Intelligent Diagnosis Method and System for Building Electrical Safety Risks Based on Big Data

Abstract

The present invention relates to the technical field of building electrical safety, and discloses an intelligent diagnosis method and system for building electrical safety risks based on big data, which includes: collecting the operation condition data of the building electrical system in real time; cleaning the collected operation condition data to remove noise, and normalizing the denoised operation condition data to obtain preprocessed data; extracting features from the preprocessed data, extracting time-domain features and frequency-domain features, obtaining the energy distribution at different scales through wavelet transform, capturing electrical signal anomalies, and obtaining multi-dimensional electrical feature data; inputting the multi-dimensional electrical feature data into an intelligent diagnosis model, diagnosing the building electrical safety state through the intelligent diagnosis model, and outputting a safety diagnosis result; determining the building electrical safety risk level according to the safety diagnosis result, and sending corresponding warning information based on the safety risk level; the present invention accurately and real-time identifies building electrical safety hazards and ensures the safety of building electricity consumption.

Description

Technical Field

[0001] The present invention relates to the technical field of building electrical safety, and specifically relates to an intelligent diagnosis method and system for building electrical safety risks based on big data. Background Art

[0002] In modern buildings, the electrical system is complex and huge, covering multiple subsystems such as lighting, power, sockets, and fire protection electricity. Traditional electrical safety inspections mostly rely on manual regular inspections, with measurements by experience and simple instruments. This method has low efficiency and strong subjectivity, and it is difficult to capture potential risks in real time and comprehensively. For example, problems such as wire aging, overload, short circuit, and grounding faults may instantly cause electrical fires or equipment damage, and hidden dangers are difficult to detect in a timely manner during the interval of manual inspections. Summary of the Invention

[0003] The purpose of the present invention is to solve the above problems and design an intelligent diagnosis method and system for building electrical safety risks based on big data.

[0004] The first aspect of the present invention provides an intelligent diagnosis method for building electrical safety risks based on big data. The intelligent diagnosis method for building electrical safety risks based on big data includes the following steps:

[0005] Collect the original parameter data of the building electrical system in real time through various sensors, interface with the building automation system to obtain the operation status information, connect to the power quality monitor to collect the power quality index data, and integrate the original parameter data, operation status information, and power quality index data to obtain the operation condition data of the building electrical system;

[0006] Clean the collected operation condition data to remove noise, and perform normalization processing on the denoised operation condition data to obtain the preprocessed data;

[0007] Extract features from the preprocessed data, extract time-domain features and frequency-domain features, obtain the energy distribution at different scales through wavelet transform, capture electrical signal anomalies, and obtain multi-dimensional electrical feature data;

[0008] Input the multi-dimensional electrical feature data into the intelligent diagnosis model, diagnose the building electrical safety status through the intelligent diagnosis model, and output the safety diagnosis result;

[0009] Determine the building electrical safety risk level according to the safety diagnosis result, and send corresponding warning information based on the safety risk level.

[0010] Optionally, in the first implementation manner of the first aspect of the present invention, the cleaning of the collected operation condition data to remove noise and the normalization processing of the denoised operation condition data to obtain the preprocessed data include:

[0011] Obtain the collected operating condition data, traverse the operating condition data to check whether there are missing values in each data record, and use linear interpolation to fill the detected missing values. For each position where a missing value is located, estimate the missing value through linear calculation based on the valid data points adjacent before and after it.

[0012] Use a filtering algorithm based on statistical characteristics to detect outliers in the filled data, calculate the mean and standard deviation of each data type, calculate the absolute value of the difference between each data point and the mean. If the absolute value of the difference of a certain data point is greater than the threshold, the corresponding data point is determined as an outlier, where the threshold is three times the standard deviation.

[0013] Use smoothing processing to correct the detected outliers, and use a median filter to remove noise from the data after missing value and outlier processing to obtain denoised data.

[0014] Use the Min-Max normalization method to normalize the denoised data to obtain preprocessed data.

[0015] Optionally, in the second implementation manner of the first aspect of the present invention, perform feature extraction on the preprocessed data, extract time-domain features and frequency-domain features, obtain the energy distribution at different scales through wavelet transform, capture electrical signal anomalies, and obtain multi-dimensional electrical feature data, including:

[0016] Obtain the preprocessed data, arrange the preprocessed data in chronological order to obtain a data sequence.

[0017] Calculate the mean, standard deviation, and effective value of the data sequence, traverse the data sequence to find the maximum value, minimum value, and peak value, obtain the peak factor based on the ratio of the peak value to the effective value, calculate the difference between adjacent voltage values, and calculate the standard deviation of the difference to obtain the voltage fluctuation standard deviation, and integrate to obtain time-domain features.

[0018] Apply the fast Fourier transform to convert the time-domain features of the data sequence into frequency-domain features. For the frequency-domain sequence, calculate the frequency positions of each harmonic according to the fundamental frequency, extract the amplitudes at the corresponding frequency positions, and statistically calculate the ratio of the amplitudes of each harmonic to the fundamental amplitude to obtain the harmonic spectrum distribution.

[0019] Determine the low-frequency range and extract the frequency-domain components within this range, and analyze the changes in the amplitude and phase of the low-frequency components over time, where the low-frequency range is below 100 Hz.

[0020] Obtain the energy distribution at different scales through wavelet transform, capture electrical signal anomalies, and obtain multi-dimensional electrical feature data.

[0021] Optionally, in the third implementation manner of the first aspect of the present invention, obtaining the energy distribution at different scales through wavelet transform, capturing electrical signal anomalies, and obtaining multi-dimensional electrical feature data includes:

[0022] Performing N-layer wavelet decomposition on the data sequence using Haar wavelet to decompose the signal into an approximation component and a detail component;

[0023] Calculating the energy of the approximation component and the detail component of each layer, integrating the extracted time-domain features, frequency-domain features, and the energy distribution at different scales obtained by wavelet transform to form a multi-dimensional feature vector as the final multi-dimensional electrical feature data.

[0024] Optionally, in the fourth implementation manner of the first aspect of the present invention, inputting the multi-dimensional electrical feature data into an intelligent diagnosis model, diagnosing the electrical safety state of the building through the intelligent diagnosis model, and outputting a safety diagnosis result, including:

[0025] Constructing an intelligent diagnosis model based on a DNN network and an LSTM network, and inputting the multi-dimensional electrical feature data into the intelligent diagnosis model;

[0026] Inputting the multi-dimensional electrical feature data into the input layer of the DNN network, performing layer-by-layer calculations in the hidden layer of the DNN network. For each neuron in each layer, taking the output of the previous layer as the input, performing weighted summation, adding a bias, and introducing a non-linear factor through the ReLU activation function to obtain the output of this layer. After calculations through multiple hidden layers, obtaining the output of the output layer of the DNN network;

[0027] Taking the output of the output layer of the DNN network as the input of the LSTM network. At each time step, the LSTM network calculates based on the current input, the hidden state of the previous moment, and the cell state to capture the time-series correlation of electrical parameters. After processing all time steps, obtaining the final output of the LSTM network;

[0028] Calculating the probabilities of the building electrical being in different safety states through the Softmax function for the final output of the LSTM network. According to the calculated probabilities, selecting the safety state with the highest probability as the current safety state of the building electrical, and organizing the determined safety state and the corresponding probability values to output the safety diagnosis result.

[0029] Optionally, in the fifth implementation manner of the first aspect of the present invention, constructing the intelligent diagnosis model based on the DNN network and the LSTM network includes:

[0030] Obtain the multimodal data of historical multi-source sensors, use the dynamic feature selection algorithm to screen the features of the collected multimodal data, evaluate the importance of each feature, obtain the features highly relevant to the building electrical safety diagnosis, and remove redundant features, where the multimodal data is the operating state of the building electrical system under different working conditions in the past;

[0031] Adopt the attention mechanism to dynamically adjust the weights of the data obtained after feature screening, and use the adaptive noise reduction method combining wavelet transform and generative adversarial network for noise reduction processing to obtain the training set;

[0032] Input the training set into the DNN network for training, and introduce the self-attention mechanism into the DNN network to focus on the mutual relationship between different features;

[0033] Take the output of the DNN network as the input of the LSTM network. The LSTM network introduces Peephole connections and Zoneout regularization. The Peephole connection adds additional connections to the forget gate, input gate, and output gate, and can receive the information of the cell state;

[0034] Adopt a differentiable gating mechanism to construct a dynamic feature fusion layer, fuse the features output by the DNN network and the LSTM network, and dynamically allocate weights according to the importance of different features in the current diagnosis task;

[0035] Construct a four-dimensional optimization objective function, use the MOSHO algorithm for hyperparameter optimization to obtain the optimal solution, and build an intelligent diagnosis model.

[0036] Optionally, in the sixth implementation manner of the first aspect of the present invention, the constructing a four-dimensional optimization objective function, using the MOSHO algorithm for hyperparameter optimization to obtain the optimal solution, and building an intelligent diagnosis model includes:

[0037] Determine that the four indicators of the four-dimensional optimization objective function are diagnostic accuracy, inference speed, memory occupancy, and model stability, and assign weights to the four indicators;

[0038] Randomly initialize a certain number of particles in the hyperparameter space of the model to be optimized. Each particle represents a set of hyperparameter combinations, and assign an initial velocity and position to each particle;

[0039] According to the initial hyperparameter combinations and the corresponding objective function values, initialize the training of the Bayesian model, apply the hyperparameter combinations represented by each particle in the current particle swarm to the intelligent diagnosis model for training and evaluation, and calculate the corresponding four-dimensional optimization objective function values;

[0040] For each particle, compare the objective function value at the current position of each particle with the objective function value at the historical optimal position. If the objective function value at the current position is better, update the historical optimal position of the particle.

[0041] Compare the objective function values at the historical optimal positions of all particles, and select the position with the best objective function value as the global optimal position.

[0042] Introduce quantum behavior simulation to improve the update rule of the particle swarm. Use the trained Bayesian model to predict the hyperparameter combination that obtains a better objective function value in the hyperparameter space, and add the predicted hyperparameter combination as a new particle to the particle swarm.

[0043] Add the hyperparameter combination of the newly added particle and the corresponding objective function value to the training data of the Bayesian model, and retrain the Bayesian model.

[0044] Check whether the current iteration number reaches the upper limit. If the termination condition is met, stop the iteration; otherwise, return to continue the iterative optimization.

[0045] After the iteration ends, the hyperparameter combination represented by the global optimal position is the optimal solution of the four-dimensional optimization objective function, and use the optimal hyperparameter combination to construct an intelligent diagnosis model.

[0046] The second aspect of the present invention provides an intelligent diagnosis system for building electrical safety risks based on big data. The intelligent diagnosis system for building electrical safety risks based on big data includes a data acquisition module, a data preprocessing module, a feature extraction module, an intelligent diagnosis module, and an early warning module. Among them, the data acquisition module is used to collect the original parameter data of the building electrical system in real time through a variety of sensors, connect with the building automation system to obtain the operation status information, connect with the power quality monitor to collect the power quality index data, and integrate the original parameter data, operation status information, and power quality index data to obtain the operation condition data of the building electrical system.

[0047] The data preprocessing module is used to clean the collected operation condition data to remove noise, and perform normalization processing on the denoised operation condition data to obtain the preprocessed data.

[0048] The feature extraction module is used to extract features from the preprocessed data, extract time-domain features and frequency-domain features, obtain the energy distribution at different scales through wavelet transform, capture electrical signal anomalies, and obtain multi-dimensional electrical feature data.

[0049] The intelligent diagnosis module is used to input the multi-dimensional electrical feature data into the intelligent diagnosis model, diagnose the building electrical safety status through the intelligent diagnosis model, and output the safety diagnosis result.

[0050] An early warning module, configured to determine the building electrical safety risk level according to the safety diagnosis result, and send corresponding early warning information based on the safety risk level.

[0051] Optionally, in the first implementation manner of the second aspect of the present invention, the data preprocessing module includes a filling sub-module, a calculation sub-module, a correction sub-module, and a normalization processing sub-module. Among them, the filling sub-module is configured to obtain the collected operating condition data, traverse the operating condition data to check whether there are missing values in each data record, and fill the detected missing values using the linear interpolation method. For each position where a missing value is located, the missing value is estimated through linear calculation based on the valid data points adjacent before and after it.

[0052] The calculation sub-module is configured to detect outliers in the filled data using a filtering algorithm based on statistical characteristics, calculate the mean and standard deviation of each data type, calculate the absolute value of the difference between each data point and the mean. If the absolute value of the difference of a certain data point is greater than the threshold, the corresponding data point is determined as an outlier, where the threshold is three times the standard deviation.

[0053] The correction sub-module is configured to correct the detected outliers using smoothing processing, and use a median filter to remove noise from the data after missing value and outlier processing to obtain denoised data.

[0054] The normalization processing sub-module is configured to perform normalization processing on the denoised data using the Min-Max normalization method to obtain preprocessed data.

[0055] Optionally, in the second implementation manner of the second aspect of the present invention, the feature extraction module includes an arrangement sub-module, a traversal sub-module, a conversion sub-module, an analysis sub-module, and a capture sub-module. Among them, the arrangement sub-module is configured to obtain the preprocessed data and arrange the preprocessed data in chronological order to obtain a data sequence.

[0056] The traversal sub-module is configured to calculate the mean, standard deviation, and effective value of the data sequence, traverse the data sequence to find the maximum value, minimum value, and peak value, obtain the peak factor based on the ratio of the peak value to the effective value, calculate the difference between adjacent voltage values, and calculate the standard deviation of the difference to obtain the voltage fluctuation standard deviation, and integrate to obtain time-domain features.

[0057] The conversion sub-module is configured to apply the fast Fourier transform to convert the time-domain features of the data sequence into frequency-domain features. For the frequency-domain sequence, calculate the frequency positions of each harmonic according to the fundamental frequency, extract the amplitudes at the corresponding frequency positions, and count the proportion of the amplitudes of each harmonic to the fundamental amplitude to obtain the harmonic spectrum distribution.

[0058] An analysis sub-module, which is used to determine a low-frequency range, extract frequency-domain components within this range, and analyze the changes in the amplitude and phase of the low-frequency components over time, where the low-frequency range is below 100 Hz.

[0059] A capture sub-module, which is used to obtain the energy distribution at different scales through wavelet transform, capture electrical signal anomalies, and obtain multi-dimensional electrical characteristic data.

[0060] In the technical solution provided by the present invention, the original parameter data of the building electrical system is collected in real time through multiple sensors, and is docked with the building automation system to obtain the operating state information. The power quality monitor is connected to collect the power quality index data. The original parameter data, the operating state information, and the power quality index data are integrated to obtain the operating condition data of the building electrical system; the collected operating condition data is cleaned to remove noise, and the denoised operating condition data is normalized to obtain the preprocessed data; feature extraction is performed on the preprocessed data to extract time-domain features and frequency-domain features, the energy distribution at different scales is obtained through wavelet transform, electrical signal anomalies are captured, and multi-dimensional electrical characteristic data is obtained; the multi-dimensional electrical characteristic data is input into the intelligent diagnosis model, the building electrical safety state is diagnosed through the intelligent diagnosis model, and the safety diagnosis result is output; according to the safety diagnosis result, the building electrical safety risk level is determined, and the corresponding warning information is sent based on the safety risk level; the present invention accurately and real-time identifies building electrical safety hazards and ensures the safety of building electricity use. Description of the Drawings

[0061] By reading the detailed description of the preferred embodiments below, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only for the purpose of showing the preferred embodiments and are not considered to be a limitation of the present invention.

[0062] Figure 1 Schematic diagram of the first embodiment of the intelligent diagnosis method for building electrical safety risks based on big data provided by the embodiment of the present invention;

[0063] Figure 2 Schematic diagram of the second embodiment of the intelligent diagnosis method for building electrical safety risks based on big data provided by the embodiment of the present invention;

[0064] Figure 3 Schematic diagram of the third embodiment of the intelligent diagnosis method for building electrical safety risks based on big data provided by the embodiment of the present invention;

[0065] Figure 4 Schematic diagram of the structure of the intelligent diagnosis system for building electrical safety risks based on big data provided by the embodiment of the present invention. Detailed Embodiments

[0066] The terms "first," "second," "third," "fourth," and so forth (if any) in the description and claims of the present invention and in the accompanying drawings are used to distinguish similar items and are not necessarily used to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate, so that the embodiments described herein can be implemented in an order other than that shown or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, apparatus, product, or device that includes a series of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, product, or device.

[0067] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 A schematic diagram of a first embodiment of a method for intelligent diagnosis of building electrical safety risks based on big data provided by an embodiment of the present invention, the method specifically comprising the following steps:

[0068] Step 101: Parameter raw data of the building electrical system is collected in real time through various sensors, and the system is connected to the building automation system to obtain operating status information. The system is connected to the power quality monitor to collect power quality index data. The parameter raw data, operating status information, and power quality index data are integrated to obtain operating condition data of the building electrical system.

[0069] In this embodiment, the current transformer focuses on capturing the magnitude of the current in the line and continuously monitors the real-time value of the current, such as the normal current during stable operation and the current fluctuations caused by sudden load changes; the voltage transformer obtains the voltage situation and accurately measures the voltage value at each point; the temperature sensor is attached to the surface of the electrical equipment or key line parts to sense the temperature changes of the equipment and the line in real time, because excessive temperature is often the prelude to electrical faults. Problems such as line overload and poor contact will cause the temperature to rise sharply; the leakage sensor closely monitors the leakage current; the building automation system (BAS) controls the operating status of many aspects of the electrical equipment. The opening and closing status of the circuit breaker directly reflects the on / off situation of the circuit. If the circuit breaker opens and closes frequently, it may indicate problems such as short circuit and overload in the downstream line; frequent operation of the contactor may imply unstable operation of the equipment it controls or disorder in the control logic; equipment running continuously for a long time is more likely to have problems such as fatigue wear and overheating; the power quality monitor conducts a comprehensive detection of the power quality, and the harmonic content is one of its key detection indicators. With the use of a large number of non-linear loads (such as electronic equipment and frequency conversion devices) in modern buildings, the harmonic problem has become increasingly prominent. Excessive harmonics will interfere with the normal operation of other electrical equipment and even damage the equipment; the monitoring of the three-phase unbalance degree is also crucial. Three-phase unbalance may cause excessive load on a certain phase, resulting in consequences such as line heating and equipment damage; the power factor reflects the energy efficiency of the power system. A low power factor not only wastes electric energy but may also overload the power supply equipment and affect the power supply stability.

[0070] Step 102: Clean the collected operating condition data to remove noise, and perform normalization processing on the denoised operating condition data to obtain preprocessed data;

[0071] In this embodiment, the collected operating condition data is obtained, and the operating condition data is traversed to check whether there are missing values in each data record. The linear interpolation method is used to fill the detected missing values. For the position of each missing value, based on the adjacent valid data points before and after it, the missing value is estimated through linear calculation; the outlier of the filled data is detected by a filtering algorithm based on statistical characteristics. The mean and standard deviation of each data type are calculated, and the absolute value of the difference between each data point and the mean is calculated. If the absolute value of the difference of a certain data point is greater than the threshold, the corresponding data point is determined as an outlier, where the threshold is three times the standard deviation; the detected outliers are corrected by smoothing processing, and the median filter is used to remove the noise from the data after missing value and outlier processing to obtain denoised data; the Min-Max normalization method is used to perform normalization processing on the denoised data to obtain preprocessed data.

[0072] Step 103: Perform feature extraction on the pre-processed data to extract time domain features and frequency domain features, obtain energy distribution at different scales through wavelet transform, capture electrical signal anomalies, and obtain multi-dimensional electrical feature data;

[0073] Step 104: Input the multi-dimensional electrical characteristic data into the intelligent diagnosis model, diagnose the building electrical safety status through the intelligent diagnosis model, and output the safety diagnosis result;

[0074] Step 105: Determine the building electrical safety risk level according to the safety diagnosis result, and send corresponding warning information based on the safety risk level.

[0075] In the embodiment of the present invention, through the intelligent diagnosis model, whether it is a small leakage caused by line aging, a gradual change of electrical parameters caused by long-term overload, or complex hidden faults such as harmonic interference caused by nonlinear loads, they can be accurately identified, greatly shortening the maintenance time and reducing power outage losses; implement hierarchical management of building electrical facilities, and maintain regular monitoring for low-risk areas to avoid waste of resources caused by excessive operation and maintenance; when in a medium-risk area, start a detailed diagnosis process, specifically check for potential hidden dangers, and reasonably allocate manpower and material resources for precise maintenance; when it is determined to be a high-risk area, immediately take emergency measures such as power outages, and organize professional teams for emergency repairs. This risk-based dynamic operation and maintenance strategy enables operation and maintenance resources to be effectively managed. The optimal configuration ensures that operation and maintenance efficiency is maximized and operation and maintenance costs are reduced while ensuring electrical safety. It effectively prevents safety accidents such as electrical fires, equipment damage, and power outages caused by various factors such as line failures, equipment aging, and power quality problems, providing a solid guarantee for the safety of life and property of people in the building. Through real-time monitoring and intelligent diagnosis, it ensures that key electrical systems such as lighting, air conditioning, elevators, and fire protection are always in good operating condition, avoiding the failure of these systems due to electrical faults, thereby ensuring the normal use of the building and improving the experience and satisfaction of building users. Whether it is a commercial building, a residential building, or a public facility building, it can play a stable role in its professional field.

[0076] See also Figure 2 , a schematic diagram of a second embodiment of a method for intelligent diagnosis of building electrical safety risks based on big data provided by an embodiment of the present invention, the method comprising:

[0077] Step 201: Obtain pre-processed data, and arrange the pre-processed data in chronological order to obtain a data sequence;

[0078] Step 202: Calculate the mean, standard deviation, and effective value of the data sequence, traverse the data sequence to find the maximum, minimum, and peak values, obtain the crest factor based on the ratio of the peak value to the effective value, calculate the difference between adjacent voltage values and the standard deviation of the difference to obtain the voltage fluctuation standard deviation, and integrate them to obtain the time domain characteristics;

[0079] Step 203: Apply the fast Fourier transform to convert the time-domain features of the data sequence into frequency-domain features. For the frequency-domain sequence, calculate the frequency positions of each harmonic according to the fundamental frequency, extract the amplitudes at the corresponding frequency positions, and statistically calculate the ratio of the amplitudes of each harmonic to the fundamental amplitude to obtain the harmonic frequency spectrum distribution.

[0080] Step 204: Determine the low-frequency range, extract the frequency-domain components within this range, and analyze the variations of the amplitudes and phases of the low-frequency components over time.

[0081] In this embodiment, the low-frequency range is below 100 Hz.

[0082] Step 205: Obtain the energy distribution at different scales through wavelet transform, capture electrical signal anomalies, and obtain multi-dimensional electrical feature data.

[0083] In this embodiment, the Haar wavelet is used to perform N-layer wavelet decomposition on the data sequence, and the signal is decomposed into an approximation component and a detail component; calculate the energies of the approximation component and the detail component at each layer, and integrate the extracted time-domain features, frequency-domain features, and the energy distributions at different scales obtained by wavelet transform to form a multi-dimensional feature vector as the final multi-dimensional electrical feature data.

[0084] Please refer to Figure 3 , the schematic diagram of the third embodiment of the intelligent diagnosis method for building electrical safety risks based on big data provided by the embodiment of the present invention. The method includes:

[0085] Step 301: Construct an intelligent diagnosis model based on the DNN network and the LSTM network, and input the multi-dimensional electrical feature data into the intelligent diagnosis model.

[0086] In this embodiment, multi-modal data of historical multi-source sensors is obtained, and a dynamic feature selection algorithm is used to screen the collected multi-modal data, evaluate the importance of each feature, obtain features highly relevant to building electrical safety diagnosis, and remove redundant features, where the multi-modal data is the operating state of the building electrical system under different working conditions in the past; the attention mechanism is used to dynamically adjust the weights of the data obtained after feature screening, and an adaptive noise reduction method combining wavelet transform and generative adversarial network is used for noise reduction processing to obtain a training set; the training set is input into the DNN network for training, and a self-attention mechanism is introduced into the DNN network to focus on the mutual relationship between different features; the output of the DNN network is used as the input of the LSTM network, and a Peephole connection and Zoneout regularization are introduced into the LSTM network, where the Peephole connection adds additional connections to the forget gate, input gate, and output gate and can receive information of the cell state; a differentiable gating mechanism is used to construct a dynamic feature fusion layer to fuse the features output by the DNN network and the LSTM network, and weights are dynamically assigned according to the importance of different features in the current diagnosis task; a four-dimensional optimization objective function is constructed, and the MOSHO algorithm is used for hyperparameter optimization to obtain the optimal solution to construct an intelligent diagnosis model.

[0087] In this embodiment, the four indicators of the four-dimensional optimization objective function are determined as diagnostic accuracy, inference speed, memory occupancy, and model stability, and weights are assigned to the four indicators; a certain number of particles are randomly initialized in the hyperparameter space of the model to be optimized, each particle represents a set of hyperparameter combinations, and an initial velocity and position are assigned to each particle; according to the initial hyperparameter combinations and the corresponding objective function values, the Bayesian model is initialized for training, and the hyperparameter combinations represented by each particle in the current particle swarm are applied to the intelligent diagnosis model for training and evaluation, and the corresponding four-dimensional optimization objective function values are calculated; for each particle, compare the objective function value at the current position of each particle with the objective function value at the historical optimal position. If the objective function value at the current position is better, update the historical optimal position of the particle; compare the objective function values at the historical optimal positions of all particles, and select the position with the optimal objective function value as the global optimal position; the update rule of the particle swarm is improved by introducing quantum behavior simulation, and using the trained Bayesian model, hyperparameter combinations with better objective function values are predicted in the hyperparameter space, and the predicted hyperparameter combinations are added to the particle swarm as new particles; the hyperparameter combinations of the newly added particles and the corresponding objective function values are added to the training data of the Bayesian model, and the Bayesian model is retrained; check whether the current number of iterations reaches the upper limit. If the termination condition is met, stop the iteration; otherwise, return to continue the iterative optimization; after the iteration ends, the hyperparameter combination represented by the global optimal position is the optimal solution of the four-dimensional optimization objective function, and the intelligent diagnosis model is constructed using the optimal hyperparameter combination.

[0088] Step 302: The multi-dimensional electrical feature data is input into the input layer of the DNN network, and layer-by-layer calculations are performed in the hidden layer of the DNN network. For each neuron in each layer, the output of the previous layer is used as the input, weighted summation is performed, and a bias is added. A non-linear factor is introduced through the ReLU activation function to obtain the output of this layer. After calculations through multiple hidden layers, the output of the output layer of the DNN network is obtained;

[0089] Step 303: The output of the output layer of the DNN network is used as the input of the LSTM network. At each time step, the LSTM network calculates based on the current input, the hidden state at the previous moment, and the cell state to capture the time series correlation of electrical parameters. After processing all time steps, the final output of the LSTM network is obtained;

[0090] Step 304: The final output of the LSTM network is used to calculate the probabilities of the building electrical being in different safety states through the Softmax function. According to the calculated probabilities, the safety state with the highest probability is selected as the current safety state of the building electrical. The determined safety state and the corresponding probability values are sorted out to output the safety diagnosis result.

[0091] Please refer to Figure 4 , the structural schematic diagram of the intelligent diagnosis system for building electrical safety risks based on big data provided by the embodiments of the present invention. The system includes a data acquisition module, a data preprocessing module, a feature extraction module, an intelligent diagnosis module, and an early warning module. Among them, the data acquisition module is used to collect the original parameter data of the building electrical system in real time through a variety of sensors, connect with the building automation system to obtain the operation status information, connect with the power quality monitor to collect the power quality index data, and integrate the original parameter data, operation status information, and power quality index data to obtain the operation condition data of the building electrical system;

[0092] The data preprocessing module is used to clean the collected operation condition data to remove noise, and perform normalization processing on the denoised operation condition data to obtain the preprocessed data;

[0093] The feature extraction module is used to extract features from the preprocessed data, extract time-domain features and frequency-domain features, obtain the energy distribution at different scales through wavelet transform, capture electrical signal anomalies, and obtain multi-dimensional electrical feature data;

[0094] The intelligent diagnosis module is used to input the multi-dimensional electrical feature data into the intelligent diagnosis model, diagnose the safety state of the building electrical through the intelligent diagnosis model, and output the safety diagnosis result;

[0095] The early warning module is used to determine the building electrical safety risk level according to the safety diagnosis result, and send corresponding early warning information based on the safety risk level.

[0096] In this embodiment, the data preprocessing module includes a filling sub-module, a calculation sub-module, a correction sub-module, and a normalization processing sub-module. Among them, the filling sub-module is used to obtain the collected operating condition data, traverse the operating condition data to check whether there are missing values in each data record, and fill the detected missing values using the linear interpolation method. For each position where a missing value is located, the missing value is estimated through linear calculation based on the valid data points adjacent before and after it.

[0097] The calculation sub-module is used to detect outliers in the filled data based on a filtering algorithm of statistical characteristics, calculate the mean and standard deviation of each data type, calculate the absolute value of the difference between each data point and the mean, and if the absolute value of the difference of a certain data point is greater than the threshold, the corresponding data point is determined as an outlier, where the threshold is three times the standard deviation.

[0098] The correction sub-module is used to correct the detected outliers by smoothing processing, and use a median filter to remove noise from the data after missing value and outlier processing to obtain denoised data.

[0099] The normalization processing sub-module is used to perform normalization processing on the denoised data using the Min-Max normalization method to obtain preprocessed data.

[0100] In this embodiment, the feature extraction module includes an arrangement sub-module, a traversal sub-module, a conversion sub-module, an analysis sub-module, and a capture sub-module. Among them, the arrangement sub-module is used to obtain the preprocessed data and arrange the preprocessed data in chronological order to obtain a data sequence.

[0101] The traversal sub-module is used to calculate the mean, standard deviation, and effective value of the data sequence, traverse the data sequence to find the maximum value, minimum value, and peak value, obtain the peak factor based on the ratio of the peak value to the effective value, calculate the difference between adjacent voltage values, and calculate the standard deviation of the difference to obtain the voltage fluctuation standard deviation, and integrate to obtain time-domain features.

[0102] The conversion sub-module is used to apply the fast Fourier transform to convert the time-domain features of the data sequence into frequency-domain features. For the frequency-domain sequence, calculate the frequency positions of each harmonic according to the fundamental frequency, extract the amplitudes at the corresponding frequency positions, and count the ratio of the amplitudes of each harmonic to the fundamental amplitude to obtain the harmonic frequency spectrum distribution.

[0103] The analysis sub-module is used to determine the low-frequency range, extract the frequency-domain components within this range, and analyze the changes in the amplitude and phase of the low-frequency components over time, where the low-frequency range is below 100 Hz.

[0104] The capture sub-module is used to obtain the energy distribution at different scales through wavelet transform, capture electrical signal anomalies, and obtain multi-dimensional electrical feature data.

[0105] The foregoing has shown and described the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments, and the above embodiments and the descriptions in the specification are only preferred examples of the present invention, and are not used to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of the present invention claimed is defined by the appended claims and their equivalents.

Claims

1. An intelligent diagnosis method for building electrical safety risks based on big data, characterized in that, The intelligent diagnosis method for building electrical safety risks based on big data includes the following steps: Obtain the operating condition data of the building electrical system through parameter raw data, operating status information, and power quality index data; Clean and normalize the collected operating condition data to obtain preprocessed data; Arrange the preprocessed data in chronological order to obtain a data sequence; Obtain the effective value and peak value of the data sequence, obtain the peak factor based on the ratio of the effective value to the peak value, obtain the voltage fluctuation standard deviation using the standard deviation of the difference between adjacent voltage values, and obtain the time-domain characteristics based on the peak factor and the voltage fluctuation standard deviation; After converting the time-domain characteristics of the data sequence into frequency-domain characteristics, determine the frequency positions of each harmonic according to the fundamental frequency, extract the amplitudes at the corresponding frequency positions, and count the proportion of the amplitudes of each harmonic to the fundamental amplitude to obtain the harmonic spectrum distribution; Perform multi-scale decomposition on the data sequence through wavelet transform to obtain the energy distribution at different scales, capture abnormal fluctuations in the electrical signal, and integrate the time-domain characteristics, frequency-domain characteristics, and energy distribution at different scales to obtain multi-dimensional electrical characteristic data; Input the multi-dimensional electrical characteristic data into the intelligent diagnosis model, diagnose the building electrical safety status through the intelligent diagnosis model, and output the safety diagnosis result; Determine the building electrical safety risk level according to the safety diagnosis result, and send corresponding warning information based on the safety risk level; The intelligent diagnosis model is constructed based on the DNN network and the LSTM network, and specifically includes: Perform feature screening on the collected operating status data under different working conditions; First, perform noise reduction processing on the operating status data after feature screening using an adaptive noise reduction method that combines wavelet transform and generative adversarial network, then dynamically adjust the weights of the data after noise reduction processing using an attention mechanism, and obtain a training set using the processed data; Input the training set into the DNN network for training, and introduce a self-attention mechanism into the DNN network; Use the output of the DNN network as the input of the LSTM network. The LSTM network introduces Peephole connections and Zoneout regularization. Among them, the Peephole connection adds connections for receiving cell state information to the forget gate, input gate, and output gate; Adopt a differentiable gating mechanism to fuse the features output by the DNN network and the LSTM network; Randomly initialize a certain number of particles in the hyperparameter space of the model to be optimized; Compare the objective function values of the historical optimal positions of all particles, and select the position with the optimal objective function value as the global optimal position; The hyperparameter combination represented by the global optimal position is the optimal solution of the four-dimensional optimization objective function, and use the optimal hyperparameter combination to construct the intelligent diagnosis model.

2. The intelligent diagnosis method for building electrical safety risks based on big data according to claim 1, characterized in that, The cleaning and normalization processing of the collected operating condition data to obtain preprocessed data includes: Obtain the collected operating condition data, traverse the operating condition data to check whether there are missing values in each data record, and use linear interpolation to fill in the detected missing values. For the location of each missing value, estimate the missing value through linear calculation based on the valid data points adjacent before and after it; Detect outliers in the filled data using a filtering algorithm based on statistical characteristics. First, calculate the mean and standard deviation of each data type, and then calculate the absolute value of the difference between each data point and the mean. If the absolute value of the difference of a certain data point is greater than the threshold, the corresponding data point is determined as an outlier, where the threshold is three times the standard deviation; Use smoothing processing to correct the detected outliers, and use a median filter to remove noise from the data after missing value and outlier processing to obtain denoised data; Use the Min-Max normalization method to normalize the denoised data to obtain preprocessed data.

3. The intelligent diagnosis method for building electrical safety risks based on big data according to claim 1, characterized in that Perform multi-scale decomposition on the data sequence through wavelet transform to obtain the energy distribution at different scales, and capture abnormal fluctuations in the electrical signal. Integrate the time-domain features, frequency-domain features, and energy distribution at different scales to obtain multi-dimensional electrical feature data, including: Perform N-layer wavelet decomposition on the data sequence using the Haar wavelet to decompose the signal into an approximation component and a detail component; Calculate the energy of the approximation component and the detail component of each layer, and integrate the extracted time-domain features, frequency-domain features, and the energy distribution at different scales obtained by wavelet transform to form a multi-dimensional feature vector as the final multi-dimensional electrical feature data.

4. The intelligent diagnosis method for building electrical safety risks based on big data according to claim 1, characterized in that, Input the multi-dimensional electrical feature data into the intelligent diagnosis model, diagnose the safety state of the building electrical through the intelligent diagnosis model, and output the safety diagnosis result, including: Construct an intelligent diagnosis model based on the DNN network and the LSTM network, and input the multi-dimensional electrical feature data into the intelligent diagnosis model; The multi-dimensional electrical feature data is input into the input layer of the DNN network and is calculated layer by layer in the hidden layer of the DNN network. For each neuron in each layer, the output of the previous layer is used as the input, weighted summation is performed, and a bias is added. The ReLU activation function is used to introduce non-linearity to obtain the output of this layer. After calculation through multiple hidden layers, the output of the output layer of the DNN network is obtained; Use the output of the output layer of the DNN network as the input of the LSTM network. At each time step, the LSTM network calculates based on the current input and combines the hidden state and cell state of the previous moment to capture the time series correlation of the electrical parameters. After processing all time steps, the final output of the LSTM network is obtained; Calculate the probability of the building electrical being in different safety states through the Softmax function for the final output of the LSTM network. According to the calculated probability, select the safety state with the highest probability as the current safety state of the building electrical, and organize the determined safety state and the corresponding probability value to output the safety diagnosis result.

5. The intelligent diagnosis method for building electrical safety risks based on big data according to claim 4, characterized in that The construction of the intelligent diagnosis model based on the DNN network and the LSTM network further includes: Construct a four-dimensional optimization objective function, use the MOSHO algorithm for hyperparameter optimization to obtain the optimal solution, and construct an intelligent diagnosis model.

6. The intelligent diagnosis method for building electrical safety risks based on big data according to claim 5, wherein The construction of the four-dimensional optimization objective function, using the MOSHO algorithm for hyperparameter optimization to obtain the optimal solution, and constructing an intelligent diagnosis model includes: Determine the four indicators of the four-dimensional optimization objective function as diagnostic accuracy, inference speed, memory occupancy, and model stability, and assign weights to the four indicators; Randomly initialize a certain number of particles in the hyperparameter space of the model to be optimized. Each particle represents a set of hyperparameter combinations, and assign an initial velocity and position to each particle; According to the initial hyperparameter combinations and the corresponding objective function values, initialize the training of the Bayesian optimization algorithm. Apply the hyperparameter combinations represented by each particle in the current particle swarm to the intelligent diagnosis model for training and evaluation, and calculate the corresponding four-dimensional optimization objective function values; For each particle, compare the objective function value at the current position of each particle with the objective function value at the historical optimal position. If the objective function value at the current position is better, update the historical optimal position of the particle; Compare the objective function values of the historical optimal positions of all particles, and select the position with the best objective function value as the global optimal position; Introduce a quantum behavior simulation mechanism, and use the trained initialized model to predict hyperparameter combinations with better objective function values in the hyperparameter space, and add the predicted hyperparameter combinations as new particles to the particle swarm; Add the hyperparameter combinations and the corresponding objective function values of the newly added particles to the training data based on the Bayesian optimization algorithm, and retrain; Check whether the current iteration number reaches the upper limit. If the termination condition is met, stop the iteration; otherwise, return to continue the iterative optimization; After the iteration ends, the hyperparameter combination represented by the global optimal position is the optimal solution of the four-dimensional optimization objective function, and use the optimal hyperparameter combination to construct an intelligent diagnosis model.

7. An intelligent diagnosis system for building electrical safety risks based on big data, characterized in that, The intelligent diagnosis system for building electrical safety risks based on big data includes a data acquisition module, a data preprocessing module, a feature extraction module, an intelligent diagnosis module, and an early warning module, where: The data acquisition module is used to obtain the operating condition data of the building electrical system through parameter raw data, operating status information, and power quality index data; The data preprocessing module is used to clean and normalize the collected operating condition data to obtain preprocessed data; A feature extraction module, which is used to arrange the preprocessed data in chronological order to obtain a data sequence; acquire the effective value and peak value of the data sequence, obtain the crest factor based on the ratio of the effective value to the peak value, obtain the voltage fluctuation standard deviation using the standard deviation of the difference between adjacent voltage values, and obtain the time-domain features based on the crest factor and the voltage fluctuation standard deviation; after converting the time-domain features of the data sequence into frequency-domain features, determine the frequency positions of each harmonic according to the fundamental frequency, extract the amplitudes at the corresponding frequency positions, and calculate the proportion of the amplitudes of each harmonic to the fundamental amplitude to obtain the harmonic spectrum distribution; perform multi-scale decomposition on the data sequence through wavelet transform to obtain the energy distribution at different scales, capture abnormal fluctuations in the electrical signal, and integrate the time-domain features, frequency-domain features, and energy distribution at different scales to obtain multi-dimensional electrical feature data; An intelligent diagnosis module, which is used to input the multi-dimensional electrical feature data into an intelligent diagnosis model, diagnose the safety state of building electricity through the intelligent diagnosis model, and output a safety diagnosis result: An early warning module, which is used to determine the safety risk level of building electricity according to the safety diagnosis result, and send corresponding early warning information based on the safety risk level; The intelligent diagnosis model is constructed based on a DNN network and an LSTM network, and specifically includes: Performing feature screening on the collected operation state data under different working conditions; First, perform noise reduction processing on the operation state data after feature screening using an adaptive noise reduction method that combines wavelet transform and a generative adversarial network, then dynamically adjust the weights of the data after noise reduction processing using an attention mechanism, and obtain a training set using the processed data; Input the training set into the DNN network for training, and introduce a self-attention mechanism into the DNN network; Use the output of the DNN network as the input of the LSTM network. The LSTM network introduces Peephole connections and Zoneout regularization. Among them, the Peephole connection adds connections for receiving information of the cell state to the forget gate, input gate, and output gate; Adopt a differentiable gating mechanism to fuse the features output by the DNN network and the LSTM network; Randomly initialize a certain number of particles in the hyperparameter space of the model to be optimized; Compare the objective function values of the historical optimal positions of all particles, and select the position with the optimal objective function value as the global optimal position; The hyperparameter combination represented by the global optimal position is the optimal solution of the four-dimensional optimization objective function, and an intelligent diagnosis model is constructed using the optimal hyperparameter combination.

8. The intelligent diagnosis system for building electrical safety risks based on big data according to claim 7, characterized in that, The data preprocessing module includes a filling sub-module, a calculation sub-module, a correction sub-module, and a normalization processing sub-module, where: The filling sub-module is used to obtain the collected operation condition data, traverse the operation condition data to check whether there are missing values in each data record, fill the detected missing values using linear interpolation, and for the position of each missing value, estimate the missing value through linear calculation based on the adjacent valid data points before and after it; A calculation sub-module, which is used to detect outliers in the filled data based on a filtering algorithm of statistical characteristics. First, calculate the mean and standard deviation of each data type, then calculate the absolute value of the difference between each data point and the mean. If the absolute value of the difference of a certain data point is greater than the threshold, the corresponding data point is determined as an outlier, where the threshold is three times the standard deviation; A correction sub-module, which is used to correct the detected outliers by using smoothing processing, and use a median filter to remove noise from the data after missing value and outlier processing to obtain denoised data; A normalization processing sub-module, which is used to perform normalization processing on the denoised data by using the Min-Max normalization method to obtain preprocessed data.

Images