Temperature sensing protection system and protection method based on low-voltage photovoltaic intelligent switch

Abstract

The invention discloses a temperature sensing protection system and protection method based on a low-voltage photovoltaic intelligent switch, belongs to the technical field of low-voltage photovoltaic power distribution systems, and aims to solve the problems that the temperature protection of the associated part of the intelligent switch is easy to trigger by mistake, the temperature fitting precision is insufficient, and the early warning processing is low in efficiency in the existing scheme. Comprising a high-incidence sensitive point identification module, a staged temperature building simulation combination module and an abnormity evaluation early warning processing module, a matrix is built through temperature and system operation parameters, and temperature high-incidence sensitive points are screened in combination with double indexes; adapting the exclusive temperature model according to starting, stable power generation and power sudden change stages, and verifying and adjusting; extracting multi-dimensional abnormal feature fusion to calculate an early warning risk value, and forming an operation and maintenance sequence in combination with part importance ranking; the protection method corresponds to system core steps. According to the invention, the temperature sensing precision and protection reliability of the intelligent switch are improved, the fault handling efficiency is optimized, the risk of equipment damage or shutdown caused by abnormal temperature of the low-voltage photovoltaic system is effectively reduced, and the low-voltage photovoltaic intelligent switch full-scene temperature protection requirement is met.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of low-voltage photovoltaic power distribution systems, and particularly relates to a temperature sensing protection system and method based on a low-voltage photovoltaic intelligent switch. BACKGROUND

[0002] The low-voltage photovoltaic intelligent switch is a core control component of the low-voltage photovoltaic power distribution system, and the temperature state of the electrical connection part associated with the switch directly determines the safety and stability of the system operation. However, the existing temperature protection schemes for such switches have the following defects: In the electrical connection part associated with the low-voltage photovoltaic intelligent switch, it is difficult to accurately identify the real heating failure and the false temperature rise caused by environmental interference, and it is also difficult to efficiently lock the key heating parts with high failure frequency and high risk level, which easily causes the temperature protection of the switch to be mis-triggered or missed-triggered. The temperature change law is significantly different in different operating stages such as start-up, stable power generation, and power mutation of the low-voltage photovoltaic intelligent switch. The existing single temperature model cannot adapt to the characteristics of each stage, resulting in insufficient temperature fitting accuracy and inability to provide reliable data support for the temperature abnormality judgment of the switch. The temperature abnormality evaluation index of the low-voltage photovoltaic intelligent switch associated part is single, which is difficult to fully reflect the severity of the abnormality, and lacks a scientific operation and maintenance sorting mechanism combining the urgency of the fault and the importance of the part, resulting in insufficient temperature warning accuracy and low fault disposal efficiency. Therefore, we propose a temperature sensing protection system and method based on a low-voltage photovoltaic intelligent switch. SUMMARY

[0003] The purpose of the present application is to provide a temperature sensing protection system and method based on a low-voltage photovoltaic intelligent switch to solve the problems raised in the background.

[0004] To achieve the above purpose, the present application provides the following technical scheme: a temperature sensing protection system based on a low-voltage photovoltaic intelligent switch, comprising: A high-frequency sensitive point identification module: collects the temperature of the electrical connection part and the core operating parameters of the photovoltaic system, constructs a matrix, and screens candidate heating sensitive points according to the temperature fluctuation value and peak value; calculates the heat sensitivity evaluation index combining the sensitive point temperature and operating parameters, and calculates the historical fault over-temperature index according to the historical fault data, and screens the temperature high-frequency sensitive points accordingly; A phased temperature modeling and fitting module: calculates the photovoltaic component operating load index and the first derivative, second derivative, and adjacent fluctuation amplitude, determines the start-up, stable power generation, and power mutation stages, integrates the continuous time periods, constructs a dedicated model for the temperature high-frequency sensitive points according to the corresponding stage, verifies the model after solving the parameters through the determination coefficient and the root mean square error, and adjusts the model if it is unqualified until it meets the requirements. The abnormality evaluation early warning treatment module: extracts the temperature rise rate, temperature fluctuation residual, multi-sensitive point coordination and trend deviation of the qualified period of the temperature high-sensitive point, fuses to form a comprehensive feature and normalizes, calculates an early warning risk value, and marks a temperature fault point; matches the importance value of the key part, calculates a priority processing evaluation value, sorts the fault processing sequence according to the priority processing evaluation value, and pushes the fault processing sequence to the operation and maintenance end.

[0005] Preferably, the specific process of screening the candidate heat-sensitive point is as follows: The distributed sensor is used to collect the temperature data of all electrical connection parts of the low-voltage photovoltaic power distribution full link in the collection period, and a temperature time sequence matrix is constructed; meanwhile, the core operation parameters of the photovoltaic system at each time are collected, and an operation parameter time sequence matrix is constructed; For each electrical connection part, the variance of the temperature value in the collection period is calculated to obtain the temperature fluctuation value, and the temperature peak value is obtained; If the temperature fluctuation value exceeds the preset temperature fluctuation reference variance, or the temperature peak value exceeds the preset temperature threshold, the part is marked as a candidate heat-sensitive point; all candidate heat-sensitive points are sorted to form a candidate heat-sensitive point set.

[0006] Preferably, the specific process of screening the temperature high-sensitive point is as follows: For each candidate heat-sensitive point, its temperature time sequence and each operation parameter time sequence are extracted from the temperature time sequence matrix and the operation parameter time sequence matrix respectively; The Pearson correlation coefficient is calculated with the temperature as the dependent variable and the operation parameter as the independent variable, and the absolute value is taken to obtain the thermal correlation coefficient, and the thermal sensitivity evaluation index is obtained by combining the preset weight coefficient of the operation parameter; if the index reaches the preset threshold, it is marked as a heat-sensitive point; The temperature-related historical fault data are called, the historical over-temperature frequency proportion is counted, the historical fault over-temperature index is obtained by combining the fault severity weight, and if the index reaches the preset threshold, it is marked as a temperature high-sensitive point; all temperature high-sensitive points are integrated to form a temperature high-sensitive point set.

[0007] Preferably, the specific process of calculating the photovoltaic module operation load index and the first derivative, the second derivative, and the adjacent fluctuation amplitude is as follows: The power generation power and loop current of the photovoltaic module at each time in the current collection period are obtained, and the operation load index corresponding to each time is obtained by combining the rated power, rated current and preset weight coefficient of the photovoltaic module; Based on the numerical differentiation method, the first derivative, the second derivative and the adjacent fluctuation amplitude of the operation load index are calculated for each collection time; The first derivative represents the variable rate at the corresponding collection time, the second derivative represents the change acceleration at the corresponding collection time, and the adjacent fluctuation amplitude represents the stability of the operation state at the corresponding collection time and the previous time, and the adjacent fluctuation amplitude of the first collection time is zero.

[0008] Preferably, the specific process of determining the start-up, stable power generation, power mutation stage and integrating continuous periods is as follows: The preset start-up load preset threshold, the running load index first-order derivative preset threshold, the running load index second-order derivative preset threshold, and the running load index adjacent fluctuation amplitude preset threshold are set. The start-up stage needs to meet four conditions of load range, load rapid rise, stable rise rate, and no large fluctuation between adjacent time points. The stable power generation stage needs to meet three conditions of slow load change rate, no obvious acceleration change, and minimal fluctuation between adjacent time points. The power mutation stage meets any condition of load sudden change or adjacent fluctuation amplitude being too large, and in case of conflict, the start-up stage is determined first. The running load index, its first-order derivative, second-order derivative, and adjacent fluctuation amplitude at each time point are obtained, the stage criteria are matched, and the running stage label is marked. The initial set is formed by merging continuous time periods with the same label. If the fragmentation period is followed by a non-fragmentation period of the same type, it is merged into the period. If it only exists on one side, it is merged into that side. If it exists independently, it is merged into the adjacent time period of the same type to obtain the final period set. For each temperature high-sensitivity point, the temperature sequence is divided according to the start and end time of each continuous period in the final period set, and a phase-temperature subsequence corresponding to each period is obtained.

[0009] Preferably, the specific process of constructing a dedicated model for the temperature high-sensitivity point according to the corresponding phase is as follows: For the phase-temperature subsequence, if its running phase label is the start-up stage, a first-order linear model is used, and the model contains a temperature intercept parameter and a temperature time coefficient specific to the stage. If the label is the stable power generation stage, a third-order polynomial model is used, and the model contains a basic temperature intercept parameter, a first-order term, a second-order term, and a third-order term coefficient of temperature change over time. If the label is the power mutation stage, a segmented linear model is used with the load mutation time as the boundary, and the corresponding dedicated model parameters are configured before and after the mutation time.

[0010] Preferably, after solving the parameters, the specific process of verification through the determination coefficient and the root mean square error is as follows: For each temperature high-sensitivity point, the least squares method is used to solve the dedicated parameters of the start-up, stable power generation, and power mutation stage models, aiming to minimize the sum of squared deviations of the fitted temperature and the actual temperature in the corresponding period. Through the double-index quantification verification of the determination coefficient and the root mean square error, preset thresholds of both are satisfied, and the model is qualified and the corresponding period is marked as a qualified period; if not satisfied, the order of the general model is adjusted, the data is repeatedly substituted into the parameter to solve and verify, until the model meets the double-index standard.

[0011] Preferably, the specific process of calculating the early warning risk value and marking the temperature fault point is as follows: For each qualified period of the temperature high-incidence sensitive point, four core abnormal features of temperature rise rate, temperature fluctuation residual, multi-sensitive point synergy and trend deviation degree are extracted; Fusion of the characteristics of each qualified period, the comprehensive temperature rise rate, the comprehensive fluctuation residual, the comprehensive synergy and the comprehensive trend deviation degree are obtained; The four integrated features are normalized and dimensionless, and the early warning risk value is calculated combined with the preset weight coefficient; If the early warning risk value reaches the preset threshold, the temperature fault early warning is triggered, and the temperature high-incidence sensitive point is marked as a temperature fault point.

[0012] Preferably, the specific process of forming the fault handling sequence and pushing to the operation and maintenance end is as follows: An importance assignment library of temperature high-incidence sensitive points is constructed, and a preset key part importance value is configured for each temperature high-incidence sensitive point in the library; the temperature fault point is matched with the assignment library to obtain the corresponding key part importance value; For each temperature fault point, the early warning risk value and the key part importance value are combined, and the priority handling evaluation value is calculated according to the preset weight coefficient; All temperature fault points are sorted in descending order of priority handling evaluation value to form a temperature fault priority handling sequence, and the sequence is sent to the fault operation and maintenance end, and the operation and maintenance end carries out operation and maintenance processing in sequence according to the sequence.

[0013] Compared with the prior art, the beneficial effects of the present application are: (1) The temperature sensing protection system and protection method based on the low-voltage photovoltaic intelligent switch, the potential heating point of the intelligent switch related electrical connection part is screened through the temperature fluctuation value and the peak value, the real electrical fault temperature rise and the environmental distortion false temperature rise are distinguished combined with the heat sensitive evaluation index, the index is calculated based on the Pearson correlation coefficient of temperature and system operation parameters and the preset weight coefficient, and then the high fault risk part is focused relying on the historical fault overtemperature index, the temperature high-incidence sensitive point is accurately identified, the intelligent switch temperature protection is avoided from being mistakenly triggered or missed, and clear targeting is provided for subsequent protection actions.

[0014] (2) The temperature sensing protection system and protection method based on the low-voltage photovoltaic intelligent switch accurately divides the period by quantifying the operating load index and its derivative characteristics, respectively adapts the exclusive temperature model and solves the parameters, and realizes the high-precision fitting of the temperature change in each stage through the double-index verification and model adjustment mechanism, so that the temperature sensing of the intelligent switch is more in line with the actual operating state, ensures that the protection action is triggered based on the real temperature anomaly, and improves the reliability of the temperature protection.

[0015] (3) The temperature sensing protection system and protection method based on the low-voltage photovoltaic intelligent switch extracts four core temperature anomaly characteristics of the high-incidence sensitive point associated with the intelligent switch, calculates the early warning risk value after normalization processing, and forms a disposal sequence combined with the importance value sorting of the parts, which realizes the accurate sensing and early warning of the temperature anomaly of the related parts of the intelligent switch, provides a scientific priority guide for operation and maintenance, makes the temperature protection disposal more timely and efficient, and effectively reduces the risk of equipment damage or shutdown of the low-voltage photovoltaic system caused by temperature anomaly of the related parts of the intelligent switch. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 The flowchart of the present application. DETAILED DESCRIPTION

[0017] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application. EMBODIMENT

[0018] Please refer to Figure 1 The temperature sensing protection system based on the low-voltage photovoltaic intelligent switch comprises a high-incidence sensitive point identification module, a staged temperature modeling and fitting module, and an abnormality evaluation and early warning disposal module. The high-incidence sensitive point identification module collects the temperature of the electrical connection part and the core operating parameters of the photovoltaic system, constructs a matrix and pre-processes it, screens the candidate heat-sensitive points according to the temperature fluctuation value and peak value, calculates the heat-sensitive evaluation index combined with the temperature of the sensitive point and the operating parameters, calculates the historical fault over-temperature index according to the historical fault data, screens the temperature high-incidence sensitive points, and organizes them into a set. The specific process is as follows: The distributed sensor collects temperature data of all electrical connection parts of the low-voltage photovoltaic power distribution full link in the collection period, and the electrical connection parts include: photovoltaic module output end, inverter terminal, intelligent switch input / output terminal, bus joint, cable connection point, fuse contact, load access end, etc. constructing a temperature time sequence matrix based on temperature data corresponding to each electrical connection position at each collection time , wherein x is a label of the electrical connection position, M is a total number of electrical connection positions in the low-voltage photovoltaic power distribution full link, and y is a label of the collection time, N is a total number of collection times in the collection period, Tyx is a temperature value corresponding to the xth electrical connection position at the yth collection time; performing a preprocessing operation on data in the temperature time sequence matrix; collecting core operating parameters of the photovoltaic system at each collection time, the core operating parameters including power generation power, loop current, bus voltage monitored by the intelligent switch, and the like, and constructing an operating parameter time sequence matrix ; m is a label of the core operating parameter, G is a total number of operating parameters; performing a preprocessing operation on data in the operating parameter time sequence matrix; for each electrical connection position, calculating a variance of the temperature value corresponding to each collection time in the collection period to obtain a temperature fluctuation value; and simultaneously obtaining a temperature peak value of the electrical connection position in the collection period; presetting a temperature fluctuation reference variance and a temperature threshold; if the temperature fluctuation value of the electrical connection position is greater than the corresponding preset temperature fluctuation reference variance or the temperature peak value is greater than the corresponding preset temperature threshold, marking the electrical connection position as a candidate heat-sensitive point; sorting all candidate heat-sensitive points in the collection period to form a candidate heat-sensitive point set; for each candidate heat-sensitive point, extracting a temperature time sequence thereof from the temperature time sequence matrix, and simultaneously extracting a time sequence corresponding to each operating parameter from the operating parameter time sequence matrix; taking the temperature of the sensitive point as a dependent variable and each operating parameter at the corresponding time as an independent variable, respectively calculating a Pearson correlation coefficient and taking an absolute value to obtain a heat correlation coefficient of the candidate heat-sensitive point with respect to each operating parameter , and using the formula: , to obtain a heat sensitivity evaluation index RP, wherein is a preset weight coefficient corresponding to the mth operating parameter; wherein the greater the heat sensitivity evaluation index, the higher the degree of coincidence of the candidate heat-sensitive point with the synchronous fluctuation of each core operating parameter; and if the temperature is abnormal, it is more likely to be a temperature rise driven by a real electrical fault rather than a false temperature rise caused by outdoor exposure, electromagnetic interference, or the like; presetting a heat sensitivity evaluation index threshold, and if the heat sensitivity evaluation index corresponding to the candidate heat-sensitive point is greater than or equal to the corresponding preset threshold, marking the candidate heat-sensitive point as a heat-sensitive point; For each heat-sensitive point, the temperature-related historical fault data of the low-voltage photovoltaic intelligent switch is called; in each collection period corresponding to each historical temperature-related fault condition, the number of collection time points at which the current heat-sensitive point temperature value exceeds the preset temperature threshold is counted, and then divided by the total number of collection time points in the corresponding collection period to obtain the historical over-temperature frequency proportion under the corresponding historical temperature-related fault condition; A fault severity weight corresponding to each historical temperature-related fault condition is given; the higher the fault severity, the greater the weight, and the sum of the fault severity weights of all temperature-related fault conditions is equal to one; The historical over-temperature frequency proportion of the heat-sensitive point in each historical temperature-related fault condition is multiplied by the corresponding fault severity weight, and then the product is accumulated to obtain the historical fault over-temperature index of the heat-sensitive point; A preset historical fault over-temperature index threshold is set, and if the historical fault over-temperature index of the heat-sensitive point is greater than or equal to the corresponding preset threshold, the heat-sensitive point is marked as a temperature high-frequency sensitive point; All temperature high-frequency sensitive points are integrated to form a temperature high-frequency sensitive point set.

[0019] It should be noted that the initial screening stage is based on the temperature time sequence matrix and the operating parameter time sequence matrix, the temperature fluctuation value is obtained by calculating the variance of the temperature value in the collection period of each electrical connection part, and the temperature peak value is extracted, and the parts exceeding the preset reference variance or temperature threshold are marked as candidate heat-sensitive points, so as to quickly narrow the monitoring range and ensure the efficiency of screening; The precise screening stage realizes precise identification through double index calculation: one is the heat-sensitive evaluation index, which extracts the temperature time sequence and each operating parameter time sequence of the candidate point from the two matrices, calculates the Pearson correlation coefficient with temperature as the dependent variable and operating parameters as the independent variable, and takes the absolute value to obtain the heat correlation coefficient, and then sums the operating parameter preset weight coefficient to obtain the index, which can effectively quantify the synchronization fluctuation coincidence degree of the temperature of the candidate point and the system operating parameters, accurately exclude false temperature rise caused by environmental distortion such as outdoor exposure and electromagnetic interference, and lock the heat-sensitive point driven by real electrical fault; the second is the historical fault over-temperature index, which calls the temperature-related historical fault data, counts the historical over-temperature frequency proportion of the candidate point under each fault condition, and calculates the sum by combining the fault severity weight to obtain the key parts with high fault frequency and high severity, and improve the pertinence of identification; The final integrated set of temperature-sensitive points provides a clear target for the phased temperature modeling and fitting module's operational stage division and dedicated model construction. This eliminates the need for modeling to cover all electrical components, significantly improving modeling efficiency and fitting accuracy. Simultaneously, it lays a precise data foundation for feature extraction and risk value calculation in the anomaly assessment, early warning, and handling module. This ensures that subsequent early warning and handling can directly target high-risk areas, significantly reducing the probability of misjudgment and missed judgment in the entire system, and improving the overall efficiency of fault prediction and maintenance.

[0020] The phased temperature modeling and fitting module calculates the photovoltaic module's operating load index, first derivative, second derivative, and adjacent fluctuation amplitudes. It identifies the startup, stable power generation, and power surge stages and integrates continuous time periods. For each corresponding stage, it constructs a dedicated model for temperature-sensitive high-incidence points. After solving the parameters, it verifies the model using the coefficient of determination and root mean square error. If the model is unqualified, it is adjusted until it meets the standards. The specific process is as follows: Obtain the power generation of the photovoltaic modules at each acquisition time within the current acquisition period. With loop current And using the formula: The operating load index corresponding to each data collection time is obtained. ; in, and These are the rated power and rated current of the photovoltaic module, respectively, with a1 and a2 being preset weighting coefficients; For each data acquisition moment, the derivative characteristics are calculated using numerical differentiation methods. Specifically, the first and second derivatives of the operating load index at each acquisition moment, as well as the adjacent fluctuation amplitudes, are calculated. First derivative Characterization The rate of change at the y-th acquisition time (reflecting how fast the operating state changes): Using the formula: , in, The time interval between adjacent data collection moments; Second derivative Characterization The change in acceleration at the y-th acquisition time (reflecting the degree of drastic change in operating state); Using the formula: ; Adjacent fluctuation amplitude: characterizes the stability of the operating state between the y-th acquisition time and the previous time; Using the formula: When y=1, Zero; The criteria for determining the startup phase, stable power generation phase, and power surge phase of a low-voltage photovoltaic power distribution system are established as follows: The criteria for determining the start-up phase must be met simultaneously: ( (This refers to the startup load threshold, corresponding to the component startup power ratio). (Indicates a rapid increase in load); (This indicates a steady rate of ascent without drastic acceleration or deceleration.) (Indicates that there are no significant fluctuations between adjacent time points); The criteria for determining a stable power generation phase must simultaneously meet the following: (This indicates that the rate of load change is extremely slow); (Indicates no significant change in acceleration); (This indicates that the fluctuations between adjacent time points are minimal); During the power surge phase, any of the following conditions must be met: (This indicates a sudden and drastic increase / decrease in load). (This indicates significant load fluctuations between adjacent time points); When the startup phase and the mutation phase conflict, the startup phase takes precedence (because the startup phase is a specific process of the initial operation of the system). in, Preset threshold for startup load, A threshold is preset for the first derivative of the operating load index. Preset threshold for the second derivative of the operating load index. Preset thresholds for adjacent fluctuations in the operating load index; The system acquires the operating load index, its first and second derivatives, and adjacent fluctuation amplitudes at each acquisition time, and matches them with the criteria for determining the start-up phase, stable power generation phase, and power burst phase. It then outputs the corresponding operating phase (start-up phase, stable power generation phase, and power burst phase) and labels it. Starting from the first collection moment of the collection cycle, sequentially determine whether the labels of the current moment and the previous moment's running stage are consistent; If multiple consecutive time points meet the criteria, have the same and adjacent labels, they are merged into a single continuous time period, and the start and end times of that time period are recorded. After the traversal is complete, all consecutive time periods will be integrated according to the time order to obtain the initial time period set; Traverse the initial time period set, preset a minimum continuous duration threshold. If the continuous duration of the initial time period set is less than the minimum continuous duration threshold, it is determined to be a fragmented time period. If the fragmented period is preceded and followed by two non-fragmented periods of the same type, then the fragmented period is merged into the preceding and following periods. If a fragmented time period exists only on one side and is not fragmented, it is directly merged into the non-fragmented time period on that side. If a fragmented time period exists independently, it is merged into the nearest adjacent time period of the same type based on its label type (determined by a combination of label similarity and time distance). After integration, a complete and continuous set of final time periods for the operational phases is obtained; For each high-temperature sensitive point Based on the complete temperature time series data within its collection period, the temperature sequence is divided into stages according to the start and end times of each continuous time period in the final time period set of the operation stage, to obtain a stage-temperature subsequence that corresponds one-to-one with each time period in the set; where i is the label of the high temperature point, i=1,2,...,n, and n is the total number of high temperature sensitive points. If the stage label of the continuous time period corresponding to the stage-temperature subsequence is the start-up stage, then substituting it into the first-order linear model, the expression is: ,in, , and This refers to the start and end times of the continuous period. Sensitive points for high temperature The fitted temperature at time y during the startup phase. and Sensitive points for high temperature The specific startup phase model parameters are: the temperature intercept parameter in the startup phase and the temperature-time coefficient in the restart phase. If the operating stage label of the stage-temperature subsequence corresponding to the continuous time period is the stable power generation stage, then substituting it into the third-order polynomial model, the expression is: ,in, Sensitive points for high temperature Let y be the fitted temperature at time y during the stable power generation phase; , , , Sensitive points for high temperature The exclusive model parameters are: the basic temperature intercept parameter of the high temperature sensitivity point in the stable power generation stage, the coefficient of the first term of temperature changing with time, the coefficient of the second term of temperature changing with time, and the coefficient of the third term of temperature changing with time. If the stage label of the stage-temperature subsequence corresponding to the continuous time period is the power mutation stage, then substituting it into the piecewise linear model with the mutation time as the boundary, the expression is: ,in, Sensitive points for high temperature The fitted temperature at time y during the power mutation phase. The moment of load change within this continuous period (determination method: when the power fluctuation amplitude at a certain moment exceeds the preset threshold, that moment is the moment of load change). , Sensitive points for high temperature Dedicated model parameters prior to the moment of load mutation; , Sensitive points for high temperature Dedicated model parameters after the moment of load mutation; For each high-temperature sensitive point All models aim to minimize the sum of squared deviations between the fitted temperature and the actual temperature within the corresponding time period. The least squares method is used to solve for the specific parameters of the general model at each stage. The specific solution process is as follows: Solving for specific parameters during the startup phase, and constructing the objective function as follows: ,in, Sensitive points for high temperature The objective function value corresponding to the startup phase represents the minimum sum of squared deviations between the actual temperature and the fitted temperature during this phase. Sensitive points for high temperature The actual temperature value at time y during the startup phase; and Taking the partial derivative with respect to 0 and setting it equal to 0, we obtain the parameter estimates. and ( , for , (Estimation results) Solving for specific parameters during the stable power generation phase, the objective function is constructed as follows: , in, Sensitive points for high temperature The objective function value corresponding to the stable power generation stage represents the minimum value of the sum of squares of the deviations between the actual temperature and the fitted temperature during this stage. right , , , Find the partial derivative and set it equal to 0 to obtain the parameter estimate. , , , ; Power mutation phase: Construct objective functions for the data before and after the mutation time respectively: The objective function before the mutation was: ; The objective function after mutation is: ; For the parameters in the objective function before mutation , and the parameters in the objective function after mutation , Find the partial derivatives and set them equal to 0. Then, solve the system of equations to obtain the parameter estimates. , , , ; For each high-temperature sensitive point In each runtime phase, the model is quantitatively verified using both the coefficient of determination and the root mean square error to ensure that the model fitting accuracy meets the subsequent early warning requirements. Coefficient of determination The expression is: ,in, Sensitive points for high temperature The temperature value at the y-th data collection time. Sensitive points for high temperature The fitted temperature at the y-th acquisition time, Sensitive points for high temperature During the period The average temperature inside; The coefficient of determination characterizes the model's ability to explain the temperature change trend at temperature-sensitive points with high incidence rates; the closer the value is to 1, the better the fit. 2. Root mean square error The expression is:

[0021] in, For time period Number of samples within; The average deviation between the fitted temperature value and the actual value at the temperature sensitive point where high temperature incidence occurs; the smaller the value, the higher the fitting accuracy. Preset determination coefficient threshold With root mean square error threshold If the runtime model simultaneously meets the following two conditions, then the runtime model is deemed qualified, and the runtime segment is marked as a qualified time period: ; ; If any of the above conditions are not met, the model is deemed unqualified. The order of the general model is adjusted, and the corresponding time period data of the high-temperature sensitivity point is re-substituted after adjustment. The above-mentioned process of solving the specific parameters and quantification verification is repeated until the model meets the dual-index verification standard, ensuring that the model of each high-temperature sensitivity point has a reliable fitting effect in each operating period.

[0022] Furthermore, the order of the general model can be adjusted, specifically including: During the stable power generation phase: The core type of the general model is a polynomial model. If the third-order polynomial model fails to accurately adapt to the nonlinear temperature fluctuation characteristics, the model order can be reduced to a second-order polynomial model to optimize the fitting accuracy by reducing the number of polynomial terms. Start-up phase: The core type of the general model is a first-order linear model. If the fitting deviation is due to insufficient adaptation between the linear constraints and the heating trend, the constraints of the linear model can be optimized (specifically, the fitting weights of the temperature-time coefficients can be adjusted) to enhance the model's fit to the linear heating law. During the power mutation phase: The core type of the general model is a piecewise linear model. If the linear fitting of the two segments before and after the mutation fails to accurately capture the temperature rate jump characteristics, the accuracy of segment node determination can be optimized (i.e., the identification threshold of the load mutation moment can be refined to ensure that the mutation moment is completely matched with the actual power mutation moment), or the fitting interval boundary of the two linear models can be adjusted (so that the interval completely corresponds to the stable temperature change interval before and after the mutation) to improve the adaptability of the piecewise fitting.

[0023] It should be noted that the operating load index is calculated based on the power generation and loop current of the photovoltaic module, combined with the rated power, rated current and preset weighting coefficients. Then, the first derivative, second derivative and adjacent fluctuation amplitude are obtained by numerical differentiation method. These three factors quantify the speed, intensity and stability of the change in the operating state, providing an objective and detailed quantitative basis for subsequent stage judgment, and avoiding the subjectivity of dividing stages based solely on experience. The startup phase requires meeting four conditions, including load range and rapid rise; the stable power generation phase requires meeting three conditions, including extremely slow rate of change; and the power surge phase requires meeting any one of the conditions, namely, sudden load change or excessive fluctuation. In case of conflict, the startup phase is prioritized. At the same time, by merging continuous time periods and processing fragmented time periods, a complete and continuous final time period set is formed, ensuring the consistency of temperature data characteristics in each phase and eliminating data interference for targeted modeling. A first-order linear model is used to address the linear temperature rise during the start-up phase, a third-order polynomial model is used to address the nonlinear fluctuations during the steady phase, and a piecewise linear model is used to address the rate jumps during the abrupt change phase. The least squares method is used to solve for the specific parameters of each model (with the goal of minimizing the sum of squares of the deviations between the fitted temperature and the actual temperature), so that the model can accurately adapt to the temperature change patterns of different phases. The model is quantitatively verified using two indicators: the coefficient of determination (characterizing the explanatory power of the trend) and the root mean square error (characterizing the fitting deviation). When the threshold is not reached, the model is adjusted in a targeted manner: the polynomial order is reduced in the stable phase, the linear constraints are optimized in the startup phase, and the segment nodes are refined or the fitting interval is adjusted in the mutation phase, so as to ensure that the model has high fitting accuracy for each temperature high-incidence sensitive point in each time period. Ultimately, the output qualified stage model and qualified time period provide an accurate temperature fitting data foundation for the anomaly assessment and early warning handling module to extract features such as temperature rise rate and temperature fluctuation residuals, effectively reducing the error of subsequent feature extraction and risk calculation, and significantly improving the early warning accuracy and reliability of the entire protection system.

[0024] The anomaly assessment, early warning, and handling module extracts the temperature rise rate, temperature fluctuation residual, multi-sensitive point synergy, and trend deviation during qualified periods of high-temperature sensitive points. These are then fused to form a comprehensive feature set and normalized. An early warning risk value is calculated, and temperature fault points are marked. The module matches the importance values ​​of key components, calculates priority handling evaluation values, and sorts these values ​​to form a fault handling sequence, which is then pushed to the operations and maintenance end. The specific process is as follows: Targeting high-temperature sensitive points For each qualified time period q, q=1,2,...,Q, where Q is the total number of qualified time periods, the following core abnormal feature values ​​are extracted, including: Temperature rise rate Characterization The rate of temperature increase / decrease in the q-th time period reflects the drasticness of temperature change, and is expressed as: Start-up phase: ; Stable power generation phase: ; Power mutation phase: ; Temperature fluctuation residual Characterization The degree of deviation between the actual temperature and the fitted temperature during the q-th time period reflects the temperature stability, and is expressed as: , The larger the temperature fluctuation residual, the more significant the deviation of the actual temperature from the normal trend, and the more likely there are abnormal risks such as poor local contact or overload. Multi-sensory point synergy Characterization Synchronization with temperature changes at other temperature-sensitive points is used to distinguish between single-point, occasional anomalies and system-level fault-driven anomalies. The expression is: ,in: Sensitive points for high temperature Within the acceptable time interval q, the fitted temperature at time y is... For the kth temperature-sensitive point The fitted temperature at time y within the acceptable time period q. The function for calculating the Pearson correlation coefficient; The closer the synergy of multiple sensitive points is to 1, the better. The more synchronized the temperature changes are with other temperature-sensitive points, the more likely it is caused by global factors or cascading failures such as abnormal system load or power grid fluctuations, rather than single-point environmental interference. Trend deviation Characterization The degree of deviation between the current temperature trend and the historical normal operating condition trend, excluding abnormal fluctuations, is expressed as: ,in The core parameters of the model for the current time period (taken during the startup phase) During the stable phase, take Mutation stage ), It is the statistical average value of core parameters under the same period of normal operating conditions in history (calculated based on more than 3 years of fault-free operation data). When the trend deviation exceeds the preset benchmark value, it indicates that the current temperature change pattern has deviated from the normal range and requires close attention. Sensitive points for high temperature The temperature rise rate, temperature fluctuation residual, multi-sensor synergy, and trend deviation of all qualified time periods are fused to obtain comprehensive characteristics of the entire acquisition cycle, enabling the analysis of... A comprehensive assessment of the temperature conditions is as follows: Overall temperature rise rate: Focusing on the most dramatic rate of temperature change throughout the entire cycle, highlighting key anomaly signals; Overall fluctuation residual: It captures the largest temperature deviation within the entire cycle, reflecting the most severe stability anomalies; Comprehensive synergy: By averaging, the accidental synchronicity of single-point periods is weakened, while the systematic synchronicity characteristics over the entire cycle are strengthened. Overall trend deviation: A comprehensive assessment of the overall deviation between the temperature trend throughout the entire cycle and historical normal operating conditions is conducted.

[0025] After normalizing and dimensionlessly processing the comprehensive temperature rise rate, comprehensive fluctuation residual, comprehensive synergy, and comprehensive trend deviation, the formula is used: High-temperature sensitive points were obtained. Corresponding warning risk value ; Where r1, r2, r3, and r4 are preset weight coefficients; A preset risk threshold for high-incidence sensitive points is set. If the risk value of a high-incidence sensitive point is greater than or equal to the corresponding preset threshold, a temperature fault warning for the high-incidence sensitive point is triggered, and the high-incidence sensitive point is marked as a temperature fault point. Construct a library of importance values ​​for high-temperature-incidence sensitive points, in which each high-temperature-incidence sensitive point corresponds to a preset importance value for a key component; By substituting temperature fault points into the importance assignment library of temperature-sensitive high-incidence points, the corresponding importance values ​​of key components are matched. For each temperature fault point, obtain its corresponding early warning risk value GF and critical component importance value ZY, and use the formula: The review priority processing evaluation value GY is obtained, where h1 and h2 are preset weight coefficients; All temperature fault points to be processed are obtained and sorted from largest to smallest based on the corresponding temperature fault priority processing evaluation value to obtain a temperature fault priority processing sequence, which is then sent to the fault operation and maintenance terminal. The fault operation and maintenance terminal performs operation and maintenance processing on the temperature fault points in the sequence in turn based on the temperature fault priority processing sequence.

[0026] It should be noted that for each qualified time period of high temperature sensitivity points, four core abnormal features are accurately extracted: the temperature rise rate directly reflects the severity of temperature change; the temperature fluctuation residual quantifies the deviation between the actual temperature and the fitted temperature (which can promptly detect hidden dangers such as poor local contact and overload); the synergy of multiple sensitive points is calculated through Pearson correlation coefficient, which can effectively distinguish between single-point accidental anomalies and anomalies driven by system-level faults; and the trend deviation is based on the statistical average of core parameters of more than 3 years of historical fault-free data, which can eliminate the interference of normal fluctuations. The four features comprehensively cover different dimensions of anomalies and avoid misjudgment by a single indicator. The comprehensive temperature rise rate focuses on the most dramatic changes throughout the entire cycle, highlighting key abnormal signals; the comprehensive fluctuation residual captures the maximum deviation, locking in the most serious stability problems; the comprehensive coordination takes the average value to weaken accidental synchronization and strengthen the characteristics of systematic anomalies; the comprehensive trend deviation comprehensively assesses the deviation from historical normal operating conditions, ensuring that no anomaly is missed, making the temperature status assessment more three-dimensional. By eliminating unit differences of different features through normalization and dimensionless processing, and combining the pre-set weighting coefficients to calculate the early warning risk value, the risk level of each sensitive point can be objectively measured, ensuring that high-risk parts are accurately marked as temperature failure points, and avoiding the subjectivity of risk assessment. By combining the early warning risk value with the importance of the location to calculate the priority evaluation value, and sorting them according to the value to form a processing sequence, the urgency of the fault is taken into account, as well as the impact weight of the location on the system. This allows maintenance work to focus on priorities without blindly investigating, which greatly improves maintenance efficiency and shortens the fault handling cycle. By closely linking the qualified time period and fitted temperature data of the phased temperature modeling and fitting module, the system achieves accurate extraction of abnormal features based on precise fitting. At the same time, through a closed-loop early warning-sorting-push mechanism, the abstract temperature data is transformed into actionable operation and maintenance instructions, which significantly improves the accuracy of abnormal early warning and the targeted nature of fault handling of the entire protection system, and effectively reduces the risk of equipment damage or shutdown caused by abnormal temperature in low-voltage photovoltaic systems.

[0027] Temperature sensing protection methods based on low-voltage photovoltaic smart switches include: Step 1: Collect the temperature of electrical connection parts and the core operating parameters of the photovoltaic system, construct a matrix, and screen candidate heat-sensitive points according to temperature fluctuation values ​​and peak values; calculate the thermal sensitivity evaluation index by combining the temperature of the sensitive points with the operating parameters, calculate the historical fault over-temperature index based on historical fault data, and screen high-temperature sensitive points accordingly. Step 2: Calculate the photovoltaic module operating load index and its first derivative, second derivative, and adjacent fluctuation amplitude. Determine the startup, stable power generation, and power surge stages and integrate them into continuous time periods. Construct a dedicated model for the corresponding stage as the temperature high-incidence sensitive point. After solving the parameters, verify them through the coefficient of determination and root mean square error. If the model is not qualified, adjust it until it meets the standard. Step 3: Extract the temperature rise rate, temperature fluctuation residual, synergy of multiple sensitive points and trend deviation during qualified periods of high temperature sensitivity points, integrate them to form comprehensive features and normalize them, calculate the early warning risk value, mark the temperature fault point; match the importance value of key parts, calculate the priority processing evaluation value, sort them according to the value to form a fault processing sequence and push it to the operation and maintenance end.

[0028] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A temperature sensing and protection system based on a low-voltage photovoltaic intelligent switch, characterized in that, include: High-risk sensitive point identification module: Collects the temperature of electrical connection parts and core operating parameters of photovoltaic system, constructs a matrix, and screens candidate heat-sensitive points according to temperature fluctuation value and peak value; calculates the thermal sensitivity evaluation index by combining the temperature of sensitive points with operating parameters, calculates the historical fault over-temperature index based on historical fault data, and screens high-risk temperature sensitive points accordingly. The phased temperature modeling and fitting module calculates the photovoltaic module's operating load index, first derivative, second derivative, and adjacent fluctuation amplitude, determines the startup, stable power generation, and power surge stages, integrates continuous time periods, constructs a dedicated model for temperature-sensitive points according to the corresponding stage, verifies the parameters through the coefficient of determination and root mean square error, and adjusts the model if it is not qualified until it meets the standard. Anomaly assessment and early warning module: Extracts the temperature rise rate, temperature fluctuation residual, synergy of multiple sensitive points and trend deviation during qualified periods of high temperature sensitivity points, integrates them to form comprehensive features and normalizes them, calculates early warning risk values, and marks temperature fault points; Match the importance values ​​of key components, calculate the priority evaluation values, sort them according to these values ​​to form a fault handling sequence, and push it to the operation and maintenance end.

2. The temperature sensing and protection system based on a low-voltage photovoltaic intelligent switch according to claim 1, characterized in that: The specific process for screening candidate heat-sensitive points is as follows: Distributed sensors are used to collect temperature data from all electrical connection points of the entire low-voltage photovoltaic power distribution chain during the acquisition period, and a temperature time series matrix is ​​constructed. At the same time, the core operating parameters of the photovoltaic system at each moment are collected, and an operating parameter time series matrix is ​​constructed. For each electrical connection, the variance of the temperature value within the acquisition period is calculated to obtain the temperature fluctuation value, and the temperature peak value is obtained. If the temperature fluctuation exceeds the preset temperature fluctuation baseline variance, or the temperature peak exceeds the preset temperature threshold, the location is marked as a candidate heat-sensitive point; all candidate heat-sensitive points are compiled to form a set of candidate heat-sensitive points.

3. The temperature sensing and protection system based on a low-voltage photovoltaic intelligent switch according to claim 2, characterized in that: The specific process for screening high-temperature sensitive points is as follows: For each candidate heat-sensitive point, its temperature time series and the time series of each operating parameter are extracted from the temperature time series matrix and the operating parameter time series matrix, respectively. The Pearson correlation coefficient is calculated with temperature as the dependent variable and operating parameters as independent variables, and the absolute value is taken to obtain the thermal parameter correlation coefficient. Combined with the preset weighting coefficient of operating parameters, the thermal sensitivity evaluation index is obtained. If the index reaches the preset threshold, it is marked as a heat-sensitive point. Retrieve historical fault data related to temperature, calculate the percentage of historical over-temperature frequency, and combine the fault severity weight to obtain the historical fault over-temperature index. If the index reaches a preset threshold, it is marked as a high-temperature sensitive point; all high-temperature sensitive points are integrated to form a set of high-temperature sensitive points.

4. The temperature sensing and protection system based on a low-voltage photovoltaic intelligent switch according to claim 3, characterized in that: The specific process for calculating the photovoltaic module operating load index, its first derivative, second derivative, and adjacent fluctuation amplitudes is as follows: The power generation and loop current of the photovoltaic module at each moment within the current acquisition period are obtained. Combined with the rated power, rated current and preset weighting coefficient of the photovoltaic module, the operating load index corresponding to each moment is obtained. Based on the numerical differentiation method, the first derivative, second derivative, and adjacent fluctuation amplitude of the operating load index are calculated for each data acquisition time. Among them, the first derivative represents the rate of change at the corresponding acquisition time, the second derivative represents the acceleration of change at the corresponding acquisition time, the adjacent fluctuation amplitude represents the stability of the operating state at the corresponding acquisition time and the previous time, and the adjacent fluctuation amplitude at the first acquisition time is zero.

5. The temperature sensing and protection system based on a low-voltage photovoltaic intelligent switch according to claim 4, characterized in that: The specific process for determining the startup, stable power generation, and power surge phases, and integrating the continuous time periods, is as follows: Preset thresholds for startup load, first derivative of operating load index, second derivative of operating load index, and adjacent fluctuation range of operating load index. The startup phase must simultaneously meet four conditions: load range, rapid load increase, stable rate of increase, and no significant fluctuations between adjacent time periods. During the stable power generation phase, three conditions must be met simultaneously: extremely slow load change rate, no significant acceleration change, and minimal fluctuation between adjacent time periods. The power surge phase meets either the condition of a sudden load change or an excessively large adjacent fluctuation amplitude. In case of a conflict, the start-up phase is prioritized. Obtain the operating load index and its first derivative, second derivative, and adjacent fluctuation amplitude at each time point, match the stage standard and mark the operating stage label, and merge consecutive time periods with the same label to form an initial set; If the fragmented time period is preceded and followed by non-fragmented time periods of the same type, it is merged into that time period; if it exists only on one side, it is merged into that side; if it exists independently, it is merged into the adjacent time period of the same type, thus obtaining the final set of time periods. For each temperature-sensitive point, the temperature sequence is divided according to the start and end times of each consecutive time period in the final time period set, resulting in a stage-temperature subsequence that corresponds one-to-one with each time period.

6. The temperature sensing and protection system based on a low-voltage photovoltaic intelligent switch according to claim 5, characterized in that: The specific process for constructing a dedicated model for temperature-sensitive areas according to the corresponding stage is as follows: For the stage-temperature subsequence, if its operation stage label is the start-up stage, a first-order linear model is adopted, and the model includes the temperature intercept parameter and temperature-time coefficient specific to this stage. If the label is for the stable power generation stage, a third-order polynomial model is used. The model includes the basic temperature intercept parameter, and the coefficients of the first, second, and third terms of temperature change over time. If the label is the power change phase, a piecewise linear model is adopted with the load change time as the boundary, and corresponding dedicated model parameters are configured before and after the change time.

7. The temperature sensing and protection system based on a low-voltage photovoltaic intelligent switch according to claim 6, characterized in that: After solving for the parameters, the model is verified using the coefficient of determination and root mean square error. If the model fails to meet the requirements, adjustments are made until it meets the standards. The specific process is as follows: For each temperature-sensitive point, with the goal of minimizing the sum of squared deviations between the fitted temperature and the actual temperature for the corresponding time period, the least squares method is used to solve for the specific parameters of the model for the startup, stable power generation, and power surge stages. The model is quantitatively verified using two indicators: the coefficient of determination and the root mean square error. Both thresholds are preset. If both are met, the model is considered qualified and the corresponding time period is marked as a qualified time period. If they are not met, the general model order is adjusted, and the data is substituted again to repeat the solution and verification until the model meets the dual indicator standard.

8. The temperature sensing and protection system based on a low-voltage photovoltaic intelligent switch according to claim 7, characterized in that: The specific process for calculating the early warning risk value and marking the temperature fault point is as follows: For each qualified time period of high-temperature sensitive points, four core abnormal features are extracted: temperature rise rate, temperature fluctuation residual, synergy of multiple sensitive points, and trend deviation. By integrating the characteristics of each qualified time period, we obtain the comprehensive temperature rise rate, comprehensive fluctuation residual, comprehensive synergy, and comprehensive trend deviation. The four comprehensive characteristics are normalized and dimensionless, and the warning risk value is calculated by combining the preset weight coefficients. If the warning risk value reaches the preset threshold, a temperature fault warning will be triggered, and the high-temperature sensitive point will be marked as a temperature fault point.

9. The temperature sensing and protection system based on a low-voltage photovoltaic intelligent switch according to claim 8, characterized in that: The specific process of forming a fault handling sequence and pushing it to the operation and maintenance end is as follows: Construct a library of importance assignment values ​​for high-temperature-incidence sensitive points, in which each high-temperature-incidence sensitive point is assigned a preset importance value for key components; Match the temperature fault point with the assignment library to obtain the corresponding critical component importance value; For each temperature fault point, the priority evaluation value is calculated based on the pre-set weighting coefficient, combining its early warning risk value and the importance value of key parts. All temperature fault points are sorted from largest to smallest according to the priority evaluation value to form a priority handling sequence for temperature faults. This sequence is sent to the fault operation and maintenance terminal, which then carries out operation and maintenance processing in sequence.

10. A temperature sensing and protection method based on a low-voltage photovoltaic intelligent switch, applied to the temperature sensing and protection system based on a low-voltage photovoltaic intelligent switch proposed in any one of claims 1-9, characterized in that, include: Step 1: Collect the temperature of electrical connection parts and the core operating parameters of the photovoltaic system, construct a matrix, and screen candidate heat-sensitive points according to temperature fluctuation values ​​and peak values; calculate the thermal sensitivity evaluation index by combining the temperature of the sensitive points with the operating parameters, calculate the historical fault over-temperature index based on historical fault data, and screen high-temperature sensitive points accordingly. Step 2: Calculate the photovoltaic module operating load index and its first derivative, second derivative, and adjacent fluctuation amplitude. Determine the startup, stable power generation, and power surge stages and integrate them into continuous time periods. Construct a dedicated model for the corresponding stage as the temperature high-incidence sensitive point. After solving the parameters, verify them through the coefficient of determination and root mean square error. If the model is not qualified, adjust it until it meets the standard. Step 3: Extract the temperature rise rate, temperature fluctuation residual, synergy of multiple sensitive points and trend deviation during qualified periods of high temperature sensitivity points, integrate them to form comprehensive features and normalize them, calculate the early warning risk value, and mark the temperature fault points; Match the importance values ​​of key components, calculate the priority evaluation values, sort them according to these values ​​to form a fault handling sequence, and push it to the operation and maintenance end.

Images