State monitoring method and device of power distribution room and electronic equipment

Abstract

The invention relates to the technical field of environment monitoring, and discloses a state monitoring method and device for a power distribution room and electronic equipment. The state monitoring method comprises the following steps: acquiring environmental parameters of a distribution room; the environmental parameters comprise pyrolysis particle concentration; constructing an environmental parameter time sequence matrix according to the environmental parameters; and inputting the environment parameter time sequence matrix into the trained state monitoring model to obtain the future state of the power distribution room. According to the invention, the accuracy and reliability of fire risk judgment are improved.

Description

Technical Field

[0001] This application relates to the technical field of environmental monitoring, for example, to a method, device, and electronic device for monitoring the status of a power distribution room. Background Art

[0002] A power distribution room is an important place for centralized management of power distribution, control, and protection in the power system, mainly responsible for safely and effectively delivering electric energy from the power source end (such as a transformer) to each electrical equipment or user end. Since there are usually a large number of flammable substances in the power distribution room (such as cables, insulating materials, etc.), and electrical equipment may generate high temperatures and electric sparks during operation, there is a relatively high fire risk.

[0003] In order to achieve a comprehensive, accurate, and real-time assessment of the fire risk, related technologies disclose a method for assessing the fire risk of the power grid environment based on time series analysis, including: S1. Obtain historical fire data in the power grid environment, and continuously collect power grid working parameters and environmental monitoring data; S2. Data preprocessing; S2.1 Construct a time series to convert historical fire data, power grid working parameters, and environmental monitoring data into a format that can be analyzed by a machine learning model; S2.2 Perform data normalization and missing value processing on the constructed time series; S3. Based on the information recorded in the historical fire data time series, use the unsupervised learning K-means clustering algorithm to classify the fire risk; S4. Use correlation analysis and principal component analysis to select the time series features involved, and the independent variables of the time series are power grid working parameters and environmental monitoring data; S5. Construct a time series data set, and divide the time series data set into a training set and a test set; S6. Construct a machine learning model based on the convolutional neural network CNN and the long short-term memory network LSTM; S7. Input the test set into the trained machine learning model for model evaluation.

[0004] Although related technologies consider various factors in the power grid environment, such as historical fire data, power grid working parameters, and environmental monitoring data, etc., among which, the power grid working parameters include voltage level, frequency, load, active power, reactive power, current, voltage stability, frequency stability, line loss, transformer capacity, and cable capacity, and realize the fire risk assessment of the power grid environment. However, with the replacement or performance degradation of the equipment in the power distribution room, the power grid working parameters fluctuate greatly. The large fluctuations in the power grid working parameters are likely to increase the false alarm probability. Therefore, the accuracy of the fire risk assessment method in related technologies is relatively low. Summary of the Invention

[0005] To gain a basic understanding of some aspects of the disclosed embodiments, a simple summary is provided below. This summary is not a general review, nor is it intended to identify key / important constituent elements or delineate the scope of protection of these embodiments, but rather serves as a preface to the subsequent detailed description.

[0006] Embodiments of the present disclosure provide a method, device, and electronic device for monitoring the status of a power distribution room, improving the accuracy and reliability of fire risk judgment.

[0007] In some embodiments, a method for monitoring the status of a power distribution room is provided, including: obtaining environmental parameters of the power distribution room; the environmental parameters include the concentration of pyrolysis particles; constructing a time series matrix of environmental parameters based on the environmental parameters; and inputting the time series matrix of environmental parameters into a trained status monitoring model to obtain the future status of the power distribution room.

[0008] Optionally, the environmental parameters further include temperature, humidity, hydrogen chloride gas concentration, and smoke concentration; constructing a time series matrix of environmental parameters based on the environmental parameters includes: respectively constructing target time series corresponding to each environmental parameter according to the environmental parameters; and constructing a time series matrix of environmental parameters based on the target time series corresponding to each environmental parameter.

[0009] Optionally, a trained status monitoring model is obtained in the following manner: obtaining sample data; preprocessing the sample data; the preprocessing includes data cleaning, data normalization, data annotation, unifying the sequence length, aligning the time series, and data segmentation; and training a pre-constructed LSTM neural network with the preprocessed sample data to obtain a trained status monitoring model.

[0010] Optionally, the future status includes normal operation and abnormal operation; the status monitoring method further includes: when the future status is normal operation, re-obtaining the environmental parameters of the power distribution room and constructing a time series of environmental parameters; when the future status is abnormal operation, determining a target execution strategy according to the future status and executing it to obtain status anomaly information.

[0011] Optionally, the abnormal operation includes dust interference, water vapor interference, abnormal temperature rise, and fire occurrence; determining a target execution strategy according to the future status and executing it to obtain status anomaly information includes: when the future status is fire occurrence, determining the target execution strategy as the first execution strategy and executing it: obtaining the occurrence probability corresponding to each future status output by the status monitoring model; determining the fire level according to the occurrence probability corresponding to each future status; and taking the fire level as the status anomaly information; when the future status is dust interference, water vapor interference, or abnormal temperature rise, determining the target execution strategy as the second execution strategy and executing it: determining a target risk assessment model according to the future status; inputting the environmental parameters into the target risk assessment model to obtain the cause of the status anomaly; and taking the cause of the status anomaly as the status anomaly information.

[0012] Optionally, according to the occurrence probabilities corresponding to different future states, determine the fire level, including: calculating the weighted values of the occurrence probabilities corresponding to normal operation, dust interference, water vapor interference, and abnormal temperature rise to obtain an interference parameter; calculating the fire level according to the occurrence probability corresponding to the fire occurrence and the interference parameter.

[0013] Optionally, in the case where the future state is abnormal operation, the state monitoring method further includes: determining and executing a target environment adjustment strategy according to the future state and the state abnormal information.

[0014] In some embodiments, a state monitoring device for a power distribution room is provided, including: an acquisition module configured to obtain the environmental parameters of the power distribution room; the environmental parameters include the pyrolysis particle concentration; a construction module configured to construct a time series matrix of environmental parameters according to the environmental parameters; an identification module configured to input the time series matrix of environmental parameters into a trained state monitoring model to obtain the future state of the power distribution room.

[0015] In some embodiments, a state monitoring device for a power distribution room is provided, including a processor and a memory storing program instructions, and the processor is configured to execute the state monitoring method for the power distribution room as described in the above embodiments when running the program instructions.

[0016] In some embodiments, an electronic device is provided, including: a device body; the state monitoring device for the power distribution room as described in the above embodiments is installed on the device body.

[0017] The state monitoring method, device, and electronic device for the power distribution room provided by the embodiments of the present disclosure can achieve the following technical effects:

[0018] The embodiments of the present disclosure can monitor the environmental parameters of the power distribution room in real time, especially the pyrolysis particle concentration. Pyrolysis particles are tiny particles generated by the decomposition of substances at high temperatures, and the pyrolysis particle concentration can reflect early signs of a fire. By arranging the obtained environmental parameters in chronological order and organizing them into the form of a time series matrix to form a time series matrix of environmental parameters, the changing trend of environmental parameters over time can be captured, especially the changing trend of pyrolysis particle concentration over time, providing a basis for subsequent analysis by the state monitoring model. The state monitoring model can analyze the dynamic characteristics and changing trends of the environmental parameters in the time series matrix of environmental parameters over time, so as to predict the future state of the power distribution room.

[0019] Compared with the related art, the embodiments of the present disclosure do not need to collect historical fire data and power grid operating parameters, reducing the quantity and diversity of calculation parameters, thereby reducing the calculation complexity and improving the evaluation efficiency. In addition, during the process of collecting environmental parameters, by collecting the pyrolysis particle concentration as the input of the state monitoring model to judge the future state of the distribution room, the accuracy and reliability of fire risk judgment are improved.

[0020] The above general description and the following description are only exemplary and explanatory, and are not used to limit this application. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] One or more embodiments are exemplarily illustrated by corresponding drawings. These exemplary illustrations and the drawings do not constitute limitations on the embodiments. Elements with the same reference numerals in the drawings are shown as similar elements. The drawings do not constitute a scale limitation, and in which:

[0022] Figure 1 is a schematic diagram of an electronic device provided by an embodiment of the present disclosure;

[0023] Figure 2 is a schematic diagram of a method for monitoring the state of a distribution room provided by an embodiment of the present disclosure;

[0024] Figure 3 is a schematic diagram of a method for training a state monitoring model provided by an embodiment of the present disclosure;

[0025] Figure 4 is a schematic diagram of a pre-constructed LSTM neural network provided by an embodiment of the present disclosure;

[0026] Figure 5 is a schematic diagram of a method for monitoring the state of a distribution room provided by another embodiment of the present disclosure;

[0027] Figure 6 is a schematic diagram of a device for monitoring the state of a distribution room provided by an embodiment of the present disclosure;

[0028] Figure 7 is a schematic diagram of a device for monitoring the state of a distribution room provided by another embodiment of the present disclosure. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0029] In order to understand the features and technical content of the embodiments of the present disclosure in more detail, the implementation of the embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings. The attached drawings are for reference and illustration only and are not intended to limit the embodiments of the present disclosure. In the following technical description, for the sake of explanation, numerous details are provided to give a full understanding of the disclosed embodiments. However, one or more embodiments may still be implemented without these details. In other cases, well-known structures and devices may be shown in a simplified manner to simplify the drawings.

[0030] In the description of the embodiments of the present disclosure, the terms "first", "second", etc. in the specification, claims and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances so as to implement the embodiments of the present disclosure described herein. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion.

[0031] Unless otherwise specified, the term "plurality" means two or more.

[0032] In the embodiments of the present disclosure, the character " / " indicates that the objects before and after are in an "or" relationship. For example, A / B means: A or B.

[0033] The term "and / or" is a description of the associated relationship of objects and indicates that three relationships can exist. For example, A and / or B means: the three relationships of A, B, and A and B.

[0034] The term "corresponding" may refer to an associated relationship or a binding relationship. A corresponding to B means that there is an associated relationship or a binding relationship between A and B.

[0035] It should be noted that, without conflict, the embodiments in the embodiments of the present disclosure and the features in the embodiments may be combined with each other.

[0036] Combined Figure 1 As shown, the embodiments of the present disclosure provide an electronic device 1, including a device body 10 and a state monitoring device 60(70) of the power distribution room. The state monitoring device 60(70) of the power distribution room is installed on the device body 10.

[0037] In the electronic device 1 provided by the embodiments of the present disclosure, the state monitoring device 60(70) of the power distribution room is installed on the device body 10. The installation relationship described here not only includes being placed inside the device body 10, but also includes the installation connection with other components of the electronic device 1, including but not limited to physical connection, electrical connection or signal transmission connection, etc. Those skilled in the art can understand that the state monitoring device 60(70) of the power distribution room can be adapted to the feasible device body 10, so as to implement other feasible embodiments.

[0038] Optionally, the status monitoring device 70 of the power distribution room includes a processor 700. The processor 700 can obtain the environmental parameters of the power distribution room; can construct an environmental parameter time series according to the environmental parameters; can construct an environmental parameter time series matrix according to the environmental parameter time series; can input the environmental parameter time series matrix into the trained status monitoring model to obtain the future status of the power distribution room.

[0039] Combined with Figure 1 and Figure 7 the electronic device shown, an embodiment of the present disclosure provides a status monitoring method for a power distribution room, as Figure 2 shown, the status monitoring method includes:

[0040] S201, the processor obtains the environmental parameters of the power distribution room.

[0041] The environmental parameters include the pyrolysis particle concentration. In this step, sensors can be arranged at different positions in the power distribution room to monitor the environmental parameters in real time. Based on the different monitored environmental parameters, the types of sensors can be different.

[0042] S202, the processor constructs an environmental parameter time series matrix according to the environmental parameters.

[0043] In this embodiment, the environmental parameter time series matrix constructed according to the environmental parameters is as follows:

[0044]

[0045] wherein, represents the measured value of the environmental parameter r collected by sensor n at time k. Let P represent the pyrolysis particle concentration, then the environmental parameter time series matrix corresponding to the pyrolysis particle concentration is as follows:

[0046]

[0047] wherein, represents the measured value of the pyrolysis particle concentration P collected by sensor n at time k.

[0048] S203, the processor inputs the environmental parameter time series matrix into the trained status monitoring model to obtain the future status of the power distribution room.

[0049] The state monitoring method for a power distribution room provided by the embodiments of the present disclosure can monitor the environmental parameters of the power distribution room in real time, especially the pyrolysis particle concentration. Pyrolysis particles are tiny particles generated by the decomposition of substances at high temperatures, and the pyrolysis particle concentration can reflect early signs of a fire. By arranging the obtained environmental parameters in chronological order and organizing them in the form of a time series matrix to form an environmental parameter time series matrix, it is possible to capture the changing trend of environmental parameters over time, especially the changing trend of pyrolysis particle concentration over time, providing a basis for subsequent analysis of the state monitoring model. The state monitoring model can analyze the environmental parameters in the environmental parameter time series matrix, the dynamic characteristics and changing trends evolving over time, so as to predict the future state of the power distribution room.

[0050] In the fire risk assessment methods in the related art, although various factors in the power grid environment are considered, such as historical fire data, power grid operating parameters, and environmental monitoring data, among which, the historical fire data includes the fire occurrence time, scale, duration, types and scales of affected power grid facilities, loss assessment, and casualty situations, the power grid operating parameters include voltage level, frequency, load, active power, reactive power, current, voltage stability, frequency stability, line loss, transformer capacity, and cable capacity, and the environmental monitoring data includes temperature, humidity, wind speed, and atmospheric pressure. However, the complex and diverse parameters also lead to high consumption of computing resources, complex calculation processes, low real-time assessment efficiency, and an increased false alarm probability.

[0051] Compared with the related art, the embodiments of the present disclosure do not need to collect historical fire data and power grid operating parameters, reducing the quantity and diversity of calculation parameters, thereby reducing the calculation complexity and improving the assessment efficiency. In addition, during the process of collecting environmental parameters, by collecting the pyrolysis particle concentration as the input of the state monitoring model to judge the future state of the power distribution room, the accuracy and reliability of fire risk judgment are improved.

[0052] In some embodiments, the environmental parameters further include temperature, humidity, hydrogen chloride gas concentration, and smoke concentration.

[0053] In this embodiment, the monitoring range of environmental parameters is expanded. Based on the original pyrolysis particle concentration parameter, the temperature, humidity, hydrogen chloride gas concentration, and smoke concentration are collected and monitored. The operation of equipment in the distribution room generates heat. Excessive temperature will increase the fire risk; excessive humidity will cause the insulation performance of electrical equipment to decline, increasing the risks of short circuit and fire; chlorine-containing plastics or chemicals are prone to generate hydrogen chloride gas during incomplete combustion at high temperatures, and excessive hydrogen chloride gas concentration will cause damage to equipment and personnel; smoke is a significant feature during a fire, and the level of smoke concentration can intuitively reflect the occurrence and spread of a fire. In this embodiment, by expanding the monitoring range of environmental parameters and monitoring environmental parameters closely related to fire phenomena, the environmental conditions in the distribution room can be more comprehensively understood, thereby more accurately assessing the fire risk and other potential safety hazards.

[0054] Optionally, an environmental parameter time series matrix is constructed according to the environmental parameters, including: constructing the target time series corresponding to each environmental parameter according to the environmental parameters; and constructing the environmental parameter time series matrix according to the target time series corresponding to each environmental parameter.

[0055] In this embodiment, the pyrolysis particle concentration target time series, temperature target time series, humidity target time series, hydrogen chloride gas concentration target time series, and smoke concentration target time series corresponding to the pyrolysis particle concentration, temperature, humidity, hydrogen chloride gas concentration, and smoke concentration are first constructed respectively, and then the environmental parameter time series matrix is constructed according to the pyrolysis particle concentration target time series, temperature target time series, humidity target time series, hydrogen chloride gas concentration target time series, and smoke concentration target time series.

[0056] Exemplarily, let P represent the pyrolysis particle concentration, T represent the temperature, H represent the humidity, Q represent the hydrogen chloride gas concentration, and M represent the smoke concentration. Then the pyrolysis particle concentration target time series, temperature target time series, humidity target time series, hydrogen chloride gas concentration target time series, and smoke concentration target time series corresponding to the pyrolysis particle concentration P, temperature T, humidity H, hydrogen chloride gas concentration Q, and smoke concentration M are respectively: {P tar 1 , P tar 2 , …, P tar k}, {T tar 1 , T tar 2 , …, T tar k}, {H tar 1 , H tar 2 , …, Htar k}, {Q tar 1 , Q tar 2 , …, Q tar k} and {M tar 1 , M tar 2 , …, M tar k}. Based on each target time series, the obtained environmental parameter time series matrix is constructed as follows:

[0057]

[0058] where P tar k represents the target pyrolysis particle concentration at time k, T tar k represents the target temperature at time k, H tar k represents the target humidity at time k, Q tar k represents the target hydrogen chloride gas concentration at time k, M tar k represents the target smoke concentration at time k.

[0059] In this embodiment, by first forming the target time series corresponding to each environmental parameter (pyrolysis particle concentration, temperature, humidity, hydrogen chloride gas concentration, smoke concentration), the change trend of each environmental parameter over time can be captured. Based on the target time series corresponding to each environmental parameter, an environmental parameter time series matrix is constructed, so that the environmental parameter time series matrix can comprehensively reflect the change of each environmental parameter in the distribution room over time. By performing real-time monitoring and time series change analysis on multiple environmental parameters, they can be mutually verified to further improve the accuracy of abnormal warning. For example, when the temperature and smoke concentration increase simultaneously over time, the risk of fire is higher.

[0060] Optionally, according to the environmental parameters, the target time series corresponding to each environmental parameter are constructed respectively, including: calculating the average value or weighted average value of each environmental parameter at the same moment respectively, obtaining the target values of each environmental parameter at different moments; sorting the multiple target values corresponding to the same environmental parameter in chronological order to obtain the target time series corresponding to each environmental parameter.

[0061] In this embodiment, taking to represent the target value of the environmental parameter at time k, the target values of the pyrolysis particle concentration P, temperature T, humidity H, hydrogen chloride gas concentration Q, and smoke concentration M at time k are P tark , T tar k , H tar k , Q tar k and M tar k . And is the average value or weighted average value of environmental parameters at the same moment, that is is the average value or weighted average value.

[0062] In this embodiment, there are multiple sensors simultaneously monitoring the same environmental parameter. By averaging or weighted averaging the environmental parameters collected by multiple sensors at the same moment, more accurate and stable data, that is, the target value, can be obtained, reducing the influence of the error of a single environmental parameter on the overall data and improving the accuracy and stability of the data. The target time series constructed based on the target value is more in line with the actual situation and can more accurately reflect the change trend of the environmental parameter, thereby further improving the accuracy of anomaly warning.

[0063] In some embodiments, the state monitoring model can be obtained by training a pre-constructed LSTM neural network. The LSTM (Long Short-Term Memory) neural network is a special recurrent neural network (RNN). In this embodiment, the state monitoring model is constructed based on the LSTM neural network framework to utilize the adaptive learning ability and robustness of the LSTM neural network to improve the accuracy and reliability of the state monitoring model.

[0064] In some embodiments, the pre-trained state monitoring model is obtained in the following manner. As Figure 3 shown, an embodiment of the present disclosure provides a training method for a state monitoring model, including:

[0065] S301, the processor obtains sample data.

[0066] In this embodiment, the sample data can be the environmental parameters of the power distribution room in different states collected in multiple previous experiments. Among them, different states include normal operation, dust interference, water vapor interference, abnormal temperature rise, and fire occurrence; the environmental parameters include temperature, humidity, smoke concentration, hydrogen chloride gas concentration, and pyrolysis particle concentration.

[0067] S302, the processor preprocesses the sample data.

[0068] In this step, the preprocessing includes data cleaning, data normalization, data annotation, sequence length unification, time series alignment, and data segmentation.

[0069] Among them, data cleaning of the sample data includes: removing noise, error items, duplicate items, missing values, and irrelevant information in the sample data to ensure the accuracy and integrity of the sample data.

[0070] Data normalization of the sample data includes: using the Min-Max normalization method to map each environmental parameter in the sample data from the original interval to the interval of (0, 1) to improve the convergence speed and performance of the state monitoring model. The specific formula is as follows;

[0071]

[0072] Among them, L represents the length of the time series, N represents the total number of sensors, represents the measured value of the environmental parameter r collected by sensor n at time k, maxr n represents the maximum measured value of sensor n in the time series of environmental parameter r, minr n represents the minimum measured value of sensor n in the time series of environmental parameter r, represents the value of the environmental parameter r collected by sensor n at time k after normalization.

[0073] Data annotation of the sample data includes: using One-Hot encoding to convert each state label into a binary vector and corresponding one-to-one with the environmental parameters collected under the corresponding state, so as to add labels to the sample data to indicate the state to which the sample data belongs. Exemplarily, the One-Hot encoding corresponding to different state labels is shown in Table 1:

[0074] Table 1

[0075] Status label Normal operation Dust interference Water vapor interference Abnormal temperature rise Fire occurrence One-Hot encoding <![CDATA[[1,0,0,0,0] T > <![CDATA[[0,1,0,0,0] T > <![CDATA[[0,0,1,0,0] T > <![CDATA[[0,0,0,1,0] T > <![CDATA[[0,0,0,0,1] T >

[0076] Unifying the sequence length of the sample data includes: determining the optimal sequence length; performing truncation or padding operations on the time series corresponding to each environmental parameter according to the optimal sequence length, so that the data input into the pre-constructed LSTM neural network has the same length and meets the input requirements of the LSTM neural network. Exemplarily, the optimal sequence length is 60 s (seconds). In this step, the optimal sequence length can be obtained by technicians according to the actual layout of the power distribution room scene in advance, and the specific value of the optimal sequence length in the embodiments of the present disclosure is not limited.

[0077] Time series alignment of the sample data includes: performing time series detection on the time series corresponding to each environmental parameter and aligning the timestamps to ensure that the time series corresponding to each environmental parameter are arranged in the correct order, and the time series of all environmental parameters are aligned in time for synchronous analysis.

[0078] Data splitting of the sample data includes: dividing the sample data into a training set, a validation set, and a test set. Among them, the training set, the validation set, and the test set all include the environmental parameters of the power distribution room under different states.

[0079] By dividing the sample data into a training set, a validation set, and a test set for model training, validation, and testing, the rationality of the model training, hyperparameter tuning, and evaluation processes is ensured. Exemplarily, 60% of the sample data is taken as the training set to train the LSTM neural network, 20% of the sample data is taken as the validation set to preliminarily verify the effect of the trained LSTM neural network, providing a reference for hyperparameter adjustment, and the remaining 20% of the sample data is taken as the test set to test the performance of the verified LSTM neural network.

[0080] S303, the processor uses the preprocessed sample data to train the pre-constructed LSTM neural network to obtain a trained state monitoring model.

[0081] In this step, the training set can be used to train the pre-constructed LSTM neural network to adjust the network parameters to minimize the loss function. During the training process, the validation set is used for model validation to evaluate the performance of the model and adjust the hyperparameters of the model to prevent overfitting and optimize the model performance. When the performance of the model on the validation set is stable and meets the predetermined requirements, the training is stopped and the test set is used to evaluate the final performance of the model.

[0082] The training method of the state monitoring model provided by the embodiments of the present disclosure preprocesses the sample data before model training, improving the quality and consistency of the sample data, and improving the training efficiency and accuracy of the state monitoring model. In addition, the LSTM neural network has the ability to process long sequence data. Based on the LSTM neural network, a state monitoring model is trained, enabling the state monitoring model to capture complex patterns and trends in time series data and enhancing the robustness of the state monitoring model.

[0083] In some embodiments, as Figure 4 shown, the pre-constructed LSTM neural network includes an input layer, a first hidden layer, a second hidden layer, and an output layer. The input layer is used to receive the input environmental parameter time series matrix; the first hidden layer is used to extract the time series features in the environmental parameter time series matrix, capture the long-term dependencies in the environmental parameters, and output the environmental hidden state information; the second hidden layer takes the environmental hidden state information output by the first hidden layer as input, predicts and outputs the values of each environmental parameter at future time points, that is, the predicted future values; the output layer takes the future values of each environmental parameter output by the second hidden layer as input, makes a judgment on the future state of the power distribution room and outputs.

[0084] In some embodiments, the input layer is a fully connected layer composed of 5 neurons. In this embodiment, the input layer is composed of 5 neurons, and each neuron corresponds to an environmental parameter (such as temperature, humidity, smoke concentration, hydrogen chloride gas concentration, pyrolysis particle concentration, etc.). The input layer is a fully connected layer, that is, each neuron is connected to the subsequent LSTM unit, used to receive and integrate data from different environmental parameters, convert the environmental parameters into a one-dimensional feature vector, and facilitate the processing of the subsequent LSTM unit.

[0085] In some embodiments, the first hidden layer contains 640 neurons, also known as LSTM units. Each LSTM unit follows a specific information transfer formula, including components such as a forget gate, an input gate, a candidate memory cell, and an output gate. The specific information transfer formula is as follows: C t = C t-1 ·σ(W f ·[h t-1 , x t + b f ) + σ(W i ·[h t-1 , x t + b i )·tanh(W c ·[h t-1 , x t + b c ); h t = σ(W o ·[h t-1 , x t + b o )·tanh(C t ).

[0086] Where, x t represents the input data at the current time t; C t represents the cell state at the current time t in the first hidden layer, used to store long-term memory information, and C t-1 represents the cell state at the previous time (t - 1) in the first hidden layer; W f represents the weight of the forget gate in the first hidden layer, and b f represents the bias of the forget gate in the first hidden layer; W i represents the weight of the input gate in the first hidden layer, and b i represents the bias of the input gate in the first hidden layer; W c represents the weight of the candidate memory cell in the first hidden layer, and b c represents the bias of the candidate memory cell in the first hidden layer; W o represents the weight of the output gate in the first hidden layer, and b o represents the bias of the output gate in the first hidden layer; ht represents the output of the LSTM cell state at time t in the first hidden layer, h t-1 represents the output of the LSTM cell state at time t-1 in the first hidden layer; σ and tanh represent activation functions. σ(W f ·[h t-1 ,x t +b f ) is the forget gate part of the first hidden layer, which controls the degree of forgetting of the LSTM unit for the existing features, that is, the cell state (C t-1 ) at the previous moment; σ(W i ·[h t-1 ,x t +b i ) is the input gate part of the first hidden layer, which controls the degree of update of the cell state by the new input feature (x t ), that is, controls the degree of reception of the new input feature by the cell state; tanh(W c ·[h t-1 ,x t +b c ) is the candidate memory cell part of the first hidden layer, which is used to generate new candidate memory content and update the cell state; σ(W o ·[h t-1 ,x t +b o ) is the output gate part of the first hidden layer, which controls the amount of information output based on the current cell state (C t ). The final cell state (C t ) and output (h t ) together constitute the output of the first hidden layer and are passed to the subsequent second hidden layer for processing.

[0087] The LSTM unit is the basic processing unit of the LSTM neural network and is also called a neuron. The LSTM unit is designed to process time series data and capture long-term dependencies in the data. Each LSTM unit contains a series of complex mechanisms and gating structures inside, enabling the LSTM unit to selectively forget old information, receive new information, and update its internal state based on this information. And the cell state is a key component in the LSTM unit. The cell state is a "memory line" that runs through the LSTM unit and is responsible for storing and transmitting long-term memory information. The cell state is passed between LSTM units and is not directly affected by the external input and output gates, thus ensuring the persistence and stability of information.

[0088] In this embodiment, it is specified that the first hidden layer contains 640 neurons in order to more fully extract the feature information in the environmental parameter time series matrix and improve the accuracy and robustness of the state monitoring model.

[0089] In some embodiments, the second hidden layer comprises 320 neurons. Each neuron follows the following information transfer formula: C′ t = C′ t-1 ·σ(W′ f ·[h′ t-1 , h t + b′ f ) + σ(W′ i ·[h′ t-1 , h t + b′ i )·tanh(W′ c ·[h′ t-1 , h t + b′ c );h′ t = σ(W′ o ·[h′ t-1 , h t + b′ o )·tanh(C′ t )。

[0090] Wherein, h t represents the input of the second hidden layer at the current time t, that is, the output of the first hidden layer at time t; C′ t represents the cell state of the second hidden layer at the current time t, and C′ t-1 represents the cell state of the second hidden layer at the previous time (t - 1); W′ f represents the weight of the forget gate in the second hidden layer,, b′ f represents the bias of the forget gate in the second hidden layer; W′ i represents the weight of the input gate in the second hidden layer, b′ i represents the bias of the input gate in the second hidden layer; W c ′ represents the weight of the candidate memory cell in the second hidden layer, b ′ c represents the bias of the candidate memory cell in the second hidden layer; W o ′ represents the weight of the output gate in the second hidden layer, b ′ o represents the bias of the output gate in the second hidden layer; h ′ t represents the output of the LSTM cell state of the second hidden layer at time t, and h ′ t-1 represents the output of the LSTM cell state of the second hidden layer at time t - 1. σ(W f ′ ·[h ′t-1 , h t + b ′ f ) is the forgetting gate part of the second hidden layer, which controls the degree of forgetting of the existing features by the LSTM cell, that is, the cell state (C ′ t-1 ) at the previous moment; σ(W i ′ ·[h t t-1 , h t + b ′ i ) is the input gate part of the second hidden layer, which controls the degree of update of the new input features (h t ) to the cell state, that is, controls the degree of reception of the new input features by the cell state; tanh(W c ′ ·[h ′ t-1 , h t + b ′ c ) is the candidate memory cell part of the second hidden layer, which is used to generate new candidate memory content and update the cell state; σ(W o ′ ·[h ′ t-1 , h t + b ′ o ) is the output gate part of the second hidden layer, which controls the amount of information output based on the current cell state (C ′ t ). The final cell state (C ′ t ) and the output (h ′ t ) together constitute the output of the second hidden layer and are passed to the subsequent output layer for processing.

[0091] In some embodiments, the output layer is a fully connected layer and includes 5 neurons. In this embodiment, the fully connected layer of the output layer includes 5 neurons, and each neuron corresponds to a future state of the power distribution room (such as normal operation, dust interference, water vapor interference, abnormal temperature rise, or fire occurrence).

[0092] The output layer takes the output of the second hidden layer as input and uses a Softmax classifier and an argmax function to judge and output the prediction result of the future state of the power distribution room. The Softmax classifier is used to convert the output of the fully connected layer into a probability distribution, ensuring that the sum of all output values is 1, and each value represents the probability of the corresponding future state. For each neuron (i.e., each future state), the Softmax function calculates the ratio of its output value to the sum of all neuron output values to obtain a probability value, that is, the probability corresponding to the future state. The argmax function is used to select the state with the highest probability from the probability distribution output by the Softmax classifier as the prediction result of the future state of the power distribution room. The argmax function traverses all probability values to find the index corresponding to the maximum probability value, and this index represents the predicted future state of the power distribution room.

[0093] Specifically, taking the output of the second hidden layer as input, calculate Softmax to obtain the probability vector [P norm , P dust , P mist , P temp , P fire of the future state of the power distribution room. T . Among them, P norm represents the probability value that the future state of the power distribution room is normal operation, P dust represents the probability value that the future state of the power distribution room is dust interference, P mist represents the probability value that the future state of the power distribution room is water vapor interference, P temp represents the probability value that the future state of the power distribution room is abnormal temperature rise, P fire represents the probability value that the future state of the power distribution room is fire occurrence. Using the argmax function, find the maximum probability values of P norm , P dust , P mist , P temp and P fire and set them to 1, and set the remaining probability values to 0, and convert the future state probability vector of the power distribution room into a binary vector in One-Hot encoding. Corresponding the binary vector calculated by the argmax function with the One-Hot encoding of each state (refer to Table 1 in the above-mentioned embodiment), and output the future state of the power distribution room.

[0094] Optionally, training the pre-constructed LSTM neural network includes: using the mean squared error - multi-class cross-entropy loss function as the loss function, and using the AdamW optimization algorithm to iteratively train the LSTM neural network to minimize the loss function. Among them, the calculation formula of the mean squared error - multi-class cross-entropy loss function (MSE-Categorical Cross-EntropyLoss) is as follows:

[0095]

[0096]

[0097] L total = α·L MSE + β·L CE ;

[0098] α + β = 1;

[0099] Wherein, L MSE represents the mean squared error loss, L represents the time series length, represents the prediction result of the i-th moment (future state) of the power distribution room output by the LSTM neural network, y i represents the actual result of the power distribution room at the i-th moment, L CE represents the multi-class cross-entropy loss, N represents the number of sample data in the training set, C represents the number of categories of the state label (future state), y ij represents the actual probability that the i-th sample data in the training set belongs to the j-th future state, represents the prediction probability that the i-th sample output by the LSTM neural network belongs to the j-th future state, L total represents the total loss, and α and β are the weights corresponding to the mean squared error loss and the multi-class cross-entropy loss respectively. The mean squared error loss is used to measure the error between the future value and the actual value of the environmental parameters, ensuring that the model can accurately predict the change trend of the environmental parameters. The multi-class cross-entropy loss is used to evaluate the classification performance of the model for future states, ensuring that the model can accurately identify the future states of the power distribution room.

[0100] The AdamW optimization algorithm is an optimization algorithm that can be applied in deep learning. It improves the convergence speed and generalization ability of the model by improving the processing method of weight decay. In this embodiment, the AdamW optimization algorithm is used to iteratively train the LSTM neural network to improve the training efficiency and generalization ability of the model. In this embodiment, the hyperparameters include the time series length, the number of neurons in the first hidden layer, the number of neurons in the second hidden layer, the number of iterations, the weight α corresponding to the mean squared error loss, and the weight β corresponding to the multi-class cross-entropy loss.

[0101] In some embodiments, the future states include normal operation and abnormal operation. The state monitoring method further includes: in the case where the future state is normal operation, re-obtaining the environmental parameters of the power distribution room and constructing an environmental parameter time series; in the case where the future state is abnormal operation, determining a target execution strategy according to the future state and executing it to obtain state anomaly information.

[0102] In this embodiment, when the future state is normal operation, the environmental parameters of the power distribution room are re-acquired and an environmental parameter time series is constructed to continue monitoring the future state of the power distribution room, realizing continuous automation of the environmental monitoring of the power distribution room and improving the operation and maintenance efficiency of the power distribution room. When the future state is abnormal operation, state anomaly information is obtained to learn more information about the state anomaly, so as to take preventive measures according to the specific state to avoid the occurrence of abnormal operation and ensure the safe and stable operation of the power system.

[0103] In some embodiments, the abnormal operation includes dust interference, water vapor interference, abnormal temperature rise and fire occurrence.

[0104] Optionally, a target execution strategy is determined and executed according to the future state to obtain state anomaly information, including: when the future state is a fire occurrence, determining that the target execution strategy is the first execution strategy and executing: obtaining the occurrence probability corresponding to each future state output by the state monitoring model; determining the fire level according to the occurrence probability corresponding to each future state; using the fire level as the state anomaly information.

[0105] In this embodiment, when the future state is a fire occurrence, the fire level can be further determined as the state anomaly information according to the occurrence probability corresponding to each future state. The fire level is a quantitative representation of the possibility and severity of a fire occurrence. By using the determined fire level as the state anomaly information, subsequent fire response, rescue and resource allocation can be carried out according to the change of the fire level, more accurately identifying and handling the safety hazards of the power distribution room, and improving the response efficiency and accuracy.

[0106] Optionally, the fire level is determined according to the occurrence probability corresponding to different future states, including: calculating the weighted value of the occurrence probabilities corresponding to normal operation, dust interference, water vapor interference and abnormal temperature rise to obtain an interference parameter; calculating the fire level according to the occurrence probability corresponding to the fire occurrence and the interference parameter.

[0107] In this embodiment, by calculating the weighted value of the occurrence probabilities corresponding to normal operation, dust interference, water vapor interference and abnormal temperature rise, an interference parameter is obtained to reflect the potential risk of fire occurrence in the current environmental state. Combining the fire occurrence probability and the interference parameter, the fire level is jointly calculated, realizing comprehensive consideration of the occurrence probabilities of various states such as normal operation, dust interference, water vapor interference and abnormal temperature rise, and the potential impact of each state on the fire occurrence, so as to more accurately evaluate the fire risk in the current environment and more accurately reflect the severity and urgency of the fire.

[0108] Optionally, the weighted value of the occurrence probabilities corresponding to normal operation, dust interference, water vapor interference and abnormal temperature rise is calculated according to the following formula to obtain the interference parameter G:

[0109] G = 1 - P norm ·W norm -P temp ·W temp +P dust ·W dust +P mist ·W mist ;

[0110] W norm +W dust +W mist +W temp = 1;

[0111] wherein, P norm , P dust , P mist , P temp are respectively the state probabilities corresponding to normal operation, dust interference, water vapor interference, and abnormal temperature rise output by the state monitoring model; W norm , W dust , W mist , W temp are respectively the weight coefficients corresponding to normal operation, dust interference, water vapor interference, and abnormal temperature rise.

[0112] In this embodiment, different weight coefficients are corresponding to normal operation, dust interference, water vapor interference, and abnormal temperature rise. The specific values of the weight coefficients reflect the influence degree of different states on the possibility of fire occurrence. According to the calculation formula of the interference parameter provided by the embodiments of the present disclosure, it is possible to comprehensively consider the correlation between each state of normal operation, dust interference, water vapor interference, and abnormal temperature rise and the occurrence of fire to calculate the interference parameter, comprehensively consider the potential influence of multiple states on the fire occurrence risk, and improve the comprehensiveness and accuracy of fire level judgment.

[0113] Optionally, the weight coefficients corresponding to different future states are determined by the analytic hierarchy process.

[0114] The Analytic Hierarchy Process (AHP) is a decision analysis method that combines qualitative and quantitative analysis. The AHP can treat a complex multi-objective decision problem as a system, decompose these objectives into multiple objectives or criteria, and then decompose them into several levels of multiple indicators (or criteria, constraints). The single ranking (weight) and total ranking of the hierarchy are calculated through the fuzzy quantification method of qualitative indicators, which is used as a systematic method for optimizing decisions on objectives (multiple indicators) and multiple plans. The AHP decomposes the decision problem into different hierarchical structures in the order of the overall goal, sub-goals of each level, evaluation criteria, and even specific alternative plans. Then, the priority weight of each element at each level to an element at the previous level is obtained by solving the eigenvector of the judgment matrix. Finally, the weighted sum method is used to recursively merge the final weight of each alternative plan to the overall goal.

[0115] In this embodiment, the weight coefficient is determined by the analytic hierarchy process so that the weight coefficient can more accurately reflect the impact of different future states on the fire risk, thereby improving the accuracy and reliability of fire level judgment.

[0116] Optionally, the fire level R is calculated according to the occurrence probability and interference parameters corresponding to the fire occurrence according to the following formula:

[0117]

[0118] Among them, R is the fire risk level, P fire is the state probability corresponding to the fire occurrence output by the state monitoring model, The symbol for rounding up.

[0119] In this embodiment, after calculation, the value of R is an integer greater than 0. The specific value of R is positively correlated with the severity of the fire. In this embodiment, the fire risk is further evaluated in combination with the state probability of fire occurrence and the interference parameter, so as to more accurately identify potential safety hazards and take preventive measures in time.

[0120] Optionally, a target execution strategy is determined based on the future state and executed to obtain state abnormality information, including: when the future state is dust interference, water vapor interference or abnormal temperature rise, the target execution strategy is determined to be a second execution strategy and executed: a target risk assessment model is determined based on the future state; environmental parameters are input into the target risk assessment model to obtain the cause of the state abnormality; and the cause of the state abnormality is used as the state abnormality information.

[0121] In this embodiment, a risk assessment model that matches the specific future state, that is, the target risk assessment model, can be determined according to the specific future state. After inputting the current environmental parameters into the target risk assessment model, the cause of the future state, that is, the cause of the abnormal state, can be obtained. By obtaining the cause of the abnormal state, appropriate measures can be taken in a timely manner to avoid or reduce the probability of the occurrence of this future state.

[0122] Optionally, the number of risk assessment models is multiple. Determining the target risk assessment model according to the future state includes: obtaining the state labels of each risk assessment model; using the risk assessment model corresponding to the state label that is the same as the future state as the target risk assessment model.

[0123] In this embodiment, multiple trained risk assessment models are pre-stored for identifying the corresponding cause of the abnormal state according to the environmental parameters. These risk assessment models are specifically trained and optimized for different future states (such as dust interference, water vapor interference, abnormal temperature rise, etc.). Each risk assessment model has a corresponding state label for identifying the future state applicable to this risk assessment model. By matching the predicted future state with the state labels of each risk assessment model, the risk assessment model corresponding to the state label that is the same as the future state is selected as the target risk assessment model, realizing accurate matching and application of the most suitable risk assessment model from multiple risk assessment models, improving the accuracy and reliability of the assessment of the cause of the abnormal state. Since each risk assessment model is trained and optimized for a specific future state, it can more accurately reflect the characteristics in this future state, improving the pertinence and effectiveness of the assessment of the cause of the abnormal state.

[0124] In a specific application, the number of trained risk assessment models is 3, and the corresponding state labels are dust interference, water vapor interference, and abnormal temperature rise, that is, the dust interference - risk assessment model, the water vapor interference - risk assessment model, and the abnormal temperature rise - risk assessment model. Based on the state labels corresponding to each risk assessment model, the risk assessment model corresponding to the same future state can be determined as the target risk assessment model. If the future state of the power distribution room is dust interference, the environmental parameters are input into the dust interference - risk assessment model, and the main dust particle size and concentration of the current environment are calculated and output; if the future state of the power distribution room is water vapor interference, the environmental parameters are input into the water vapor interference - risk assessment model, and the relative humidity and absolute humidity of the air are calculated and output; if the future state of the power distribution room is abnormal temperature rise, the environmental parameters are input into the abnormal temperature rise - risk assessment model, and the components, positions, and quantities of the abnormal temperature rise are output.

[0125] In a specific application, the dust interference - risk assessment model is obtained by training based on a neural network model. The dust interference - risk assessment model can take the environmental parameters measured in real - time by a dust concentration detector as input, and call the historical average dust concentration and dust particle size information in the power distribution room. Through the trained neural network model, it calculates and outputs the main dust particle size and concentration causing dust interference under the current environmental parameters.

[0126] In a specific application, the water vapor interference - risk assessment model may include a humidity calculation model. The humidity calculation model is a mathematical model based on physical principles and thermodynamic formulas for calculating the content and saturation of water vapor in the air. The humidity calculation model may include an absolute humidity calculation model and a relative humidity calculation model. Absolute humidity refers to the mass of water vapor contained in a certain volume of air. Relative humidity (RH) refers to the ratio of the water vapor pressure in the air to the saturated water vapor pressure at the same temperature. The water vapor interference - risk assessment model can take the local real - time air humidity information and the environmental parameters measured in real - time by a humidity detector in the power distribution room as input, and calculate and output the relative humidity and absolute humidity.

[0127] In a specific application, the abnormal temperature rise - risk assessment model is obtained by training based on a neural network model. The abnormal temperature rise - risk assessment model can screen out one or more temperature detectors corresponding to the temperature parameters with the highest temperature peak in the environmental parameters, and call their temperature waveform data and location information. Then, using the trained neural network model, it discriminates and outputs the components, locations, and quantities of abnormal temperature rise.

[0128] Combined with Figure 5 As shown, the embodiments of the present disclosure provide another method for monitoring the state of a power distribution room, including:

[0129] S501, the processor acquires the environmental parameters of the power distribution room.

[0130] S502, the processor constructs an environmental parameter time - series matrix according to the environmental parameters.

[0131] S503, the processor inputs the environmental parameter time - series matrix into the trained state monitoring model to obtain the future state of the power distribution room.

[0132] The future state of the power distribution room includes normal operation and abnormal operation.

[0133] S504, the processor determines whether the future state of the power distribution room is normal operation. If the future state is normal operation, step S501 is executed again; otherwise, step S505 is executed.

[0134] S505, the processor determines the target execution strategy according to the future state and executes it to obtain the state abnormal information.

[0135] S506. The processor determines and executes a target environment adjustment strategy based on the future state and the state exception information.

[0136] In the state monitoring method for a power distribution room provided by an embodiment of the present disclosure, an environment adjustment strategy library is preset, which stores environment adjustment strategies corresponding to different future states and state exception information. The environment adjustment strategy is used to control electronic devices in the power distribution room, such as fresh air fans, dehumidifiers, air conditioners, etc., to adjust the environment of the power distribution room, so as to reduce the probability of the predicted future state occurring or to make the future state return to normal operation. The embodiment of the present disclosure can, after obtaining the state exception information, select an environment adjustment strategy that matches the future state and the state exception information from the preset environment adjustment strategy library according to the specific future state and state exception information to determine the target environment adjustment strategy and execute it, so as to reduce the probability of the predicted future state occurring or to make the future state return to normal operation, and achieve the reduction or elimination of the abnormal state. The embodiment of the present disclosure realizes the automatic identification and processing of the abnormal operation state, reduces the need for manual intervention, and ensures the pertinence and effectiveness of the adjustment measures.

[0137] Exemplarily, when the future state is dust interference, the target environment adjustment strategy can be "turn on the emergency ventilation system until the dust concentration in the current environment is equal to the preset concentration threshold"; when the future state is water vapor interference, the target environment adjustment strategy can be "turn on the dehumidifier until the relative humidity and absolute humidity of the current air are equal to the corresponding humidity thresholds"; when the future state is abnormal temperature rise, the target environment adjustment strategy can be "according to the position of the abnormal temperature rise component, turn on the air conditioner and adjust the air outlet direction of the air conditioner so that the air outlet direction faces the position of the abnormal temperature rise component, or, turn on the cooling fan corresponding to the position of the abnormal temperature rise component for cooling, or, cut off the power supply of the component with abnormal temperature rise"; when the future state is a fire, the target environment adjustment strategy can be "cut off the power supply of the power distribution room, turn on the ventilation system and send a warning message to the terminal device corresponding to the target personnel". Based on the severity of the fire level, the target personnel can be the security inspection personnel of the power distribution room, all the staff of the power distribution room, all the staff of the power distribution room and the contact person of the fire brigade closest to the power distribution room, etc.

[0138] Combined with Figure 6 As shown in, an embodiment of the present disclosure provides a state monitoring device 60 for a power distribution room, including a collection module 610, a construction module 620, and an identification module 630. The collection module 610 is configured to obtain the environmental parameters of the power distribution room; the environmental parameters include the smoke concentration; the construction module 620 is configured to construct an environmental parameter time series matrix according to the environmental parameters; the identification module 630 is configured to input the environmental parameter time series matrix into the trained state monitoring model to obtain the future state of the power distribution room.

[0139] The state monitoring device 60 of the power distribution room provided by the embodiments of the present disclosure can execute the state monitoring method of the power distribution room described in the above embodiments. Therefore, the technical effects possessed by the state monitoring method of the power distribution room described in the above embodiments are all possessed by the embodiments of the present disclosure, and will not be elaborated here.

[0140] Optionally, the environmental parameters further include temperature, humidity, hydrogen chloride gas concentration, and pyrolysis particle concentration. The construction module 620 is further configured to construct a target time series corresponding to each environmental parameter according to the environmental parameters; and construct an environmental parameter time series matrix according to the target time series corresponding to each environmental parameter.

[0141] Optionally, the state monitoring device 60 of the power distribution room further includes a training module 640. The training module 640 is configured to obtain sample data; perform preprocessing on the sample data; the preprocessing includes data cleaning, data normalization, data annotation, sequence length unification, time series alignment, and data segmentation; and train a pre-constructed LSTM neural network with the preprocessed sample data to obtain a trained state monitoring model.

[0142] Optionally, the future state includes normal operation and abnormal operation. The state monitoring device 60 of the power distribution room further includes an execution module 650. The execution module 650 is configured to, when the future state is normal operation, re-obtain the environmental parameters of the power distribution room and construct an environmental parameter time series; and when the future state is abnormal operation, determine a target execution strategy according to the future state and execute it to obtain state anomaly information.

[0143] Optionally, the abnormal operation includes dust interference, water vapor interference, abnormal temperature rise, and fire occurrence. The execution module 650 is further configured to, when the future state is fire occurrence, determine the target execution strategy as the first execution strategy and execute it: obtain the occurrence probabilities corresponding to different future states output by the state monitoring model; determine the fire level according to the occurrence probabilities corresponding to different future states; and use the fire level as the state anomaly information.

[0144] The execution module 650 is further configured to, when the future state is dust interference, water vapor interference, or abnormal temperature rise, determine the target execution strategy as the second execution strategy and execute it: determine a target risk assessment model according to the future state; input the environmental parameters into the target risk assessment model to obtain the cause of the state anomaly; and use the cause of the state anomaly as the state anomaly information.

[0145] Optionally, the execution module 650 is further configured to calculate a weighted value of the occurrence probabilities corresponding to normal operation, dust interference, water vapor interference, and abnormal temperature rise to obtain an interference parameter; and calculate the fire level according to the occurrence probability corresponding to fire occurrence and the interference parameter.

[0146] Optionally, the status exception information includes the cause of the status exception and the level of the status exception. In the case where the future state is abnormal operation, the execution module 650 is further configured to determine and execute a target environment adjustment policy according to the future state and the status exception information.

[0147] As shown in combination with Figure 7 The embodiment of the present disclosure provides a status monitoring device 70 for a power distribution room, including a processor 700 and a memory 701. Optionally, the device 70 may further include a communication interface 702 and a bus 703. Among them, the processor 700, the communication interface 702, and the memory 701 can complete mutual communication through the bus 703. The communication interface 702 can be used for information transmission. The processor 700 can call the logical instructions in the memory 701 to execute the status monitoring method for the power distribution room in the above embodiment.

[0148] In addition, when the logical instructions in the above-mentioned memory 701 are implemented in the form of a software functional unit and sold or used as an independent product, they can be stored in a computer-readable storage medium.

[0149] The memory 701, as a computer-readable storage medium, can be used to store software programs and computer-executable programs, such as the program instructions / modules corresponding to the methods in the embodiments of the present disclosure. The processor 700 executes functional applications and data processing by running the program instructions / modules stored in the memory 701, that is, implements the status monitoring method for the power distribution room in the above embodiment.

[0150] The memory 701 may include a storage program area and a storage data area. Among them, the storage program area can store an operating system and application programs required for at least one function; the storage data area can store data created according to the use of the terminal device, etc. In addition, the memory 701 may include a high-speed random access memory and may also include a non-volatile memory.

[0151] The embodiment of the present disclosure provides a computer-readable storage medium storing computer-executable instructions, and the computer-executable instructions are set to execute the status monitoring method for the power distribution room described above.

[0152] The technical solution of the embodiments of the present disclosure can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes one or more instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the embodiments of the present disclosure. The foregoing storage medium may be a non-transitory storage medium, such as: a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disc, etc., which are various media that can store program codes.

[0153] The above description and drawings fully illustrate the embodiments of the present disclosure so that those skilled in the art can practice them. Other embodiments may include structural, logical, electrical, process, and other changes. Embodiments merely represent possible variations. Unless explicitly required, individual components and functions are optional, and the order of operations can vary. Parts and features of some embodiments may be included in or substituted for parts and features of other embodiments. Moreover, the terms used in this application are only for describing embodiments and are not used to limit the claims. As used in the description of the embodiments and the claims, unless the context clearly indicates otherwise, the singular forms "a", "an", and "the" are intended to also include the plural forms. Similarly, as used in this application, the term "and / or" refers to any and all possible combinations including one or more of the associated listed items. Additionally, when used in this application, the term "comprise" and its variants "comprises" and / or "comprising" etc. mean the presence of the stated features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or groups of these. Without further limitation, an element defined by the statement "comprising one..." does not exclude the presence of additional identical elements in the process, method, or device including the element. In this document, each embodiment may focus on the differences from other embodiments, and the same or similar parts among the embodiments may be referred to each other. For the methods, products, etc. disclosed in the embodiments, if they correspond to the method part disclosed in the embodiments, the relevant parts may refer to the description of the method part.

[0154] Those skilled in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner may depend on the specific application and design constraints of the technical solution. The skilled person can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of the embodiments of the present disclosure. The skilled person can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0155] In the embodiments disclosed herein, the disclosed methods, products (including but not limited to devices, equipment, etc.) can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units can be merely a logical function division, and there can be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces, and the indirect couplings or communication connections of the devices or units can be in electrical, mechanical, or other forms. 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 can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to implement this embodiment. In addition, the functional units in the embodiments of the present disclosure can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0156] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to embodiments of the present disclosure. In this regard, each block in the flowchart or block diagram may represent a module, a segment of a program, or a portion of code, which contains one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions noted in the blocks may occur in a different order than noted in the accompanying drawings. For example, two consecutive blocks may in fact be executed substantially in parallel, or they may sometimes be executed in the reverse order, depending on the functions involved. In the description corresponding to the flowcharts and block diagrams in the accompanying drawings, the operations or steps corresponding to different blocks may also occur in a different order than disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps may in fact be executed substantially in parallel, or they may sometimes be executed in the reverse order, depending on the functions involved. Each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented by a dedicated hardware-based system that performs the specified functions or actions, or may be implemented by a combination of dedicated hardware and computer instructions.

Claims

1. A method for monitoring the status of a power distribution room, characterized in that: include: Obtain environmental parameters of the power distribution room; environmental parameters include pyrolysis particle concentration; According to the environmental parameters, construct the environmental parameter time series matrix; The environmental parameter time series matrix is ​​input into the trained condition monitoring model to obtain the future state of the distribution room.

2. The condition monitoring method according to claim 1, characterized in that: Environmental parameters also include temperature, humidity, hydrogen chloride gas concentration, and smoke concentration; According to the environmental parameters, construct the environmental parameter timing matrix, including: According to the environmental parameters, the target time series corresponding to each environmental parameter are constructed respectively; According to the target time series corresponding to each environmental parameter, an environmental parameter time series matrix is ​​constructed.

3. The condition monitoring method according to claim 1 or 2, characterized in that: Obtain the trained condition monitoring model as follows: Obtain sample data; Preprocess the sample data; preprocessing includes data cleaning, data normalization, data labeling, sequence length unification, time series alignment and data segmentation; The pre-processed sample data is used to train the pre-built LSTM neural network to obtain a trained state monitoring model.

4. The condition monitoring method according to claim 1 or 2, characterized in that: The future state includes normal operation and abnormal operation; the state monitoring method also includes: When the future state is normal operation, the environmental parameters of the power distribution room are re-obtained and the environmental parameter time series is constructed; In the case that the future state is abnormal operation, the target execution strategy is determined and executed according to the future state to obtain the state abnormality information.

5. The condition monitoring method according to claim 4, characterized in that: Abnormal operation includes dust interference, water vapor interference, abnormal temperature rise and fire. Determine and execute the target execution strategy based on the future state to obtain abnormal state information, including: In the case where the future state is a fire, the target execution strategy is determined to be the first execution strategy and executed: obtaining the occurrence probability corresponding to each future state output by the state monitoring model; determining the fire level according to the occurrence probability corresponding to each future state; and using the fire level as state abnormality information; When the future state is dust interference, water vapor interference or abnormal temperature rise, the target execution strategy is determined to be the second execution strategy and executed: the target risk assessment model is determined according to the future state; the environmental parameters are input into the target risk assessment model to obtain the cause of the abnormal state; the cause of the abnormal state is used as the abnormal state information.

6. The condition monitoring method according to claim 5, characterized in that: Determine the fire level based on the probability of occurrence of different future states, including: Calculate the weighted values ​​of the occurrence probabilities of normal operation, dust interference, water vapor interference and abnormal temperature rise to obtain interference parameters; The fire grade is calculated based on the occurrence probability and interference parameters corresponding to the fire.

7. The condition monitoring method according to claim 4, characterized in that: In the case where the future state is abnormal operation, the condition monitoring method also includes: Based on the future status and status exception information, determine the target environment adjustment strategy and execute it.

8. A state monitoring device for a power distribution room, characterized in that: include: A collection module is configured to obtain environmental parameters of the power distribution room; the environmental parameters include pyrolysis particle concentration; The building module is configured to build an environment parameter timing matrix according to the environment parameters; The identification module is configured to input the environmental parameter time series matrix into the trained state monitoring model to obtain the future state of the distribution room.

9. A state monitoring device for a power distribution room, comprising a processor and a memory storing program instructions, characterized in that: The processor is configured to execute the method for monitoring the status of a power distribution room as described in any one of claims 1 to 7 when running the program instructions.

10. An electronic device, characterized in that: include: Equipment body; The status monitoring device for the power distribution room as described in claim 8 or 9 is installed on the equipment body.

Images