Distributed power distribution cabinet fire alarm system based on multi-source data fusion

Abstract

The invention relates to the technical field of fire alarm, and particularly discloses a distributed power distribution cabinet fire alarm system based on multi-source data fusion, which adopts a time sequence analysis technology based on deep learning to carry out time sequence modeling on the temperature in a power distribution cabinet, the environment temperature and load current data so as to capture the time sequence change trend of the temperature, the environment temperature and the load current data; based on the power distribution cabinet environment temperature time sequence characteristics, environment interference filtering is carried out on the cabinet internal temperature time sequence characteristics to obtain pure characterization of the cabinet internal temperature time sequence change mode, and then time sequence interaction response excitation analysis is carried out on the power distribution cabinet internal temperature time sequence change characteristics after load current and environment interference filtering. Therefore, the intelligent early warning prompt of the fire risk is realized based on the time sequence correlation response mode between the load and the temperature of the power distribution cabinet. Therefore, the accuracy of fire alarm can be effectively improved, the situations of false alarm and missing alarm are reduced, and the safe operation of the power distribution cabinet is ensured.

Description

Technical Field

[0001] The present application relates to the field of fire alarm technology, and more specifically, to a distributed distribution cabinet fire alarm system based on multi-source data fusion. Background Art

[0002] In modern power systems, the distribution cabinet is a key node in power transmission and distribution, and its operation safety is directly related to the reliability of power supply and the safety of users' lives and property. However, due to the dense electrical equipment and complex loads inside the distribution cabinet, and the long-term operation state, it is very easy to cause fires due to overload, short circuit, poor contact, etc. Once a distribution cabinet fire occurs, it will not only cause power supply interruption and affect the normal order of production and life, but may also cause significant property losses and casualties, posing a serious threat to social economy and security. Traditional distribution cabinet fire alarm methods mainly rely on single sensor data, such as temperature sensors, smoke sensors, etc., which have obvious limitations and cannot fully and accurately reflect the complex operating status inside the distribution cabinet. For example, the method based on the temperature threshold inside the cabinet triggers the alarm. Since the temperature change inside the distribution cabinet is affected by multiple interferences such as ambient temperature fluctuations and heat dissipation conditions, single temperature monitoring is prone to false alarms (such as false triggering in high temperature environments in summer) or missed alarms (such as delayed temperature rise at the beginning of overload); and smoke sensors can only play a role after the fire develops to a certain stage and produces smoke, making it difficult to achieve early warning. In addition, as one of the important indicators of the operating status of the distribution cabinet, the abnormal changes in load current data are often early signals of electrical faults. For example, when overloaded, the current will continue to exceed the rated value, and the instantaneous current of short circuit will soar sharply. However, under the single sensor alarm system, the load current data and the temperature data inside the distribution cabinet are isolated from each other, and an effective correlation analysis model has not been established, making it difficult to improve the accuracy and timeliness of the alarm through comprehensive analysis of multiple parameters.

[0003] Therefore, a distributed distribution cabinet fire alarm system based on multi-source data fusion is expected. Summary of the invention

[0004] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a distributed distribution cabinet fire alarm system based on multi-source data fusion, which uses a timing analysis technology based on deep learning to perform timing modeling on the temperature, ambient temperature and load current data in the distribution cabinet to capture the timing change trend of the three. Then, based on the timing characteristics of the ambient temperature of the distribution cabinet, the environmental interference of the internal temperature timing characteristics of the cabinet is filtered out to obtain a pure representation of the timing change pattern of the internal temperature of the cabinet, and then the timing interaction response excitation analysis is performed on the timing change characteristics of the internal temperature of the distribution cabinet after filtering the load current and environmental interference, so as to realize intelligent early warning prompts for fire risks based on the timing correlation response mode between the load and temperature of the distribution cabinet. In this way, the accuracy of fire alarms can be effectively improved, false alarms and missed alarms can be reduced, and the safe operation of the distribution cabinet can be ensured.

[0005] Accordingly, according to one aspect of the present application, a distributed power distribution cabinet fire alarm system based on multi-source data fusion is provided, comprising: The distribution cabinet data acquisition module is used to obtain the distribution cabinet internal temperature time series data set, the distribution cabinet ambient temperature time series data set and the distribution cabinet load current time series data set collected by the sensor group; A distribution cabinet data timing analysis module, used for performing time series encoding on the distribution cabinet internal temperature timing data set, the distribution cabinet ambient temperature timing data set and the distribution cabinet load current timing data set to obtain the distribution cabinet internal temperature timing implicit features, the distribution cabinet ambient temperature timing implicit features and the distribution cabinet load current timing implicit features; An ambient temperature interference filtering module is used to filter the ambient temperature interference of the temperature time series implicit characteristics in the power distribution cabinet based on the ambient temperature time series implicit characteristics of the power distribution cabinet to obtain the environmental factor filtering cabinet temperature time series implicit characteristics; The fire warning analysis module is used to perform load-temperature response excitation analysis on the environmental factors filtering out the implicit characteristics of the temperature timing in the cabinet and the implicit characteristics of the load current timing of the distribution cabinet to determine whether to generate a fire warning prompt signal.

[0006] In the above-mentioned distributed distribution cabinet fire alarm system based on multi-source data fusion, the distribution cabinet data time series analysis module is used to: The temperature time series data set inside the distribution cabinet, the ambient temperature time series data set of the distribution cabinet, and the load current time series data set of the distribution cabinet are respectively time-series encoded based on the time series convolutional neural network model to obtain the implicit time series features of the temperature inside the distribution cabinet, the implicit time series features of the ambient temperature of the distribution cabinet, and the implicit time series features of the load current of the distribution cabinet.

[0007] In the above-mentioned distributed power distribution cabinet fire alarm system based on multi-source data fusion, the ambient temperature interference filtering module is used to: The temperature time series difference feature between the temperature time series implicit feature in the power distribution cabinet and the temperature time series implicit feature of the ambient temperature of the power distribution cabinet is calculated to obtain the temperature time series implicit feature in the cabinet after filtering out the environmental factors.

[0008] In the above-mentioned distributed power distribution cabinet fire alarm system based on multi-source data fusion, the fire warning analysis module includes: A load temperature response excitation analysis unit, used for performing load-temperature response excitation analysis based on inter-modal decoupling constraints on the environmental factor-filtered cabinet temperature time series implicit features and the distribution cabinet load current time series implicit features to obtain a load-cabinet temperature excitation response time series encoding vector; The warning analysis result generating unit is used to input the load-cabinet temperature excitation response time series coding vector into the classifier-based fire warning module to obtain a warning analysis result, and the warning analysis result is used to indicate whether a fire warning prompt signal is generated.

[0009] In the above-mentioned distributed power distribution cabinet fire alarm system based on multi-source data fusion, the load temperature response excitation analysis unit includes: A feature fine-grained interactive response analysis subunit, used to perform feature interactive response analysis based on principal component decomposition on the environmental factor-filtered cabinet temperature time series implicit features and the distribution cabinet load current time series implicit features to obtain a set of fine-grained response interactive coding vectors between load-cabinet temperature time series principal component modes; An information independence constraint subunit is used to filter out the information independence between the implicit characteristics of the cabinet temperature time series and the implicit characteristics of the load current time series of the distribution cabinet based on the environmental factors, and calculate the independence soft constraint factors of the fine-grained response interaction coding vectors between the load-cabinet temperature time series principal component modes in the set of fine-grained response interaction coding vectors between the load-cabinet temperature time series principal component modes to obtain a set of soft constraint factors for independence between the load-cabinet temperature time series principal component modes; An adaptive aggregation subunit is used to adaptively aggregate a set of fine-grained response interaction coding vectors between the load-cabinet temperature time series principal component modes based on a set of independence soft constraint factors between the load-cabinet temperature time series principal component modes to obtain the load-cabinet temperature excitation response time series coding vector.

[0010] In the above-mentioned distributed power distribution cabinet fire alarm system based on multi-source data fusion, the characteristic fine-grained interactive response analysis subunit is used to: Performing principal component analysis on the implicit characteristics of the load current time series of the distribution cabinet to obtain a set of principal component characteristic encoding vectors of the load current time series of the distribution cabinet; Performing a linear transformation on each distribution cabinet load current time series principal component feature coding vector in the set of the distribution cabinet load current time series principal component feature coding vectors so that the vectors have the same feature scale as the environmental factor filtered cabinet temperature time series implicit feature to obtain a set of distribution cabinet load current time series principal component linear transformation feature coding vectors; The environmental factors are filtered out, the implicit features of the cabinet temperature time series, and each distribution cabinet load current time series principal component linear transformation feature coding vector in the set of the distribution cabinet load current time series principal component linear transformation feature coding vectors are input into the feature interactive response unit to obtain a set of fine-grained response interactive coding vectors between the load and cabinet temperature time series principal component modes.

[0011] In the above-mentioned distributed power distribution cabinet fire alarm system based on multi-source data fusion, the information independence constraint subunit includes: The second-level sub-unit of independence modeling is used to input the characteristic coding vectors of the principal component linear transformation of the load current time series of the distribution cabinet in the set of characteristic coding vectors of the principal component linear transformation of the load current time series of the distribution cabinet after filtering out the implicit characteristic of the time series of the temperature in the cabinet by the environmental factors into the inter-modal independence modeling unit to obtain the set of inter-modal independence coding matrices of the principal component of the load-cabinet temperature time series; The second-level subunit for calculating the independence soft constraint factor is used to calculate the load-cabinet temperature time series principal component modal independence soft constraint factor based on the characteristic distribution of each load-cabinet temperature time series principal component modal independence coding matrix in the set of the load-cabinet temperature time series principal component modal independence coding matrix to obtain the set of the load-cabinet temperature time series principal component modal independence soft constraint factor.

[0012] In the above-mentioned distributed power distribution cabinet fire alarm system based on multi-source data fusion, the independence modeling secondary subunit is used to: Performing nonlinear activation processing on the implicit characteristics of the temperature time series of the cabinet after the environmental factors are filtered out and the linear transformation characteristic coding vector of the main component of the load current time series of the distribution cabinet to obtain the activation coding vector of the temperature time series characteristics of the cabinet after the environmental factors are filtered out and the activation coding vector of the main component characteristics of the load current time series of the distribution cabinet; Calculate the product of the transposed vector of the load current time series principal component feature activation coding vector of the distribution cabinet and the environmental factor filtered cabinet temperature time series feature activation coding vector, and divide the product result by the square root of the characteristic scale value of the environmental factor filtered cabinet temperature time series feature activation coding vector to obtain the load-cabinet temperature time series principal component modal independence coding matrix.

[0013] In the above-mentioned distributed power distribution cabinet fire alarm system based on multi-source data fusion, the independence soft constraint factor calculation secondary subunit is used to: Performing topological structure stability optimization on the load-cabinet temperature time series principal component modal independence coding matrix to obtain an optimized load-cabinet temperature time series principal component modal independence coding matrix; The square of the F-norm of the optimized load-cabinet temperature time series principal component modal independence encoding matrix is ​​calculated as the load-cabinet temperature time series principal component modal independence soft constraint factor.

[0014] Compared with the prior art, the distributed distribution cabinet fire alarm system based on multi-source data fusion provided by the present application uses a time series analysis technology based on deep learning to perform time series modeling on the temperature, ambient temperature and load current data in the distribution cabinet to capture the time series change trend of the three. Then, based on the time series characteristics of the ambient temperature of the distribution cabinet, the environmental interference of the cabinet internal temperature time series characteristics is filtered out to obtain a pure representation of the cabinet internal temperature time series change pattern, and then the time series interactive response excitation analysis is performed on the time series change characteristics of the internal temperature of the distribution cabinet after the load current and environmental interference are filtered out, thereby realizing intelligent early warning prompts for fire risks based on the time series correlation response mode between the load and temperature of the distribution cabinet. In this way, the accuracy of fire alarms can be effectively improved, false alarms and missed alarms can be reduced, and the safe operation of the distribution cabinet can be ensured. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0016] Figure 1 It is a block diagram of a distributed distribution cabinet fire alarm system based on multi-source data fusion according to an embodiment of the present application.

[0017] Figure 2 Schematic diagram of data flow of a distributed power distribution cabinet fire alarm system based on multi-source data fusion according to an embodiment of the present application.

[0018] Figure 3 It is a block diagram of a fire warning analysis module in a distributed distribution cabinet fire alarm system based on multi-source data fusion according to an embodiment of the present application.

[0019] Figure 4 The present invention is a block diagram of a load temperature response excitation analysis unit in a distributed power distribution cabinet fire alarm system based on multi-source data fusion according to an embodiment of the present application.

[0020] Figure 5It is a block diagram of an information independence constraint subunit in a distributed power distribution cabinet fire alarm system based on multi-source data fusion according to an embodiment of the present application. DETAILED DESCRIPTION

[0021] As shown in this application and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "include" and "comprise" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0022] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0023] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. Meanwhile, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0024] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0025] It is worth noting that in this application, all actions to obtain data are carried out in compliance with the relevant data protection laws and policies of the country where the data is located, and with the authorization given by the owner of the corresponding device.

[0026] In view of the technical problems described in the above background art, the present application proposes a distributed power distribution cabinet fire alarm system based on multi-source data fusion. It uses a time series analysis technology based on deep learning to perform time series modeling on the temperature inside the power distribution cabinet, the ambient temperature, and the load current data to capture the time series change trends of the three. Then, based on the time series characteristics of the ambient temperature of the power distribution cabinet, the time series characteristics of the temperature inside the cabinet are filtered to remove environmental interference, so as to obtain a pure representation of the time series change pattern of the temperature inside the cabinet. Furthermore, through the time series interaction response excitation analysis of the time series change characteristics of the temperature inside the power distribution cabinet after filtering the load current and environmental interference, an intelligent early warning prompt for fire risk is realized based on the time series correlation response pattern between the load and temperature of the power distribution cabinet. In this way, the accuracy of fire alarm can be effectively improved, the situation of false alarms and missed alarms can be reduced, and the safe operation of the power distribution cabinet can be ensured.

[0027] Figure 1 It is a block diagram of a distributed power distribution cabinet fire alarm system based on multi-source data fusion according to an embodiment of the present application. Figure 2 It is a schematic diagram of data flow of a distributed power distribution cabinet fire alarm system based on multi-source data fusion according to an embodiment of the present application. As Figure 1 and Figure 2 shown, the distributed power distribution cabinet fire alarm system 100 based on multi-source data fusion includes: a power distribution cabinet data acquisition module 110, configured to acquire a time series data set of the temperature inside the power distribution cabinet, a time series data set of the ambient temperature of the power distribution cabinet, and a time series data set of the load current of the power distribution cabinet collected by a sensor group; a power distribution cabinet data time series analysis module 120, configured to perform time series encoding on the time series data set of the temperature inside the power distribution cabinet, the time series data set of the ambient temperature of the power distribution cabinet, and the time series data set of the load current of the power distribution cabinet to obtain a time series hidden feature of the temperature inside the power distribution cabinet, a time series hidden feature of the ambient temperature of the power distribution cabinet, and a time series hidden feature of the load current of the power distribution cabinet; an ambient temperature interference filtering module 130, configured to filter the environmental temperature interference from the time series hidden feature of the temperature inside the power distribution cabinet based on the time series hidden feature of the ambient temperature of the power distribution cabinet to obtain a time series hidden feature of the temperature inside the cabinet with environmental factors filtered; a fire warning analysis module 140, configured to perform load-temperature response excitation analysis on the time series hidden feature of the temperature inside the cabinet with environmental factors filtered and the time series hidden feature of the load current of the power distribution cabinet to determine whether a fire warning prompt signal is generated.

[0028] In the above-mentioned distributed distribution cabinet fire alarm system based on multi-source data fusion, the distribution cabinet data acquisition module 110 is used to obtain the temperature time series data set inside the distribution cabinet, the distribution cabinet ambient temperature time series data set and the distribution cabinet load current time series data set collected by the sensor group. It should be understood that the traditional distribution cabinet fire alarm relies on a single sensor (such as temperature or smoke), which is susceptible to environmental interference (such as false alarms caused by high temperatures in summer) or response lag (such as smoke requires the fire to develop to a certain stage). In response to this limitation, the present application adopts a distributed architecture to deploy temperature sensors and current sensors at key heating parts inside the distribution cabinet (such as circuit breakers and transformer joints) to comprehensively and real-time monitor the temperature and load current data of key parts inside the distribution cabinet; at the same time, an ambient temperature sensor is deployed outside the distribution cabinet to monitor the ambient temperature of the distribution cabinet. By synchronously collecting temperature data, load current data and ambient temperature data of key parts inside the distribution cabinet, we can achieve comprehensive perception of the complex operating status inside the distribution cabinet, and provide full-dimensional input covering electrical, direct risk indicators and environmental interference for subsequent fire alarm modeling and analysis, which helps to more effectively capture the precursor signals of fire risks.

[0029] In the specific implementation process, in order to achieve effective monitoring of the internal temperature of the distribution cabinet, high-precision temperature sensors need to be deployed in key heating parts inside the distribution cabinet, such as circuit breakers, transformer joints, etc. These sensors can not only monitor the temperature changes of key points inside the distribution cabinet in real time, but also record this information in the form of time series data to form a time series data set of the temperature inside the distribution cabinet. This process is crucial to capturing the changing trend of the temperature inside the distribution cabinet, especially in the early stage of overload or short circuit fault, when the temperature rise has not yet reached the threshold to trigger the traditional alarm. This temperature monitoring is particularly critical. In addition, in order to further improve the accuracy of temperature data, it is possible to consider using temperature sensors of different types or different sensitivities for cross-validation, thereby effectively reducing the possibility of false alarms.

[0030] The ambient temperature sensor is responsible for collecting the temperature changes of the environment in which the distribution cabinet is located and generating a time series data set of the ambient temperature of the distribution cabinet. Considering that the internal and external temperatures of the distribution cabinet may be affected by many factors, such as seasonal changes, temperature differences between day and night, and surrounding heat sources, accurate measurement of the ambient temperature helps the subsequent data processing stage to accurately distinguish whether the temperature changes inside the distribution cabinet are caused by the operation of internal electrical equipment or by the influence of the external environment. This step is of great significance for filtering out ambient temperature interference and obtaining a purer internal temperature feature of the distribution cabinet, which helps to more effectively identify potential safety hazards inside the distribution cabinet.

[0031] In addition, in the specific implementation process, in order to achieve effective monitoring of the load current of the distribution cabinet, current sensors need to be installed on the key circuit nodes of the distribution cabinet. These current sensors can monitor the load current changes of the distribution cabinet in real time and generate a time series data set of the load current of the distribution cabinet. As one of the important indicators of the operating status of the distribution cabinet, the abnormal changes of the load current data are often early signals of electrical faults. For example, in the case of overload, the current will continue to exceed the rated value; and at the moment of short circuit, the current will soar sharply. By real-time monitoring and recording of these current changes, not only can potential electrical faults be discovered in time, but also an important basis for subsequent fire warnings can be provided. It is worth noting that when selecting current sensors, the actual working environment and current range of the distribution cabinet should be fully considered to ensure the measurement accuracy and stability of the sensor so as to accurately reflect the operating status of the distribution cabinet.

[0032] In the above-mentioned distributed distribution cabinet fire alarm system based on multi-source data fusion, the distribution cabinet data timing analysis module 120 is used to perform time series encoding on the temperature timing data set in the distribution cabinet, the ambient temperature timing data set in the distribution cabinet, and the load current timing data set in the distribution cabinet to obtain the implicit timing characteristics of the temperature in the distribution cabinet, the implicit timing characteristics of the ambient temperature of the distribution cabinet, and the implicit timing characteristics of the load current of the distribution cabinet. In a specific example of the present application, the distribution cabinet data timing analysis module 120 is used to: perform time series encoding based on the timing convolutional neural network model on the temperature timing data set in the distribution cabinet, the ambient temperature timing data set in the distribution cabinet, and the load current timing data set in the distribution cabinet to obtain the implicit timing characteristics of the temperature in the distribution cabinet, the implicit timing characteristics of the ambient temperature of the distribution cabinet, and the implicit timing characteristics of the load current of the distribution cabinet. It should be understood that the present application takes into account that the traditional threshold comparison method or statistical feature (such as mean, slope) extraction method is often difficult to effectively capture the dynamic change trend of data when processing time series data, and is susceptible to noise interference. Therefore, the present application adopts a time series convolutional neural network model based on deep learning to extract features from the time series data set of the temperature in the distribution cabinet, the time series data set of the ambient temperature of the distribution cabinet, and the time series data set of the load current of the distribution cabinet. A person of ordinary skill in the art should know that the temporal convolutional neural network (TCN) model is a deep learning model specially designed for processing time series data. It captures local features in the data through convolution operations, and uses causal convolution and dilated convolution structures to ensure that the model can capture long-distance dependencies when processing time series data, thereby effectively extracting deep features from the time series data of the temperature in the distribution cabinet, the ambient temperature, and the load current. Specifically, taking the temperature time series data set in the distribution cabinet as an example, the TCN model processes the input temperature time series data set in the distribution cabinet by sliding a fixed-size convolution kernel in the time dimension, performing convolution operations with a continuous temperature value at each time step, integrating local temperature information to extract features such as temperature rising trends and short-term temperature fluctuation cycles, and obtaining implicit time series features of the temperature in the distribution cabinet. Similarly, for the distribution cabinet ambient temperature time series data set and the distribution cabinet load current time series data set, the TCN model can capture the time series variation characteristics of the ambient temperature (such as day and night temperature difference changes) and the time series variation characteristics of the load current (such as periodic load fluctuations, abnormal current peaks, etc.) through similar processing procedures, thereby obtaining the distribution cabinet ambient temperature time series implicit features and the distribution cabinet load current time series implicit features, providing key feature inputs for subsequent fire risk analysis.

[0033] In the above-mentioned distributed distribution cabinet fire alarm system based on multi-source data fusion, the ambient temperature interference filtering module 130 is used to filter the ambient temperature interference of the temperature time series implicit characteristics in the distribution cabinet based on the ambient temperature time series implicit characteristics of the distribution cabinet to obtain the environmental factor filtering cabinet temperature time series implicit characteristics. In a specific example of the present application, the ambient temperature interference filtering module 130 is used to: calculate the temperature time series difference characteristics between the implicit time series characteristics of the temperature in the distribution cabinet and the implicit time series characteristics of the ambient temperature of the distribution cabinet to obtain the environmental factor filtering cabinet temperature time series implicit characteristics. Specifically, considering that in the actual operating environment, the temperature in the distribution cabinet is significantly affected by the ambient temperature. Taking the outdoor distribution cabinet in summer as an example, the ambient temperature may be as high as 40°C or above. Even if the internal operation of the distribution cabinet is normal, its internal temperature will increase due to environmental heat conduction. If the fire risk is judged only based on the temperature change in the cabinet, it is very easy to make a misjudgment in this case. Therefore, in order to separate the part caused by environmental factors from the temperature change in the cabinet and obtain the temperature change characteristics that purely reflect the internal operating state of the distribution cabinet, the present application calculates the temperature time series difference characteristics between the implicit characteristics of the temperature time series in the distribution cabinet and the implicit characteristics of the ambient temperature time series of the distribution cabinet to eliminate the interference of ambient temperature fluctuations on the temperature in the distribution cabinet. It should be understood that when the temperature in the cabinet rises due to the increase in ambient temperature, at the implicit feature level, the change trends of the two in certain dimensions are similar, for example, both are manifested as an increase in numerical values. Through differential operations, this similar change part caused by environmental factors can be eliminated from the implicit characteristics of the temperature in the cabinet, and the temperature change representation caused only by internal factors such as the operation of electrical equipment inside the distribution cabinet and the heating of contact resistance is obtained, that is, the environmental factors filter out the implicit characteristics of the temperature time series in the cabinet, thereby improving the fire warning system's perception of the actual operating conditions inside the distribution cabinet and improving the accuracy and reliability of fire warnings.

[0034] In the above-mentioned distributed power distribution cabinet fire alarm system based on multi-source data fusion, the fire warning analysis module 140 is used to perform load-temperature response excitation analysis on the environmental factors, removing the implicit characteristics of the cabinet temperature time series and the implicit characteristics of the load current time series of the power distribution cabinet, to determine whether to generate a fire warning prompt signal. Figure 3 FIG. 1 is a block diagram of a fire warning analysis module in a distributed power distribution cabinet fire alarm system based on multi-source data fusion according to an embodiment of the present application. Figure 3As shown, the fire warning analysis module 140 includes: a load temperature response excitation analysis unit 141, which is used to perform a load-temperature response excitation analysis based on the inter-modal decoupling constraint on the environmental factors to filter out the implicit characteristics of the cabinet temperature timing and the implicit characteristics of the load current timing of the distribution cabinet to obtain a load-cabinet temperature excitation response timing coding vector; a warning analysis result generation unit 142, which is used to input the load-cabinet temperature excitation response timing coding vector into a classifier-based fire warning module to obtain a warning analysis result, and the warning analysis result is used to indicate whether a fire warning prompt signal is generated.

[0035] Specifically, the load temperature response excitation analysis unit 141 is used to perform load-temperature response excitation analysis based on inter-modal decoupling constraints on the environmental factors to filter out the implicit characteristics of the cabinet temperature timing and the implicit characteristics of the load current timing of the distribution cabinet to obtain a load-cabinet temperature excitation response timing encoding vector. It should be understood that in the actual operation of the distribution cabinet, there is a close and complex dynamic relationship between the load current and the internal temperature. When the load current is overloaded, the current passing through the electrical equipment increases. According to Joule's law, the heating power of the equipment increases significantly, which leads to a rapid increase in the temperature inside the distribution cabinet. However, traditional feature fusion methods (such as feature splicing) are difficult to capture the nonlinear dynamic response between load current and temperature (such as the current rises faster than the temperature rise in the initial overload period, and the temperature rise lags and extends when the heat dissipation is poor), and noise may be introduced due to feature redundancy. In this regard, this application proposes a load-temperature response excitation analysis method based on inter-modal decoupling constraints to capture the complex coupling relationship between load current and temperature data, and introduces a soft constraint mechanism to make the implicit characteristics of load current and the implicit characteristics of cabinet temperature after filtering out environmental factors maintain inter-modal independence during the interaction process, thereby effectively avoiding feature redundancy and noise interference, and deeply mining the interactive response pattern between load current and temperature in the distribution cabinet that changes over time, providing more comprehensive and in-depth information basis for fire risk warning. Among them, Figure 4 FIG. 1 is a block diagram of a load temperature response excitation analysis unit in a distributed power distribution cabinet fire alarm system based on multi-source data fusion according to an embodiment of the present application. Figure 4As shown, the load temperature response excitation analysis unit 141 includes: a feature fine-grained interactive response analysis subunit 1411, which is used to perform feature interactive response analysis based on principal component decomposition on the implicit characteristics of the temperature time series in the cabinet filtered by the environmental factors and the implicit characteristics of the load current time series of the distribution cabinet to obtain a set of fine-grained response interactive coding vectors between the main component modes of the load-cabinet temperature time series; an information independence constraint subunit 1412, which is used to calculate the load-cabinet based on the information independence between the implicit characteristics of the temperature time series in the cabinet filtered by the environmental factors and the implicit characteristics of the load current time series of the distribution cabinet. The independence soft constraint factors of the fine-grained response interaction coding vectors between the main component modes of the internal temperature time series of each load-cabinet are used to obtain a set of soft constraint factors for independence between the main component modes of the load-cabinet temperature time series; an adaptive aggregation subunit 1413 is used to adaptively aggregate the set of fine-grained response interaction coding vectors between the main component modes of the load-cabinet temperature time series based on the set of soft constraint factors for independence between the main component modes of the load-cabinet temperature time series to obtain the load-cabinet temperature excitation response timing coding vector.

[0036] More specifically, the feature fine-grained interactive response analysis subunit 1411 is used to perform principal component analysis on the implicit features of the load current time series of the distribution cabinet to obtain a set of principal component feature encoding vectors of the load current time series of the distribution cabinet, which is expressed as follows:

[0037] in, Indicates the implicit characteristics of the load current timing of the distribution cabinet, represents the principal component analysis function, represents the transpose of a vector, for The characteristic scale value of express The covariance matrix of Indicates that The matrix composed of the set of principal component characteristic coding vectors of the load current time series of the distribution cabinet obtained by eigenvalue decomposition is Indicates that by The diagonal matrix composed of the set of principal component eigenvalues ​​of the load current time series of the distribution cabinet obtained by eigenvalue decomposition is , , represents each distribution cabinet load current time series principal component eigenvalue in the set of distribution cabinet load current time series principal component eigenvalues, is the number of principal component eigenvalues ​​of the load current time series of the distribution cabinet, , , Represents each distribution cabinet load current time series principal component feature coding vector in the set of distribution cabinet load current time series principal component feature coding vectors.

[0038] That is, considering that the time series changes of the load current of the distribution cabinet are forward-looking and driving for temperature changes, therefore, in order to more finely model the excitation response mechanism of the load current to temperature, the present application projects the implicit feature space of the load current time series of the distribution cabinet to a low-dimensional principal component axis system formed by the eigenvectors of the covariance matrix through orthogonal transformation, thereby constructing a feature subspace that maximizes the information entropy, and realizing the information purification and modal decoupling of the implicit features of the load current time series of the distribution cabinet. Specifically, after the principal component axis system is sorted according to the variance contribution rate, the head principal component can concentrate on characterizing the variation mode with the highest correlation with the fire risk in the implicit features of the load current time series of the distribution cabinet, while the tail component mostly corresponds to environmental noise or normal operating fluctuations. Through this feature reconstruction method, the set of the main component feature encoding vectors of the load current time series of the distribution cabinet after dimensionality reduction not only retains the core dynamic characteristics of the load current, but also strips away redundant information.

[0039] More specifically, the feature fine-grained interactive response analysis subunit 1411 is further used to: perform linear transformation on each distribution cabinet load current time series principal component feature coding vector in the set of the distribution cabinet load current time series principal component feature coding vectors so that it has the same feature scale as the environmental factor filtered cabinet temperature time series implicit feature to obtain a set of distribution cabinet load current time series principal component linear transformation feature coding vectors, which is expressed as follows:

[0040] in, represents the linear transformation function, Represents the set of characteristic encoding vectors of the principal component linear transformation of the load current time series of the distribution cabinet, Respectively represent each distribution cabinet load current time series principal component linear transformation feature coding vector in the set of the distribution cabinet load current time series principal component linear transformation feature coding vector.

[0041] That is, the principal component feature encoding vector of the load current time series of the distribution cabinet is transformed by an affine transformation in the feature space through a parameterized projection matrix, so that its implicit expression and the implicit features of the temperature time series in the cabinet after filtering out environmental factors are aligned in the feature distribution dimension. Specifically, the principal component feature encoding vector of the load current time series of the distribution cabinet is mapped to a reduced or increased dimension through a linear layer, so that the dimension of each principal component feature matches the dimension of the temperature implicit feature. The principal component features of the load current time series of the distribution cabinet after linear transformation and the implicit features of the temperature time series in the cabinet after filtering out environmental factors achieve equivalence of feature expression in a unified multimodal interactive space, which helps to further analyze the complex response relationship between the load current and the temperature in the cabinet.

[0042] More specifically, the feature fine-grained interactive response analysis subunit 1411 is further used to: filter out the implicit features of the cabinet temperature time series of the environmental factors and the linear transformation feature coding vectors of the main components of the load current time series of the distribution cabinet, and input each distribution cabinet load current time series principal component linear transformation feature coding vector into the feature interactive response unit to obtain the set of fine-grained response interactive coding vectors between the load and cabinet temperature time series principal component modes, which is expressed by the formula:

[0043] in, The first one in the set of principal component linear transformation feature coding vectors of the load current time series of the distribution cabinet is represented The characteristic encoding vector of the principal component linear transformation of the load current time series of the distribution cabinet is: represents dot product, Indicates point addition, Indicates point division, represents vector concatenation, and represent the weight matrix and bias term of the feature interaction response unit respectively, express and The load-cabinet temperature time series principal component modal fine-grained response interaction encoding vector.

[0044] That is, the present application performs multi-dimensional feature interaction encoding on the linear transformation feature coding vector of the principal component of the load current time series of the distribution cabinet and the implicit characteristics of the temperature time series filtered out by environmental factors, and introduces a neural network to learn the complex nonlinear relationship between the two, so as to achieve modeling of the fine-grained response mechanism between the load current and the temperature in the cabinet, capture the high-order interaction effects between the principal component characteristics of each load current time series and the temperature time series characteristics in the cabinet, and thus generate a set of fine-grained response interaction coding vectors between the load-cabinet temperature time series principal component modes.

[0045] Figure 5FIG. 1 is a block diagram of an information independence constraint subunit in a distributed power distribution cabinet fire alarm system based on multi-source data fusion according to an embodiment of the present application. Figure 5 As shown, the information independence constraint subunit 1412 includes: an independence modeling secondary subunit 14121, which is used to filter out the implicit characteristics of the cabinet temperature time series of the environmental factors and the linear transformation feature coding vectors of the load current time series principal components of each distribution cabinet in the set of the linear transformation feature coding vectors of the load current time series of the distribution cabinet, and input them into the inter-modal independence modeling unit to obtain a set of load-cabinet temperature time series principal component modal independence coding matrices; an independence soft constraint factor calculation secondary subunit 14122, which is used to calculate the load-cabinet temperature time series principal component modal independence soft constraint factor based on the characteristic distribution of each load-cabinet temperature time series principal component modal independence coding matrix in the set of the load-cabinet temperature time series principal component modal independence coding matrix to obtain the set of load-cabinet temperature time series principal component modal independence soft constraint factors.

[0046] In a specific example of the present application, the independence modeling secondary subunit 14121 is used to: perform nonlinear activation processing on the environmental factor filtering cabinet temperature time series implicit features and the distribution cabinet load current time series principal component linear transformation feature coding vector to obtain the environmental factor filtering cabinet temperature time series feature activation coding vector and the distribution cabinet load current time series principal component feature activation coding vector; calculate the product between the transposed vector of the distribution cabinet load current time series principal component feature activation coding vector and the environmental factor filtering cabinet temperature time series feature activation coding vector, and divide the product result by the square root of the characteristic scale value of the environmental factor filtering cabinet temperature time series feature activation coding vector to obtain the load-cabinet temperature time series principal component mode independence coding matrix, which is expressed as follows:

[0047] in, and is a nonlinear activation function, It means that the environmental factors filter out the implicit characteristics of the temperature time series inside the cabinet. for The characteristic scale value of for and The inter-modal independence encoding matrix of the principal component of the load-cabinet temperature time series.

[0048] That is, the present application introduces a nonlinear activation function to perform nonlinear activation processing on the linear transformation feature coding vector of the load current time series of the distribution cabinet after the linear transformation of the principal component, breaking through the limitations of linear transformation, and mapping the statistical characteristics of the load current and the temperature time series pattern to a high-order nonlinear representation space, so that the two form an interpretable correlation structure in the nonlinear feature space. Subsequently, the product between the transposed vector of the main component feature activation coding vector of the load current time series of the distribution cabinet and the environmental factor filtered out cabinet temperature time series feature activation coding vector is calculated, and the feature scale is normalized to construct a measurable load-cabinet temperature time series principal component modal independence coding matrix, so that the model can dynamically perceive the correlation response strength between the load current anomaly and the cabinet temperature anomaly.

[0049] In particular, in a preferred example of the present application, the independence soft constraint factor calculation secondary subunit 14122 is used to: perform topological structure stability optimization on the load-cabinet temperature time series principal component mode independence coding matrix to obtain an optimized load-cabinet temperature time series principal component mode independence coding matrix, which is expressed as:

[0050] in, Respectively Each eigenvalue of for The eigenvectors of represents the matrix multiplication operation, represents the first-order normative load-cabinet temperature time series principal component modal independence encoding vector, It represents the inter-modal independence coding matrix of the principal component of the optimized load-cabinet temperature time series.

[0051] Here, considering that the asymmetry of the load-cabinet temperature time series principal component modal independence encoding matrix (such as the lag of the current on temperature) may lead to matrix structure instability (such as eigenvalue divergence), affecting the model robustness. This application further optimizes the phase accumulation of the canonical potential to achieve topological robustness optimization of the load-cabinet temperature time series principal component modal independence encoding matrix, maintain the local structural stability of the matrix, and ensure the balanced expression of long-range correlations (such as cumulative temperature rise caused by continuous overload) and short-range fluctuations (such as instantaneous current spikes). In the specific operation, the eigenvector of the independence coding matrix between the principal component modes of the load-cabinet temperature time series is used as the topological normative potential representation, and the correlation phase of the load-temperature mode is dynamically calibrated. While retaining the asymmetric characteristics, the high-order disturbance noise is eliminated through the first-order normative potential term, so that the mutation mode of the load current and the nonlinear response of the temperature in the cabinet can maintain stable interaction in the normalized correlation phase space, so that the optimized independence coding matrix between the principal component modes of the load-cabinet temperature time series can accurately characterize the essential correlation mode excited by the load-temperature response, so that the model can more accurately separate the real fault signal from the random interference signal.

[0052] In a specific example of the present application, the independence soft constraint factor calculation secondary subunit 14122 is further used to: calculate the square of the F norm of the optimized load-cabinet temperature time series principal component mode independence encoding matrix as the load-cabinet temperature time series principal component mode independence soft constraint factor, which is expressed by the formula:

[0053] in, Represents the matrix The square of the norm, express The corresponding soft constraint factor for the independence of the principal component modes of the load-cabinet temperature time series.

[0054] That is, by calculating and optimizing the square of the F-norm of the independence coding matrix between the principal component modes of the load-cabinet temperature time series, the degree of independence between the principal component characteristics of the load time series and the characteristic modes of the temperature time series in the cabinet is quantified, which provides effective guidance for the subsequent dynamic aggregation of fine-grained response interaction coding vectors between the principal component modes of the load-cabinet temperature time series, thereby helping to enhance the model's ability to capture the complex correlation between load current anomalies and cabinet temperature anomalies, and improve the robustness of the model in a noisy environment.

[0055] More specifically, the adaptive aggregation subunit 1413 is expressed as follows:

[0056] in, is the normalized exponential function, Represents the load-cabinet temperature excitation response timing encoding vector.

[0057] Here, by introducing a dynamic adaptive aggregation mechanism based on independence soft constraints, we can break through the limitations of simple splicing or average fusion of traditional multimodal features. Under the premise of retaining the fine-grained interactive response between the main component modes, we can adjust the weight distribution of each fine-grained response in real time according to the temporal dynamic correlation strength of the load and temperature characteristics. By selectively strengthening the strong correlation characteristics between abnormal load current and abnormal temperature fluctuations, we can suppress weak correlation or redundant information such as ambient temperature residual interference and normal load fluctuations, and finally generate a load-cabinet temperature excitation response temporal encoding vector with global timing awareness.

[0058] Specifically, the warning analysis result generating unit 142 is used to input the load-cabinet temperature excitation response time sequence coding vector into the fire warning module based on the classifier to obtain the warning analysis result, and the warning analysis result is used to indicate whether a fire warning prompt signal is generated. Specifically, the classifier is constructed using a deep neural network. After a large amount of training and optimization, the classifier performs feature pattern learning on the input load-cabinet temperature excitation response time sequence coding vector to identify the time sequence interactive response mode of the load current and the temperature in the distribution cabinet, such as the temperature rise mode caused by overload, the temperature fluctuation mode under periodic load changes, and the temperature abnormality mode accompanied by abnormal current peak, and classifies and judges the operating state of the distribution cabinet based on this. When the classifier identifies a potential high-risk fire mode, such as a sharp rise in temperature accompanied by an abnormal current peak, it is determined to be a fire warning state, and a fire warning prompt signal is automatically triggered, so as to promptly notify relevant personnel to take countermeasures, thereby effectively preventing the occurrence of fire accidents and ensuring the safe operation of the distribution system.

[0059] In the specific implementation process, when the early warning analysis results generate a fire warning prompt signal, it indicates that there is a potential fire risk in the distribution cabinet. At this time, a series of emergency response measures need to be initiated immediately. For example, when it is detected that the load current continues to exceed the rated value or fluctuates sharply, and is accompanied by an abnormal temperature increase trend, this indicates that there may be electrical fault hazards such as overload and short circuit. In this case, the system will first issue a clear fire warning prompt signal to notify relevant personnel to pay attention to potential risks. At the same time, the system will evaluate the possible risk level based on the currently collected data and recommend corresponding emergency response plans. These plans may include cutting off the relevant circuits to prevent further electrical failures and heat accumulation; evacuating personnel to ensure the safety of all personnel on site; and starting automatic fire extinguishing devices, such as gas fire extinguishing systems or water sprinkler systems, to quickly control the fire and reduce losses.

[0060] In order to respond to this emergency more effectively, the system's real-time monitoring function can also be combined to dynamically adjust the early warning strategy. By continuously monitoring the changes in various parameters in the distribution cabinet, the system can respond to any emergencies in a timely manner. For example, if the temperature in a specific area continues to rise, even if initial response measures have been taken, the system can further call on more resources to enhance the cooling effect or isolate the affected area to prevent the fire from spreading to other parts. In addition, the system can also work with external fire departments to provide accurate location information and on-site situation descriptions to help firefighters arrive at the scene more quickly and carry out rescue work.

[0061] On the other hand, if the early warning analysis result is that no fire warning prompt signal is generated, it means that the current operating status of the distribution cabinet is within the normal range and there is no obvious fire risk. In this case, although there is no need to take emergency measures immediately, it is still necessary to remain vigilant and continuously monitor the operating status of the distribution cabinet. The system will continue to collect data from various sensors, including temperature, load current and ambient temperature information inside the distribution cabinet to ensure that any subtle changes can be captured in real time. For example, under normal operating conditions, even if the external ambient temperature changes, as long as the temperature changes inside the distribution cabinet conform to the expected pattern and there is no abnormal fluctuation in the load current, the system can be considered to be in a stable state. In addition, regular inspection and maintenance are also essential links to prevent potential problems from occurring.

[0062] In the specific implementation process, in addition to the above-mentioned cutting off of circuits, evacuating personnel, and activating fire extinguishing devices, it is also necessary to consider how to minimize the impact of business interruptions. For example, the continuity of critical services can be maintained by intelligently dispatching backup power or redistributing loads. At the same time, the organization should have a complete set of emergency plans that clearly define the specific responsibilities of each department and individual in the event of a fire, ensuring that everyone knows clearly what to do in an emergency. In addition, fire safety training and drills should be conducted regularly to familiarize employees with escape routes and methods of using fire extinguishing equipment, and improve the overall ability to respond to emergencies. Through such comprehensive preparation and rapid response, the safety of personnel lives can be protected to the greatest extent, property losses can be reduced, and normal operations can be resumed as soon as possible.

[0063] In summary, the distributed distribution cabinet fire alarm system based on multi-source data fusion based on the embodiment of the present application is explained, which uses the timing analysis technology based on deep learning to perform timing modeling on the temperature, ambient temperature and load current data in the distribution cabinet to capture the timing change trend of the three. Then, based on the timing characteristics of the ambient temperature of the distribution cabinet, the environmental interference of the internal temperature timing characteristics of the cabinet is filtered out to obtain a pure representation of the timing change pattern of the internal temperature of the cabinet, and then the timing interaction response excitation analysis is performed on the timing change characteristics of the internal temperature of the distribution cabinet after filtering the load current and environmental interference, so as to realize the intelligent early warning prompt of the fire risk based on the timing correlation response mode between the load and temperature of the distribution cabinet. In this way, the accuracy of fire alarm can be effectively improved, the situation of false alarms and missed alarms can be reduced, and the safe operation of the distribution cabinet can be ensured.

[0064] The basic principle of the present invention is described above in conjunction with specific embodiments. However, it should be pointed out that the advantages, strengths, effects, etc. mentioned in the present invention are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. must be possessed by each embodiment of the present invention. In addition, the specific details of the above embodiments are only for the purpose of illustration and facilitation of understanding, rather than limitation, and the above details do not limit the present invention to being implemented by adopting the above specific details.

[0065] In the above embodiments, the description of each embodiment has its own emphasis. For the parts that are not described or recorded in detail in a certain embodiment, please refer to the relevant description of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed system and method can be implemented in other ways. For example, the system embodiment described above is only schematic. For example, the unit division is only a logical function division, and there may be other division methods in actual implementation. The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0066] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference to a figure in a claim should not be considered as limiting the claim to which it relates.

[0067] In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. The multiple units stated in the system claims can also be implemented by one unit through software or hardware.

[0068] Finally, it should be noted that the above description has been given for the purpose of illustration and description. In addition, the above embodiments are only used to illustrate the technical solution of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A distributed distribution cabinet fire alarm system based on multi-source data fusion, characterized in that: include: The distribution cabinet data acquisition module is used to obtain the distribution cabinet internal temperature time series data set, the distribution cabinet ambient temperature time series data set and the distribution cabinet load current time series data set collected by the sensor group; A distribution cabinet data timing analysis module, used for performing time series encoding on the distribution cabinet internal temperature timing data set, the distribution cabinet ambient temperature timing data set and the distribution cabinet load current timing data set to obtain the distribution cabinet internal temperature timing implicit features, the distribution cabinet ambient temperature timing implicit features and the distribution cabinet load current timing implicit features; An ambient temperature interference filtering module is used to filter the ambient temperature interference of the temperature time series implicit characteristics in the power distribution cabinet based on the ambient temperature time series implicit characteristics of the power distribution cabinet to obtain the environmental factor filtering cabinet temperature time series implicit characteristics; The fire warning analysis module is used to perform load-temperature response excitation analysis on the environmental factors filtering out the implicit characteristics of the temperature timing in the cabinet and the implicit characteristics of the load current timing of the distribution cabinet to determine whether to generate a fire warning prompt signal.

2. The distributed power distribution cabinet fire alarm system based on multi-source data fusion according to claim 1 is characterized in that: The power distribution cabinet data timing analysis module is used to: The temperature time series data set inside the distribution cabinet, the ambient temperature time series data set of the distribution cabinet, and the load current time series data set of the distribution cabinet are respectively time-series encoded based on the time series convolutional neural network model to obtain the implicit time series features of the temperature inside the distribution cabinet, the implicit time series features of the ambient temperature of the distribution cabinet, and the implicit time series features of the load current of the distribution cabinet.

3. The distributed power distribution cabinet fire alarm system based on multi-source data fusion according to claim 2 is characterized in that: The ambient temperature interference filtering module is used to: The temperature time series difference feature between the temperature time series implicit feature in the power distribution cabinet and the temperature time series implicit feature of the ambient temperature of the power distribution cabinet is calculated to obtain the temperature time series implicit feature in the cabinet after filtering out the environmental factors.

4. The distributed power distribution cabinet fire alarm system based on multi-source data fusion according to claim 3 is characterized in that: The fire warning analysis module comprises: A load temperature response excitation analysis unit, used for performing load-temperature response excitation analysis based on inter-modal decoupling constraints on the environmental factor-filtered cabinet temperature time series implicit features and the distribution cabinet load current time series implicit features to obtain a load-cabinet temperature excitation response time series encoding vector; The warning analysis result generating unit is used to input the load-cabinet temperature excitation response time series coding vector into the classifier-based fire warning module to obtain a warning analysis result, and the warning analysis result is used to indicate whether a fire warning prompt signal is generated.

5. The distributed power distribution cabinet fire alarm system based on multi-source data fusion according to claim 4 is characterized in that: The load temperature response excitation analysis unit comprises: A feature fine-grained interactive response analysis subunit, used to perform feature interactive response analysis based on principal component decomposition on the environmental factor-filtered cabinet temperature time series implicit features and the distribution cabinet load current time series implicit features to obtain a set of fine-grained response interactive coding vectors between load-cabinet temperature time series principal component modes; An information independence constraint subunit is used to filter out the information independence between the implicit characteristics of the cabinet temperature time series and the implicit characteristics of the load current time series of the distribution cabinet based on the environmental factors, and calculate the independence soft constraint factors of the fine-grained response interaction coding vectors between the load-cabinet temperature time series principal component modes in the set of fine-grained response interaction coding vectors between the load-cabinet temperature time series principal component modes to obtain a set of soft constraint factors for independence between the load-cabinet temperature time series principal component modes; An adaptive aggregation subunit is used to adaptively aggregate a set of fine-grained response interaction coding vectors between the load-cabinet temperature time series principal component modes based on a set of independence soft constraint factors between the load-cabinet temperature time series principal component modes to obtain the load-cabinet temperature excitation response time series coding vector.

6. The distributed power distribution cabinet fire alarm system based on multi-source data fusion according to claim 5 is characterized in that: The feature fine-grained interactive response analysis subunit is used to: Performing principal component analysis on the implicit characteristics of the load current time series of the distribution cabinet to obtain a set of principal component characteristic encoding vectors of the load current time series of the distribution cabinet; Performing a linear transformation on each distribution cabinet load current time series principal component feature coding vector in the set of the distribution cabinet load current time series principal component feature coding vectors so that the vectors have the same feature scale as the environmental factor filtered cabinet temperature time series implicit feature to obtain a set of distribution cabinet load current time series principal component linear transformation feature coding vectors; The environmental factors are filtered out, the implicit features of the cabinet temperature time series, and each distribution cabinet load current time series principal component linear transformation feature coding vector in the set of the distribution cabinet load current time series principal component linear transformation feature coding vectors are input into the feature interactive response unit to obtain a set of fine-grained response interactive coding vectors between the load and cabinet temperature time series principal component modes.

7. The distributed power distribution cabinet fire alarm system based on multi-source data fusion according to claim 6 is characterized in that: The information independence constraint subunit includes: The second-level sub-unit of independence modeling is used to input the characteristic coding vectors of the principal component linear transformation of the load current time series of the distribution cabinet in the set of characteristic coding vectors of the principal component linear transformation of the load current time series of the distribution cabinet after filtering out the implicit characteristic of the time series of the temperature in the cabinet by the environmental factors into the inter-modal independence modeling unit to obtain the set of inter-modal independence coding matrices of the principal component of the load-cabinet temperature time series; The second-level subunit for calculating the independence soft constraint factor is used to calculate the load-cabinet temperature time series principal component modal independence soft constraint factor based on the characteristic distribution of each load-cabinet temperature time series principal component modal independence coding matrix in the set of the load-cabinet temperature time series principal component modal independence coding matrix to obtain the set of the load-cabinet temperature time series principal component modal independence soft constraint factor.

8. The distributed power distribution cabinet fire alarm system based on multi-source data fusion according to claim 7 is characterized in that: The independence modeling secondary subunit is used to: Performing nonlinear activation processing on the implicit characteristics of the temperature time series of the cabinet after the environmental factors are filtered out and the linear transformation characteristic coding vector of the main component of the load current time series of the distribution cabinet to obtain the activation coding vector of the temperature time series characteristics of the cabinet after the environmental factors are filtered out and the activation coding vector of the main component characteristics of the load current time series of the distribution cabinet; Calculate the product of the transposed vector of the load current time series principal component feature activation coding vector of the distribution cabinet and the environmental factor filtered cabinet temperature time series feature activation coding vector, and divide the product result by the square root of the characteristic scale value of the environmental factor filtered cabinet temperature time series feature activation coding vector to obtain the load-cabinet temperature time series principal component modal independence coding matrix.

9. The distributed power distribution cabinet fire alarm system based on multi-source data fusion according to claim 8 is characterized in that: The independence soft constraint factor calculation secondary subunit is used for: Performing topological structure stability optimization on the load-cabinet temperature time series principal component modal independence coding matrix to obtain an optimized load-cabinet temperature time series principal component modal independence coding matrix; The square of the F-norm of the optimized load-cabinet temperature time series principal component modal independence encoding matrix is ​​calculated as the load-cabinet temperature time series principal component modal independence soft constraint factor.

Images