring center preprocesses the cable temperature data, and the temperature prediction module in the monitoring center makes a prediction based on the cable temperature data to generate a corresponding temperature prediction result.

[0154] S404: The alarm module in the monitoring center determines the anomaly based on the cable temperature data and the temperature prediction results, and generates corresponding alarm information.

[0155] S405: The navigation module in the monitoring center generates corresponding navigation information based on the abnormal target location information contained in the alarm information and the location information of the personnel receiving the alarm information.

[0156] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of this application.

Claims

1. An intrinsically safe monitoring system for branch cables in a combiner box, characterized in that, include: Temperature sensor network, distributed data acquisition device, monitoring center; The temperature sensor network is equipped with multiple temperature sensors, which are used to detect the cable temperature data of each branch cable in the combiner box at least. The distributed data acquisition device is equipped with multiple data acquisition devices, which are connected to the temperature sensor network and are used to collect cable temperature data of each branch cable in the preset area. The monitoring center includes a data processing module, a temperature prediction module, an alarm module, and a navigation module; The data processing module receives the cable temperature data fed back by the distributed data acquisition device and preprocesses the cable temperature data. The temperature prediction module makes predictions based on the cable temperature data and generates corresponding temperature prediction results. The alarm module determines anomalies based on the cable temperature data and the temperature prediction results, and generates corresponding alarm information. The navigation module generates corresponding navigation information based on the abnormal target location information contained in the alarm information and the location information of the personnel receiving the alarm information; The temperature sensor network includes multiple sub-networks, each corresponding to a single preset area; each sub-network contains at least one first temperature sensor and multiple second temperature sensors; the distance between the first temperature sensor and the combiner box is less than a first preset distance, used to detect the ambient temperature data corresponding to the combiner box; for the same branch cable, the second temperature sensors are respectively set at different positions on the branch cable, used to detect the cable temperature data corresponding to the combiner box; The temperature prediction module generates the temperature prediction result corresponding to the second temperature sensor by using a pre-trained regression model based on the cable temperature data and the ambient temperature data. The training process of the regression model includes: generating a corresponding time period based on the ambient temperature data; collecting corresponding cable temperature data through the second temperature sensor within the time period to form a corresponding time series, and performing stationarity processing on the time series; performing autocorrelation analysis and partial autocorrelation analysis on the time series, and selecting a regression model with a matching architecture based on the analysis results; wherein, the architecture of the regression model includes: autoregressive model, moving average model, autoregressive moving average model, and autoregressive integral moving average model; predicting model parameters based on the selected regression model and the analysis results, and fitting the model based on the predicted model parameters; evaluating the fitting results, and optimizing the regression model based on the evaluation results; In the temperature sensor network, for each branch cable, the connection position between the branch cable and the combiner box is determined; a second temperature sensor is set within a second preset distance of the connection position to collect the temperature data of the first cable; The temperature prediction module identifies the upstream equipment connected to the branch cable and determines the equipment type and status of the upstream equipment. Based on the equipment type and status, it corrects the temperature prediction result of the first cable temperature data. The correction process includes: selecting a power generation device from the upstream equipment based on the equipment type, and determining the number and type of the power generation device; if the number of devices is single, determining a first correction coefficient corresponding to the power generation device based on a first preset coefficient corresponding to the power generation type and the difference between the equipment status and a preset normal state, and correcting the temperature prediction result of the first cable temperature data using the first correction coefficient; if the number of devices is at least two, determining a corresponding first correction coefficient for each power generation device; if multiple power generation types exist among all power generation devices, adjusting the first correction coefficient for each power generation device based on the number of power generation types to obtain a second correction coefficient, and correcting the temperature prediction result of the first cable temperature data using the second correction coefficient; wherein the correction degree of the second correction coefficient is higher than that of the first correction coefficient.

2. The system according to claim 1, characterized in that, At least one second temperature sensor is installed at a third preset distance from the connection point to collect the temperature data of the second cable at its own location. The data processing module preprocesses and categorizes the cable temperature data in each preset area; Based on the classification results, for a single branch cable, determine the number of branches of the branch cable currently being detected simultaneously by the second temperature sensor corresponding to the temperature data of the first cable, and determine the number of historical anomalies at the location corresponding to each second cable temperature data. When the alarm module makes an anomaly determination based on the temperature data, it adjusts the first anomaly determination threshold of the first cable temperature data according to the number of branches, and adjusts the second anomaly determination threshold of the second cable temperature data according to the number of historical anomalies. The more branches there are, the higher the first anomaly determination threshold becomes; the more historical anomalies there are, the lower the second anomaly determination threshold becomes.

3. The system according to claim 2, characterized in that, The temperature prediction module generates a corresponding temperature prediction result for each cable temperature data. The alarm module performs anomaly determination on each cable temperature data according to the temperature prediction result; Identify the third temperature sensor corresponding to the abnormal cable temperature data, and identify other temperature sensors within the influence range of the third temperature sensor. The abnormal target location information corresponding to the third temperature sensor is used as the primary alarm target, and the abnormal target location information corresponding to other temperature sensors within the influence range is used as the secondary alarm target, and corresponding alarm information is generated. Specifically, for the second temperature sensor installed within the second preset distance of the connection location, its influence range includes the junction box where it is located and one or more corresponding branch cables; for the second temperature sensor installed outside the third preset distance of the connection location, its influence range includes one or more corresponding branch cables.

4. The system according to claim 3, characterized in that, The navigation module abstracts the map of the current region into a spatial map; Generate corresponding nodes and edges in the spatial graph; wherein, the nodes include primary nodes and secondary nodes, which correspond to the primary alarm target and the secondary alarm target, respectively, and the edges correspond to roads in the current region; Based on the personnel location information, corresponding personnel nodes are generated in the spatial diagram; For each node, calculate the shortest spatial distance between that node and other nodes in the spatial graph; Using the personnel nodes as the starting point and the main nodes as the ending point, a first navigation path is generated based on the corresponding shortest spatial distance. Using the personnel nodes as the starting point and the main nodes as the ending point, and by arranging and combining the shortest spatial distances between the nodes, one or more secondary nodes are used as path nodes to generate their respective corresponding second navigation paths. Among all the second navigation paths, the second navigation path whose spatial distance is greater than the spatial distance corresponding to the first navigation path by a fourth preset distance is deleted, and among the remaining second navigation paths, the second navigation path with the most path nodes is selected as the third navigation path; Based on the first navigation path and the third navigation path, corresponding navigation information is generated.

5. The system according to claim 1, characterized in that, The system also includes: a cloud platform; The cloud platform is used to back up and record at least one of the following: cable temperature data, temperature prediction results, anomaly determination results, alarm information, and navigation information.

6. A method for intrinsically safe monitoring of branch cables in combiner boxes, characterized in that, The method of performing safety monitoring using the intrinsically safe monitoring system for junction box branch cables as described in any one of claims 1 to 5 includes: The temperature data of each branch cable in the combiner box is detected by multiple temperature sensors in the temperature sensor network. The cable temperature data is collected by multiple data acquisition devices in the distributed data acquisition system and transmitted to the monitoring center. The cable temperature data is preprocessed by the data processing module in the monitoring center, and then predicted by the temperature prediction module in the monitoring center based on the cable temperature data to generate the corresponding temperature prediction result. The alarm module in the monitoring center determines anomalies based on the cable temperature data and the temperature prediction results, and generates corresponding alarm information. The navigation module in the monitoring center generates corresponding navigation information based on the abnormal target location information contained in the alarm information and the location information of the personnel receiving the alarm information. The temperature sensor network includes multiple sub-networks, each corresponding to a single preset area; each sub-network contains at least one first temperature sensor and multiple second temperature sensors; the distance between the first temperature sensor and the combiner box is less than a first preset distance, used to detect the ambient temperature data corresponding to the combiner box; for the same branch cable, the second temperature sensors are respectively set at different positions on the branch cable, used to detect the cable temperature data corresponding to the combiner box; The temperature prediction module generates the temperature prediction result corresponding to the second temperature sensor by using a pre-trained regression model based on the cable temperature data and the ambient temperature data. The training process of the regression model includes: generating a corresponding time period based on the ambient temperature data; collecting corresponding cable temperature data through the second temperature sensor within the time period to form a corresponding time series, and performing stationarity processing on the time series; performing autocorrelation analysis and partial autocorrelation analysis on the time series, and selecting a regression model with a matching architecture based on the analysis results; wherein, the architecture of the regression model includes: autoregressive model, moving average model, autoregressive moving average model, and autoregressive integral moving average model; predicting model parameters based on the selected regression model and the analysis results, and fitting the model based on the predicted model parameters; evaluating the fitting results, and optimizing the regression model based on the evaluation results; In the temperature sensor network, for each branch cable, the connection position between the branch cable and the combiner box is determined; a second temperature sensor is set within a second preset distance of the connection position to collect the temperature data of the first cable; The temperature prediction module identifies the upstream equipment connected to the branch cable and determines the equipment type and status of the upstream equipment. Based on the equipment type and status, it corrects the temperature prediction result of the first cable temperature data. The correction process includes: selecting a power generation device from the upstream equipment based on the equipment type, and determining the number and type of the power generation device; if the number of devices is single, determining a first correction coefficient corresponding to the power generation device based on a first preset coefficient corresponding to the power generation type and the difference between the equipment status and a preset normal state, and correcting the temperature prediction result of the first cable temperature data using the first correction coefficient; if the number of devices is at least two, determining a corresponding first correction coefficient for each power generation device; if multiple power generation types exist among all power generation devices, adjusting the first correction coefficient for each power generation device based on the number of power generation types to obtain a second correction coefficient, and correcting the temperature prediction result of the first cable temperature data using the second correction coefficient; wherein the correction degree of the second correction coefficient is higher than that of the first correction coefficient.

Images