Method and system for avoiding false alarm in thermal depolymerization detection of organic insulating material

Abstract

A method and system for avoiding false alarm in thermal depolymerization detection of an organic insulating material, and the method comprises the steps: obtaining historical data of environmental parameters in a power distribution cabinet, evaluating the corresponding thermal depolymerization condition of the organic insulating material, and defining the data obtained through evaluation as a thermal depolymerization factor; performing adaptive normalization on each environment parameter in the historical data; performing data fusion on the normalized environment parameters in four dimensions to obtain a fusion feature data set in four dimensions; mutual information and correlation coefficients of the fusion feature data sets in the four dimensions and the corresponding thermal depolymerization evaluation factors are calculated respectively, and the fusion feature data set with the maximum influence is comprehensively selected; and establishing a prediction model based on a BP neural network, inputting a data fusion result into the prediction model to obtain a predicted thermal depolymerization evaluation factor, and giving an alarm if the predicted thermal depolymerization evaluation factor exceeds a set alarm factor. According to the invention, the influence between environment parameters is fully considered, and the alarm efficiency and accuracy are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of safety monitoring, and more specifically, relates to a method for detecting thermal depolymerization of an organic insulating material to avoid false alarms. Background Art

[0002] Organic insulating materials are used in a variety of electrical cabinets, such as distribution cabinets, battery cabinets, and server cabinets. These electrical cabinets may be put into use under various complex working conditions, such as dusty, humid and misty environments. The detection of thermal depolymerization of organic insulating materials in the electrical cabinets will inevitably cause interference, making it difficult to achieve effective prevention and control.

[0003] It can be seen from this that how to create a method to detect the thermal depolymerization of organic insulating materials and avoid false alarms is crucial to accurately detect the thermal depolymerization of organic insulating materials and achieve application and promotion.

[0004] CN110706444A discloses a comprehensive pyrolytic particle electrical fire monitoring method, device and system, including: S1, obtaining the real-time values ​​of 7 parameters in the low-voltage distribution cabinet, including temperature T, temperature change rate VT, thermal decomposition gas concentration Cgas, thermal decomposition gas concentration change rate Vgas, PM1.0 particle concentration C1.0, PM2.5 particle concentration C2.5 and PM10 particle concentration C10; obtaining the thresholds d1, d2, d3, d4, d5, d6, d7 corresponding to the 7 parameters; S2, assigning different dynamic weights to each parameter and combining the parameter values ​​exceeding the threshold with the dynamic weight to calculate the processing results; S3, issuing an alarm prompt according to the processing results. The present invention obtains the above 7 parameters as important parameters for early warning evaluation of safety in the low-voltage distribution cabinet, and integrates and analyzes these 7 parameters according to threshold setting and dynamic weight analysis. However, this patent performs data fusion on the basis of setting thresholds, without considering the mutual influence of the seven parameters, and only adopts one data fusion optimization method, which is not universal enough for different distribution cabinets and different situations. Summary of the invention

[0005] In order to solve the deficiencies in the prior art, the present invention provides a method for detecting the thermal depolymerization of an organic insulating material and avoiding false alarms.

[0006] The present invention adopts the following technical solution.

[0007] A first aspect of the present invention provides a method for detecting thermal depolymerization of an organic insulating material and avoiding false alarms, comprising:

[0008] Obtain historical data of environmental parameters in the power distribution cabinet. The historical data of each environmental parameter constitutes a historical data set. The thermal depolymerization of the organic insulating material corresponding to the historical data is evaluated. The evaluated data is defined as a thermal depolymerization factor, and the thermal depolymerization evaluation factor is used as a label.

[0009] Adaptively normalize each environmental parameter in the historical data set;

[0010] The normalized environmental parameters are fused in four dimensions, and the fused data in each dimension are combined into a set to obtain a fused feature data set in four dimensions;

[0011] The mutual information and correlation coefficients of the fused feature data sets and the corresponding thermal disaggregation evaluation factors in the four dimensions are calculated respectively, and the fused feature data set with the greatest influence is selected based on the correlation coefficient and mutual information;

[0012] A prediction model based on BP neural network was established, and the model was trained by selecting the most influential fusion feature data set and the corresponding thermal depolymerization evaluation factor;

[0013] The environmental parameters in the distribution cabinet are detected in real time, and then adaptively normalized. Then, data fusion of the same dimension as the selected fusion feature data set with the greatest influence is adopted. The result of data fusion is input into the prediction model to obtain the predicted thermal depolymerization assessment factor. If it exceeds the set alarm factor, an alarm is issued.

[0014] Preferably, the environmental parameters in the distribution cabinet include temperature, temperature change rate, thermal decomposition gas concentration, thermal decomposition gas concentration change rate, type 1 particle concentration, type 2 particle concentration and type 3 particle concentration; the type 1 particle concentration is the mass concentration value obtained by converting the density of industrial metal particles as equivalent particles; the type 2 particle concentration is the mass concentration value obtained by converting the density of pollutants in the air as equivalent particles; the type 3 particle concentration is the concentration of particles with a diameter of 1.0um to 10.0um.

[0015] Preferably, the adaptive normalization is:

[0016]

[0017] Among them, P i is the set of all historical data of the i-th environmental parameter, p i,j is the jth data of the i-th environmental parameter, p i ' ,j For p i,j The normalized result.

[0018] Preferably, the data fusion of the normalized environmental parameters in four dimensions is specifically as follows:

[0019]

[0020]

[0021]

[0022] Among them, f 1,j 、f 2,j 、f 3,j 、f 4,j are the fusion data of the jth data of all environmental parameters in the first, second, third, and fourth dimensions, respectively, p i ' ,j is the normalized result of the jth data of the ith environmental parameter, i = {1, 2, 3, 4, 5, 6, 7}, and the 1st to 7th environmental parameters are temperature, temperature change rate, thermal decomposition gas concentration, thermal decomposition gas concentration change rate, type 1 particle concentration, type 2 particle concentration, and type 3 particle concentration, respectively; w i,j is the weight of the jth data of the ith environmental parameter; I is the number of types of environmental parameters.

[0023] Preferably, the weight ω of the jth data of the i-th environmental parameter is i,j The selection rules are:

[0024] Calculate the influence of temperature and temperature change rate:

[0025]

[0026] where ω 1 ' ,j is the influence of temperature, ω 2 ' ,j is the influence ratio of temperature change rate, p 1,j is the jth data of temperature; p 1 ' ,j is the jth data of the normalized temperature; p 2 ' ,j The normalized temperature change rate is the jth data; K is the total number of historical data of each environmental parameter obtained;

[0027] Calculate the influence ratio of thermal decomposition gas concentration and thermal decomposition gas concentration change rate:

[0028]

[0029] where ω 3 ' ,j is the specific gravity affected by the concentration of thermal decomposition gas, ω4 ' ,j is the specific gravity affected by the change rate of the concentration of the thermal decomposition gas, p 3,j is the jth data of the concentration of thermal decomposition gas; p 3 ' ,j is the jth data of the normalized thermal decomposition gas concentration; p′ 4,j The normalized rate of change of the thermal decomposition gas concentration is the jth data;

[0030] Calculate the influence ratio of type 1 particle concentration, type 2 particle concentration and type 3 particle concentration:

[0031]

[0032] Among them, ω′ o,j is the influence ratio of the parameter of the oth link; p′ o,j is the jth data of the oth link parameter after normalization; o∈[5,6,7];

[0033] Normalize the impact ratio of all environmental parameters:

[0034]

[0035] ω i,j is the weight of the jth data of the i-th environmental parameter.

[0036] Preferably, the calculation of the correlation coefficients between the fused feature data sets in the four dimensions and the corresponding thermal depolymerization assessment factors is specifically:

[0037] The correlation coefficient calculation formula is:

[0038]

[0039] Among them, r h is the correlation coefficient between the fused feature dataset on the hth dimension and the corresponding thermal depolymerization assessment factor; K is the total number of historical data obtained for each environmental parameter; f h,j is the fusion data of the jth data of all environmental parameters in the hth dimension; j is the corresponding j-th thermal depolymerization assessment factor.

[0040] Preferably, the mutual information between the fused feature data set in four dimensions and the corresponding thermal depolymerization evaluation factor is calculated as follows:

[0041] The thermal depolymerization evaluation factors are composed into a factor set Y, and the mutual information calculation formula is:

[0042]

[0043] Among them, Ih is the mutual information between the fused feature dataset and the corresponding thermal disaggregation evaluation factor in the hth dimension; F h is the fusion feature data set on the hth dimension; y is a data in the thermal depolymerization evaluation factor set Y; f h is a data on the fusion feature dataset on the hth dimension; p(f h ), p(y) are F h , the marginal probability distribution function of Y; p(f h ,y) is F h , the joint probability distribution function of Y.

[0044] Preferably, the method of comprehensively selecting the fusion feature data set with the greatest influence based on the correlation coefficient and the mutual information is specifically:

[0045] Compute mutual information corrected for the correlation coefficient:

[0046] I h ′=αI h

[0047]

[0048] Among them, I h ′ is the mutual information between the fused feature dataset and the corresponding thermal disaggregation evaluation factor in the hth dimension; α is the correction coefficient;

[0049] I h The fused feature dataset with the largest ′ is the fused feature dataset with the greatest influence.

[0050] Preferably, the loss function used by the BP neural network is:

[0051]

[0052] Where K is the total number of historical data obtained for each environmental parameter, y n is the nth thermal depolymerization assessment factor, y n ′ is the nth evaluation factor predicted by BP neural network.

[0053] The second aspect of the present invention provides a system for detecting the thermal depolymerization of an organic insulating material and avoiding false alarms using the method described in the first aspect of the present invention, comprising: a data acquisition module, a data preprocessing module, a data fusion module, a module for selecting and fusion feature data sets, a prediction model building module, and a prediction module; specifically:

[0054] Data acquisition module: obtains historical data of environmental parameters in the power distribution cabinet. The historical data of each environmental parameter constitutes a historical data set, and evaluates the thermal depolymerization of organic insulating materials corresponding to the historical data. The evaluated data is defined as a thermal depolymerization factor, and the thermal depolymerization evaluation factor is used as a label; and the environmental parameters in the power distribution cabinet are detected in real time;

[0055] Data preprocessing module: adaptively normalize each environmental parameter in the historical data set;

[0056] Data fusion module: The normalized environmental parameters are fused in four dimensions, and the fused data in each dimension are combined into a set to obtain a fused feature data set in four dimensions;

[0057] Select fusion feature dataset module: calculate the mutual information and correlation coefficient of the fusion feature datasets in four dimensions and the corresponding thermal disaggregation evaluation factors respectively, and select the fusion feature dataset with the greatest influence based on the correlation coefficient and mutual information;

[0058] Prediction model building module: establish a prediction model based on BP neural network, and train the model by selecting the most influential fusion feature data set and the corresponding thermal depolymerization evaluation factor;

[0059] Prediction module: Adaptively normalize the environmental parameters in the distribution cabinet detected in real time, and then adopt data fusion of the same dimension as the selected fusion feature data set with the greatest influence. Input the result of data fusion into the prediction model to obtain the predicted thermal depolymerization assessment factor. If it exceeds the set alarm factor, an alarm will be issued.

[0060] The beneficial effect of the present invention is that, compared with the prior art, the patent performs data fusion on the basis of setting a threshold value, does not consider the mutual influence of the seven parameters, and only adopts one data fusion optimization method, which is not universal enough for different distribution cabinets under different conditions.

[0061] The present invention performs data fusion in four dimensions on the normalized environmental parameters through linear weighting of adaptive weights and empirical formulas obtained from three experiments, fully considering the influence between the environmental parameters, and the present invention calculates the mutual information and correlation coefficients of the fused feature data sets in four dimensions and the corresponding thermal depolymerization assessment factors, and comprehensively selects the fused feature data set with the greatest influence according to the correlation coefficient and the mutual information; different distribution stations can use a more suitable data fusion method under different circumstances; the present invention predicts the thermal depolymerization assessment factor through a prediction model of a BP neural network, thereby judging whether an alarm is needed to avoid false alarms; and the algorithm has a small amount of calculation, which improves the efficiency of the alarm. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0063] In order to make the purpose, technical scheme and advantages of the present invention clearer, the technical scheme of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. The embodiments described in this application are only embodiments of a part of the present invention, rather than all embodiments. Based on the spirit of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work belong to the protection scope of the present invention.

[0064] like Figure 1 As shown, Embodiment 1 of the present invention provides a method for detecting thermal depolymerization of an organic insulating material to avoid false alarms, comprising:

[0065] Obtain historical data of environmental parameters in the power distribution cabinet. The historical data of each environmental parameter constitutes a historical data set. The thermal depolymerization of the organic insulating material corresponding to the historical data is evaluated. The evaluated data is defined as a thermal depolymerization factor, and the thermal depolymerization evaluation factor is used as a label.

[0066] Adaptively normalize each environmental parameter in the historical data set;

[0067] The normalized environmental parameters are fused in four dimensions, and the fused data in each dimension are combined into a set to obtain a fused feature data set in four dimensions;

[0068] The mutual information and correlation coefficients of the fused feature data sets and the corresponding thermal disaggregation evaluation factors in the four dimensions are calculated respectively, and the fused feature data set with the greatest influence is selected based on the correlation coefficient and mutual information;

[0069] A prediction model based on BP neural network was established, and the model was trained by selecting the most influential fusion feature data set and the corresponding thermal depolymerization evaluation factor;

[0070] The environmental parameters in the distribution cabinet are detected in real time, and then adaptively normalized. Then, data fusion of the same dimension as the selected fusion feature data set with the greatest influence is adopted. The result of data fusion is input into the prediction model to obtain the predicted thermal depolymerization assessment factor. If it exceeds the set alarm factor, an alarm is issued.

[0071] Preferably, the environmental parameters in the distribution cabinet include temperature, temperature change rate, thermal decomposition gas concentration, thermal decomposition gas concentration change rate, type 1 particle concentration, type 2 particle concentration and type 3 particle concentration; the type 1 particle concentration is the mass concentration value obtained by converting the density of industrial metal particles as equivalent particles; the type 2 particle concentration is the mass concentration value obtained by converting the density of pollutants in the air as equivalent particles; the type 3 particle concentration is the concentration of particles with a diameter of 1.0um to 10.0um.

[0072] Preferably, the adaptive normalization is:

[0073]

[0074] Among them, P i is the set of all historical data of the i-th environmental parameter, p i,j is the jth data of the i-th environmental parameter, p i ' ,j For p i,j The normalized result.

[0075] Preferably, the data fusion of the normalized environmental parameters in four dimensions is specifically as follows:

[0076]

[0077]

[0078]

[0079] Among them, f 1,j 、f 2,j 、f 3,j 、f 4,j are the fusion data of the jth data of all environmental parameters in the first, second, third, and fourth dimensions, respectively, p i ' ,j is the normalized result of the jth data of the ith environmental parameter, i = {1, 2, 3, 4, 5, 6, 7}, and the 1st to 7th environmental parameters are temperature, temperature change rate, thermal decomposition gas concentration, thermal decomposition gas concentration change rate, type 1 particle concentration, type 2 particle concentration, and type 3 particle concentration, respectively; w i,j is the weight of the jth data of the ith environmental parameter; I is the number of types of environmental parameters.

[0080] Preferably, the weight ω of the jth data of the i-th environmental parameter is i,j The selection rules are:

[0081] Calculate the influence of temperature and temperature change rate:

[0082]

[0083]

[0084] where ω 1 ' ,j is the influence of temperature, ω 2 ' ,j is the influence ratio of temperature change rate, p 1,j is the jth data of temperature; p 1 ' ,j is the jth data of the normalized temperature; p 2 ' ,j The normalized temperature change rate is the jth data; K is the total number of historical data of each environmental parameter obtained;

[0085] Calculate the influence ratio of thermal decomposition gas concentration and thermal decomposition gas concentration change rate:

[0086]

[0087]

[0088] where ω 3 ' ,j is the specific gravity affected by the concentration of thermal decomposition gas, ω 4 ' ,j is the specific gravity affected by the change rate of the concentration of the thermal decomposition gas, p 3,j is the jth data of the concentration of thermal decomposition gas; p 3 ' ,j is the jth data of the normalized thermal decomposition gas concentration; p′ 4,j The normalized rate of change of the thermal decomposition gas concentration is the jth data;

[0089] Calculate the influence ratio of type 1 particle concentration, type 2 particle concentration and type 3 particle concentration:

[0090]

[0091] Among them, ω o ' ,j is the influence proportion of the parameter in the oth link; p o ' ,j is the jth data of the oth link parameter after normalization; o∈[5,6,7];

[0092] Normalize the impact ratio of all environmental parameters:

[0093]

[0094] ω i,jis the weight of the jth data of the i-th environmental parameter.

[0095] Preferably, the calculation of the correlation coefficients between the fused feature data sets in the four dimensions and the corresponding thermal depolymerization assessment factors is specifically:

[0096] The correlation coefficient calculation formula is:

[0097]

[0098] Among them, r h is the correlation coefficient between the fused feature dataset on the hth dimension and the corresponding thermal depolymerization assessment factor; K is the total number of historical data obtained for each environmental parameter; f h,j is the fusion data of the jth data of all environmental parameters in the hth dimension; j is the corresponding j-th thermal depolymerization assessment factor.

[0099] Preferably, the mutual information between the fused feature data set in four dimensions and the corresponding thermal depolymerization evaluation factor is calculated as follows:

[0100] The thermal depolymerization evaluation factors are composed into a factor set Y, and the mutual information calculation formula is:

[0101]

[0102] Among them, I h is the mutual information between the fused feature dataset and the corresponding thermal disaggregation evaluation factor in the hth dimension; F h is the fusion feature data set on the hth dimension; y is a data in the thermal depolymerization evaluation factor set Y; f h is a data on the fusion feature dataset on the hth dimension; p(f h ), p(y) are F h , the marginal probability distribution function of Y; p(f h ,y) is F h , the joint probability distribution function of Y.

[0103] Preferably, the method of comprehensively selecting the fusion feature data set with the greatest influence based on the correlation coefficient and the mutual information is specifically:

[0104] Compute mutual information corrected for the correlation coefficient:

[0105] I′ h =αI h

[0106]

[0107] Among them, I h′ is the mutual information between the fused feature dataset and the corresponding thermal disaggregation evaluation factor in the hth dimension; α is the correction coefficient;

[0108] I h The fused feature dataset with the largest ′ is the fused feature dataset with the greatest influence.

[0109] Preferably, the loss function used by the BP neural network is:

[0110]

[0111] Where K is the total number of historical data obtained for each environmental parameter, y n is the nth thermal depolymerization assessment factor, y n ′ is the nth evaluation factor predicted by BP neural network.

[0112] Embodiment 2 of the present invention provides a system for detecting the thermal depolymerization of an organic insulating material and avoiding false alarms using the method described in Embodiment 1 of the present invention, comprising: a data acquisition module, a data preprocessing module, a data fusion module, a module for selecting a fusion feature data set, a prediction model building module, and a prediction module; specifically:

[0113] Data acquisition module: obtains historical data of environmental parameters in the power distribution cabinet. The historical data of each environmental parameter constitutes a historical data set, and evaluates the thermal depolymerization of organic insulating materials corresponding to the historical data. The evaluated data is defined as a thermal depolymerization factor, and the thermal depolymerization evaluation factor is used as a label; and the environmental parameters in the power distribution cabinet are detected in real time;

[0114] Data preprocessing module: adaptively normalize each environmental parameter in the historical data set;

[0115] Data fusion module: The normalized environmental parameters are fused in four dimensions, and the fused data in each dimension are combined into a set to obtain a fused feature data set in four dimensions;

[0116] Select fusion feature dataset module: calculate the mutual information and correlation coefficient of the fusion feature datasets in four dimensions and the corresponding thermal disaggregation evaluation factors respectively, and select the fusion feature dataset with the greatest influence based on the correlation coefficient and mutual information;

[0117] Prediction model building module: establish a prediction model based on BP neural network, and train the model by selecting the most influential fusion feature data set and the corresponding thermal depolymerization evaluation factor;

[0118] Prediction module: Adaptively normalize the environmental parameters in the distribution cabinet detected in real time, and then adopt data fusion of the same dimension as the selected fusion feature data set with the greatest influence. Input the result of data fusion into the prediction model to obtain the predicted thermal depolymerization assessment factor. If it exceeds the set alarm factor, an alarm will be issued.

[0119] The present disclosure may be a system, a method and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for causing a processor to implement various aspects of the present disclosure.

[0120] A computer-readable storage medium may be a tangible device that can hold and store instructions used by an instruction execution device. A computer-readable storage medium may be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. More specific examples of computer-readable storage media (a non-exhaustive list) include: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disk read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanical encoding device, such as a punch card or a raised structure in a groove on which instructions are stored, and any suitable combination of the foregoing. As used herein, a computer-readable storage medium is not to be interpreted as a transient signal per se, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagating through a waveguide or other transmission medium (e.g., a light pulse through a fiber optic cable), or an electrical signal transmitted through a wire.

[0121] The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to each computing / processing device, or downloaded to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include copper transmission cables, optical fiber transmissions, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers. The network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in the computer-readable storage medium in each computing / processing device.

[0122] The computer program instructions for performing the operation of the present disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-related instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages, such as Smalltalk, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. Computer-readable program instructions may be executed completely on a user's computer, partially on a user's computer, as an independent software package, partially on a user's computer, partially on a remote computer, or completely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., using an Internet service provider to connect via the Internet). In some embodiments, an electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA), or a programmable logic array (PLA), may be customized by utilizing the state information of the computer-readable program instructions, and the electronic circuit may execute the computer-readable program instructions, thereby realizing various aspects of the present disclosure.

[0123] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for detecting thermal depolymerization of an organic insulating material to avoid false alarms, characterized in that: include: Obtain historical data of environmental parameters in the power distribution cabinet. The historical data of each environmental parameter constitutes a historical data set. The thermal depolymerization of the organic insulating material corresponding to the historical data is evaluated. The evaluated data is defined as a thermal depolymerization factor, and the thermal depolymerization evaluation factor is used as a label. Adaptively normalize each environmental parameter in the historical data set; The normalized environmental parameters are fused in four dimensions, and the fused data in each dimension are combined into a set to obtain a fused feature data set in four dimensions; The mutual information and correlation coefficients of the fused feature data sets and the corresponding thermal depolymerization evaluation factors in the four dimensions are calculated respectively, and the fused feature data set with the greatest influence is selected based on the correlation coefficient and mutual information; A prediction model based on BP neural network was established, and the model was trained by selecting the most influential fusion feature data set and the corresponding thermal depolymerization evaluation factor; The environmental parameters in the distribution cabinet are detected in real time, and then adaptively normalized. Then, data fusion of the same dimension as the selected fusion feature data set with the greatest influence is adopted. The result of data fusion is input into the prediction model to obtain the predicted thermal depolymerization assessment factor. If it exceeds the set alarm factor, an alarm is issued.

2. A method for detecting thermal depolymerization of an organic insulating material to avoid false alarms according to claim 1, characterized in that: The environmental parameters in the distribution cabinet include temperature, temperature change rate, thermal decomposition gas concentration, thermal decomposition gas concentration change rate, type 1 particle concentration, type 2 particle concentration and type 3 particle concentration; the type 1 particle concentration is the mass concentration value obtained by converting the density of industrial metal particles as equivalent particles; the type 2 particle concentration is the mass concentration value obtained by converting the density of pollutants in the air as equivalent particles; the type 3 particle concentration is the concentration of particles with a diameter of 1.0um to 10.0um.

3. A method for detecting thermal depolymerization of an organic insulating material and avoiding false alarms according to claim 2, characterized in that: The adaptive normalization is: Among them, P i is the set of all historical data of the i-th environmental parameter, p i,j is the jth data of the i-th environmental parameter, p i ' ,j For p i,j The normalized result.

4. A method for detecting thermal depolymerization of an organic insulating material and avoiding false alarms according to claim 3, characterized in that: The data fusion of the normalized environmental parameters in four dimensions is specifically as follows: Among them, f 1,j 、f 2,j 、f 3,j 、f 4,j are the fusion data of the jth data of all environmental parameters in the first, second, third, and fourth dimensions, respectively, p′ i,j is the normalized result of the jth data of the ith environmental parameter, i = {1, 2, 3, 4, 5, 6, 7}, and the 1st to 7th environmental parameters are temperature, temperature change rate, thermal decomposition gas concentration, thermal decomposition gas concentration change rate, type 1 particle concentration, type 2 particle concentration, and type 3 particle concentration, respectively; w i,j is the weight of the jth data of the ith environmental parameter; I is the number of types of environmental parameters.

5. A method for detecting thermal depolymerization of an organic insulating material and avoiding false alarms according to claim 4, characterized in that: The weight ω of the jth data of the i-th environmental parameter i,j The selection rules are: Calculate the influence of temperature and temperature change rate: where ω1′ ,j is the influence of temperature on specific gravity, ω2′ ,j is the influence ratio of temperature change rate, p 1,j is the jth temperature data; p1′ ,j is the jth data of normalized temperature; p2′ ,j The normalized temperature change rate is the jth data; K is the total number of historical data of each environmental parameter obtained; Calculate the influence ratio of thermal decomposition gas concentration and thermal decomposition gas concentration change rate: where ω3′ ,j The specific gravity and ω4′ are affected by the concentration of thermal decomposition gas. ,j is the specific gravity affected by the change rate of the concentration of the thermal decomposition gas, p 3,j is the jth data of the concentration of the thermal decomposition gas; p3′ ,j is the jth data of the normalized thermal decomposition gas concentration; p′ 4,j The normalized rate of change of the thermal decomposition gas concentration is the jth data; Calculate the influence ratio of type 1 particle concentration, type 2 particle concentration and type 3 particle concentration: Among them, ω o ' ,j is the influence ratio of the parameter in the oth link; p o ' ,j is the jth data of the oth link parameter after normalization; o∈[5,6,7]; Normalize the impact ratio of all environmental parameters: ω i,j is the weight of the jth data of the i-th environmental parameter.

6. A method for detecting thermal depolymerization of an organic insulating material and avoiding false alarms according to claim 5, characterized in that: The calculation of the correlation coefficients between the fused feature data sets in the four dimensions and the corresponding thermal depolymerization evaluation factors is specifically: The correlation coefficient calculation formula is: Among them, r h is the correlation coefficient between the fused feature dataset on the hth dimension and the corresponding thermal depolymerization assessment factor; K is the total number of historical data obtained for each environmental parameter; f h,j is the fusion data of the jth data of all environmental parameters in the hth dimension; j is the corresponding j-th thermal depolymerization assessment factor.

7. A method for detecting thermal depolymerization of an organic insulating material and avoiding false alarms according to claim 5, characterized in that: The mutual information between the fused feature data set and the corresponding thermal disaggregation evaluation factor in the four dimensions is calculated as follows: The thermal depolymerization evaluation factors are composed into a factor set Y, and the mutual information calculation formula is: Among them, I h is the mutual information between the fused feature dataset and the corresponding thermal disaggregation evaluation factor in the hth dimension; F h is the fusion feature data set on the hth dimension; y is a data in the thermal depolymerization evaluation factor set Y; f h is a data on the fusion feature dataset in the hth dimension; p(f h ), p(y) are F h , the marginal probability distribution function of Y; p(f h ,y) is F h , the joint probability distribution function of Y.

8. A method for detecting thermal depolymerization of an organic insulating material and avoiding false alarms according to any one of claims 6 or 7, characterized in that: The method of comprehensively selecting the most influential fusion feature data set based on the correlation coefficient and mutual information is specifically: Compute mutual information corrected for the correlation coefficient: I h ′=αI h Among them, I h ′ is the mutual information between the fused feature dataset and the corresponding thermal disaggregation evaluation factor in the hth dimension; α is the correction coefficient; I h The fused feature dataset with the largest ′ is the fused feature dataset with the greatest influence.

9. The method for detecting thermal depolymerization of an organic insulating material and avoiding false alarms according to claim 1, characterized in that: The loss function used by the BP neural network is: Among them, K is the total number of historical data of each environmental parameter obtained, y n is the nth thermal depolymerization assessment factor, y n ′ is the nth evaluation factor predicted by BP neural network.

10. A system for detecting thermal depolymerization of an organic insulating material and avoiding false alarms using the method according to any one of claims 1 to 9, comprising: Data acquisition module, data preprocessing module, data fusion module, fusion feature data set selection module, prediction model construction module, prediction module; characterized by: Data acquisition module: obtains historical data of environmental parameters in the power distribution cabinet. The historical data of each environmental parameter constitutes a historical data set, and evaluates the thermal depolymerization of organic insulating materials corresponding to the historical data. The evaluated data is defined as a thermal depolymerization factor, and the thermal depolymerization evaluation factor is used as a label; and the environmental parameters in the power distribution cabinet are detected in real time; Data preprocessing module: adaptively normalize each environmental parameter in the historical data set; Data fusion module: The normalized environmental parameters are fused in four dimensions, and the fused data in each dimension are combined into a set to obtain a fused feature data set in four dimensions; Select fusion feature dataset module: calculate the mutual information and correlation coefficient of the fusion feature datasets in four dimensions and the corresponding thermal disaggregation evaluation factors respectively, and select the fusion feature dataset with the greatest influence based on the correlation coefficient and mutual information; Prediction model building module: establish a prediction model based on BP neural network, and train the model by selecting the most influential fusion feature data set and the corresponding thermal depolymerization evaluation factor; Prediction module: Adaptively normalize the environmental parameters in the distribution cabinet detected in real time, and then adopt data fusion of the same dimension as the selected fusion feature data set with the greatest influence. Input the result of data fusion into the prediction model to obtain the predicted thermal depolymerization assessment factor. If it exceeds the set alarm factor, an alarm will be issued.

Images