Full-automatic carbon emission monitoring system and carbon emission monitoring method

Abstract

The invention relates to the field of data processing systems or methods for supervision, and provides a full-automatic carbon emission monitoring system and method, and the method comprises the following steps: S1, collecting an image of a monitored device through a data collection module, and determining the type of the device; s2, using a data acquisition module to acquire equipment operation related data in real time; s3, the data processing module processes the related data; s4, the analysis module converts the vibration signal after noise reduction into an equipment power consumption value by using a similar energy state mapping model; s5, the analysis module detects and analyzes the vibration signals through fractal dimensions; s6, the analysis module performs time sequence analysis on the historical power data and the real-time monitoring data of the equipment through LSTM, predicts a future power consumption value Ppredicted (t + 1), and dynamically adjusts the prediction model by using a historical prediction error e (t); s7, the analysis module calculates the carbon emission based on the power consumption value; and S8, the communication module uploads the monitoring result to the cloud for storage, and feeds back the monitoring result to the user terminal.

Description

Technical Field

[0001] The present invention relates to the field of data processing systems or methods for monitoring purposes, and in particular to a fully automatic carbon emission monitoring system and a carbon emission monitoring method. Background Art

[0002] As the world pays more and more attention to carbon emission control, the construction industry, as a high-energy consumption field, has an increasingly urgent need for carbon emission monitoring. However, traditional carbon emission monitoring technologies mostly rely on single sensors and fixed power estimation models. Such methods have problems such as low accuracy and poor adaptability, and it is difficult to meet the monitoring requirements of diversified construction equipment and complex working conditions. In particular, the vibration characteristics of construction equipment in different environments and load conditions are significantly different. Traditional monitoring systems have difficulty capturing these dynamic changes, resulting in inaccurate carbon emission calculations.

[0003] For example, Chinese patent CN118396243B discloses a carbon emission intelligent monitoring method and system. Although it performs multi-dimensional identification of carbon emission sources in the monitoring area and generates a unique multi-dimensional identification code for each carbon emission source; monitors the carbon emission data and environmental quality data of each carbon emission source, and calculates the independent carbon emission index and independent environmental quality index of each carbon emission source respectively; calculates the comprehensive carbon emission index and comprehensive environmental quality index in the monitoring area based on the independent carbon emission index and independent environmental quality index of each carbon emission source, it ignores data processing and calibration. At the same time, existing equipment often cannot automatically adjust monitoring parameters according to different types of equipment and working conditions, resulting in poor robustness of the equipment in actual use.

[0004] In addition, the prior art also has the following defects:

[0005] 1. Traditional monitoring equipment relies on a single sensor (such as an electricity meter or a simple vibration sensor) to estimate the energy consumption and carbon emissions of the equipment, which cannot accurately reflect the dynamic changes in complex working conditions.

[0006] 2. Common algorithms are mostly based on fixed parameters (such as line energy consumption models) and cannot adapt to changes in the environment and equipment operating conditions, resulting in inaccurate monitoring results.

[0007] 3. Existing monitoring equipment is usually designed for use in specific scenarios and is difficult to work efficiently in dynamic environments (such as construction sites or scenarios where multiple devices are running simultaneously).

[0008] 4. Traditional systems are mainly based on data collection, lack intelligent analysis and feedback mechanisms, and cannot provide optimization suggestions for equipment operations.

[0009] The present invention is made to solve the common problems in the field, such as poor adaptability to a single scene, inability to realize dynamic monitoring, insufficient monitoring accuracy, low intelligence, lack of real-time correction and optimization capabilities, poor equipment adaptability, etc. Summary of the invention

[0010] The purpose of the present invention is to propose a fully automatic carbon emission monitoring system and a carbon emission monitoring method in view of the current deficiencies.

[0011] In order to overcome the shortcomings of the prior art, the present invention adopts the following technical solutions:

[0012] A carbon emission monitoring method, the carbon emission monitoring method comprising the following steps:

[0013] S1. Collect images of the monitored equipment through the data acquisition module, determine the equipment type, and dynamically initialize the corresponding carbon emission algorithm parameters;

[0014] S2. Use the data acquisition module to collect equipment operation related data in real time;

[0015] S3, the data processing module normalizes the collected equipment operation related data;

[0016] S4, the analysis module uses a similar energy state mapping model to convert the noise-reduced vibration signal into a device power consumption value;

[0017] S5, the analysis module analyzes the vibration signal through fractal dimension detection to dynamically capture the working condition characteristics of the equipment;

[0018] S6. The analysis module uses the LSTM long short-term memory network to perform time series analysis on the historical power data and real-time monitoring data of the equipment to predict the future power consumption value P. predicted (t+1), the memory of the time series data is updated through the forget gate, input gate and output gate, and the prediction model is dynamically adjusted using the historical prediction error e(t);

[0019] S7, the analysis module calculates the carbon emissions based on the power consumption value;

[0020] S8. The communication module uploads the monitoring results to the cloud storage and feeds them back to the user terminal.

[0021] Optionally, the carbon emission monitoring method further includes: image data of at least three angles of the detected device collected in step S1.

[0022] Optionally, the carbon emission monitoring method also includes: equipment operation related data includes vibration signals, noise signals and temperature and humidity data.

[0023] Optionally, the carbon emission monitoring method further includes: in step S4, a mapping relationship between the vibration signal characteristics and the power consumption is established through the historical operation data of the equipment to obtain the reference power value E0, the maximum signal strength, and the capacity state mapping coefficient κ of the equipment, and the noise-reduced vibration signal is converted into the equipment power consumption value E according to the following formula: total :

[0024]

[0025] In the formula, E mapped (t) is the instantaneous energy consumption, and T is the time window length.

[0026] Optionally, the carbon emission monitoring method further includes: in step S6, the carbon emission per unit time CO2(t) is calculated according to the following formula:

[0027] CO2(t)=β·P corrected (t);

[0028] Where P corrected (t) is the corrected instantaneous power consumption value, β is the carbon emission coefficient, and its value is a function of the type of energy used when the equipment is running;

[0029] Calculate the final carbon emissions based on the carbon emissions per unit time:

[0030]

[0031] Where CO2(t) is the carbon emission per unit time.

[0032] Optionally, the carbon emission monitoring method further includes: in step S7,

[0033] In step S7, the vibration-related data within the set time window is extracted from the historical data, and the vibration-related data includes the mapped energy value Emapped(t), the fractal dimension adjustment value ΔD, the skewness Skew, the kurtosis Kurt, the short-time mean and the short-time standard deviation, the vibration frequency distribution f(t), and the power consumption estimation initial value Praw(t), to form a time series data set Xltms:

[0034] X ltms (t) = {x t-W+1 , x t-W+2 , ..., x t};

[0035] Among them, x t-W+1 to x t is the vibration signal data of the past W time steps.

[0036] Optionally, the carbon emission monitoring method further includes: in step S5, using a fractal dimension algorithm to process the vibration signal to determine a current fractal dimension D of the vibration signal. f , and compare D f and the reference fractal dimension D ref , calculate the fractal dimension deviation ΔD;

[0037] If D f If the system exceeds the preset threshold range, an alarm will be issued to prompt that the equipment may have abnormal operating conditions.

[0038] Optionally, the carbon emission monitoring method further comprises dynamically adjusting P according to the fractal dimension deviation ΔD, skewness, kurtosis, and the mapping relationship between the vibration signal and the power. corrected .

[0039] In addition, the present invention also provides a fully automatic carbon emission monitoring system, which includes a data acquisition module, a data processing module, an analysis module, and a communication module. The data acquisition module collects data on equipment operating status and environmental conditions in real time. The data processing module cleans, preprocesses and extracts features from the collected data. The analysis module analyzes the processed data to form an analysis result. The communication module feeds back the carbon emission historical trend chart and energy-saving optimization suggestions to the user terminal.

[0040] Optionally, the data processing module processes the device operation related data collected and received based on the ESP32 development board.

[0041] The beneficial effects achieved by the present invention are:

[0042] 1. By analyzing the nonlinear characteristics of the equipment vibration signal, a new dynamic monitoring and anomaly detection method is provided, which not only improves the sensitivity and real-time performance of the system, but also enhances the system's carbon emission correction and energy-saving optimization capabilities. By combining with other data analysis and LSTM networks, fractal dimension analysis enables the system to more accurately predict future carbon emissions and optimize equipment operation modes in real time, thereby achieving high-precision carbon emission monitoring and dynamic adjustment.

[0043] 2. Through the cooperation of the data acquisition module and the data processing module, the data of the equipment operation status and environmental conditions collected in real time can be efficiently cleaned, preprocessed and feature extracted, ensuring the data quality, providing reliable input data for subsequent analysis, and ensuring that the entire system has the advantages of high-quality data processing and accurate analysis. 3. Through the cooperation of the data acquisition module and the data processing module, the analysis module and the communication module, the carbon emission historical trend chart and energy-saving optimization suggestions can form a closed-loop feedback from the device end to the cloud, and then to the user terminal, ensuring the continuous optimization of equipment energy efficiency and carbon emissions, and ensuring that the entire system has the advantages of intelligent optimization and decision support. 4. Through the cooperation of the analysis module and the communication module, the calculated carbon emissions and energy-saving optimization suggestions can be uploaded to the cloud storage in real time and fed back to the user terminal, ensuring the real-time nature of the data and the timeliness of the feedback, and ensuring that the entire system has the advantages of real-time monitoring and rapid response.

[0044] 5. Through the cooperation of data processing module and analysis module, the processed high-quality data can be used to accurately calculate the carbon emissions of equipment and predict carbon emission trends, provide accurate analysis results, and ensure that the entire system has the advantages of accurate carbon emission monitoring and optimization suggestions. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The present invention can be further understood from the following description in conjunction with the accompanying drawings. The components in the figures are not necessarily drawn to scale, but the emphasis is placed on illustrating the principles of the embodiments. In different views, the same reference numerals designate the same parts.

[0046] Figure 1 It is a block diagram of the entire system of the present invention.

[0047] Figure 2 It is a schematic diagram of the analysis flow block of the analysis module of the present invention.

[0048] Figure 3 It is a schematic diagram of the process of the carbon emission monitoring method of the present invention. DETAILED DESCRIPTION

[0049] The following is an explanation of the embodiments of the present invention through specific embodiments. Those skilled in the art can understand the advantages and effects of the present invention from the contents disclosed in this specification. The present invention can be implemented or applied through other different specific embodiments, and the details in this specification can also be modified and changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. In addition, the drawings of the present invention are only simple schematic illustrations and are not depicted according to actual sizes. It is stated in advance. The following embodiments will further explain the relevant technical contents of the present invention in detail, but the disclosed contents are not intended to limit the scope of protection of the present invention.

[0050] Embodiment 1: According to Figure 1 , Figure 2 , Figure 3 As shown, this embodiment provides a carbon emission monitoring method, which includes the following steps:

[0051] S1. Collect images of the monitored equipment through the data acquisition module, determine the equipment type, and dynamically initialize the corresponding carbon emission algorithm parameters;

[0052] The process of identifying the type of the device through image data requires the use of a classification algorithm, and the classification algorithm is a technical means well known to those skilled in the art;

[0053] Specifically, the following steps are included:

[0054] 1) Image preprocessing: Denoising: Use image denoising algorithms (such as Gaussian blur, mean filtering, etc.) to remove noise from the image to ensure the accuracy of feature extraction. Size normalization: Adjust the input image to a fixed size (such as 224×224 or 512×512) to facilitate subsequent deep learning model input. Contrast enhancement: Enhance the image contrast according to the acquisition environment to ensure that the device features are more obvious. Among them, contrast enhancement of images is a technical means well known to technicians in this field, so in this embodiment, they will not be described one by one.

[0055] 2) Classify images by using deep learning models (such as convolutional neural networks CNN, MobileNet, ResNet or VGG). The input of the network is processed image data, and the output is the category label of the device. The training data set: It is necessary to prepare an image data set containing a variety of equipment (such as tower cranes, excavators, transport vehicles, etc.), and annotate the image of each device. Training process: Use this data set to train the classification model and optimize the network weights through error back propagation.

[0056] Transmit newly acquired image data from multiple angles to the trained deep learning model, and identify and classify the devices in real time;

[0057] After the image is collected, the trained classification model is used to identify the type of equipment in real time. For example, the model can identify whether the equipment is a "tower crane" or an "excavator" and return the corresponding label. , thereby identifying the corresponding equipment type;

[0058] Optionally, the carbon emission monitoring method further includes: image data of at least three angles of the detected device collected in step S1.

[0059] Among them, in this embodiment, through multi-angle image acquisition: in order to improve the recognition accuracy, device images can be collected from different angles (front, side, top, etc.), and the images at multiple angles can be input into the classification model. The model makes more accurate predictions based on the characteristics of multiple perspectives.

[0060] At the same time, by combining image recognition with data from other sensors (such as vibration sensors, load sensors, etc.), the accuracy of equipment recognition is improved through a fusion model.

[0061] In addition, the classification results of the equipment directly affect the parameter initialization of the carbon emission algorithm. For example:

[0062] Tower crane: Initialize appropriate benchmark power value E0, dynamic carbon emission coefficient β, reference fractal dimension D according to its common load and operation mode ref ;

[0063] Excavator: Initialize the specific power mapping coefficient κ and carbon emission calculation formula according to the equipment's duty cycle and load.

[0064] S2. Use the data acquisition module to collect equipment operation related data in real time;

[0065] Optionally, the carbon emission monitoring method further includes: equipment operation related data includes but is not limited to the following signals: vibration signal, noise signal and temperature and humidity data.

[0066] The data acquisition module includes a vibration sensor, a noise sensor, a temperature sensor and a humidity sensor. The vibration sensor acquires vibration signals generated by the device during operation, the noise sensor monitors noise signals generated by the device, the temperature sensor acquires and monitors temperature data of the environment around the device, and the humidity sensor monitors humidity data of the environment around the device. The data storage device stores vibration data acquired by the vibration sensor, noise signals acquired by the noise sensor, temperature data acquired by the temperature sensor, and humidity data acquired by the humidity sensor.

[0067] S3, the data processing module normalizes the collected equipment operation related data;

[0068] The data processing module performs preprocessing on the collected equipment operation related data and the related operation related data, and the preprocessing includes filtering, denoising, normalization and other preprocessing steps to ensure data quality. Among them, filtering, denoising, normalization and other preprocessing steps are technical means well known to those skilled in the art, and therefore, they are not described one by one in this embodiment.

[0069] Meanwhile, in this embodiment, the normalization process is performed according to the following formula:

[0070]

[0071] In the formula, X normalized is the normalized data (e.g., vibration signal, noise signal, temperature and humidity data), max(X) is the maximum value in the data set, min(X) is the minimum value in the data set, and X is the original data (e.g., vibration signal, noise signal, temperature and humidity data).

[0072] Specifically, for vibration signal data:

[0073]

[0074] In the formula, S normalized is the normalized vibration signal, min(S) and max(S) are the minimum and maximum values ​​in the vibration signal data, that is, in a series of vibration signal amplitude data N(t) = [S1, S2, …, Sn], find the maximum and minimum values ​​of the vibration signal amplitude, and S is the original vibration signal data.

[0075] Normalization processing of noise signals includes:

[0076]

[0077] Where N normalized is the normalized noise signal, min(N) and max(N) are the minimum and maximum values ​​in the noise signal data, that is, in a series of noise signal amplitude data N(t) = [N1, N2, …, Nn], find the maximum and minimum values ​​of the noise signal amplitude, and N is the original noise signal data.

[0078] Normalization of temperature data includes:

[0079]

[0080] Where, T normalized is the normalized temperature data, min(T) and max(T) are the minimum and maximum values ​​in the temperature data, that is, in a series of temperature data, T(t) = [T1, T2, …, Tn], find the maximum and minimum values ​​of the temperature, and T is the original temperature data.

[0081] Normalization of humidity data:

[0082]

[0083] In the formula, H normalizedis the normalized humidity data, min(H) and max(H) are the minimum and maximum values ​​in the humidity data, that is, in a series of humidity data, H(t) = [H1, H2, …, Hn], find the maximum and minimum values ​​of the humidity, and H is the original humidity data.

[0084] S4, the analysis module uses a similar energy state mapping model to convert the noise-reduced vibration signal into a device power consumption value;

[0085] Optionally, the carbon emission monitoring method further includes: in step S4, a mapping relationship between the vibration signal characteristics and the power consumption is established through the historical operation data of the equipment to obtain the reference power value E0, the maximum signal strength, and the capacity state mapping coefficient κ of the equipment, and the noise-reduced vibration signal is converted into the instantaneous equipment power consumption value P according to the following formula: mapped (t):

[0086] P mapped (t) = κ·E mapped (t)+E0;

[0087] In the formula, E mapped (t) is the instantaneous energy consumption of the vibration signal, and T is the time window length.

[0088] Among them, the instantaneous energy consumption E of the vibration signal mapped (t) is calculated according to the following formula:

[0089] E mapped (t)=|S(t)| 2 ;

[0090] Where, S(t): represents the amplitude of the vibration signal of the equipment at time t;

[0091] In this embodiment, the vibration state of the device (such as acceleration, frequency, amplitude, etc.) is usually closely related to factors such as the workload and energy consumption of the device. When the load increases or the workload changes, the amplitude and frequency characteristics of the vibration signal usually change. By analyzing the vibration signal, the power consumption of the device can be inferred. For example, under high load, the amplitude and frequency of the vibration signal of the device may increase, so that the power demand of the device can be calculated.

[0092] The model assumes that the energy state of the vibration signal (i.e., the change in signal strength) has a certain similarity with the power consumption value of the device. The energy change of the vibration signal reflects the change in the operating state of the device, and these changes are converted into an estimate of the power consumption of the device through a specific mapping model.

[0093] The core idea of ​​energy state mapping is that by modeling the energy changes of vibration signals, the power consumption value of the equipment can be inferred.

[0094] The mapping coefficient κ can be dynamically adjusted according to the equipment type and working state. For example, the vibration characteristics of some equipment under high load may be different from those under low load, so the mapping coefficient needs to be adjusted according to different working conditions.

[0095] If there is a large deviation between the characteristics of the vibration signal and the expected power consumption value, other data (such as temperature and humidity, noise signals, etc.) can be used to further correct the mapping relationship.

[0096] Different types of equipment have different vibration characteristics, so the value range of the mapping coefficient is also different.

[0097] For heavy equipment, such as tower cranes and excavators, κ may be larger, usually in the range of 1.0 to 5.0.

[0098] For light equipment, such as power tools and appliances, κ may be small, typically in the range of 0.1 to 1.0.

[0099] The relationship between the strength of the vibration signal and power consumption changes when the device is in different load states. An increase in load usually leads to an increase in vibration intensity, so the mapping coefficients change with load changes.

[0100] Under low load conditions, the value of κ is usually smaller, indicating that the device consumes less power.

[0101] Under high load conditions, the value of κ is usually larger, indicating increased power consumption of the device.

[0102] Meanwhile, in this embodiment, the value of the mapping coefficient κ provides an example:

[0103] 1) In the scene where the tower crane lifts heavy objects:

[0104] Low load (no load or light load), when the tower crane is working with no load or light load, the vibration signal of the equipment is relatively small, the power consumption is low, and the mapping coefficient is small, then κ = 1;

[0105] High load (full load hoisting weight), when the tower crane is fully loaded, the vibration signal increases and the power consumption also increases, so the mapping coefficient increases, then κ = 3.5;

[0106] 2) During the excavation and shoveling process:

[0107] Low load (idling or light load operation), when the excavator is idling or light load operation, the vibration signal is small, the power consumption is low, and the mapping coefficient is small, then κ=1.2;

[0108] High load (shoveling, digging and other heavy load work), when the excavator is under high load, the vibration signal intensity increases, the power consumption also increases significantly, and the mapping coefficient increases accordingly, then κ = 4.0;

[0109] 3) When using electric tools (electric drills, electric saws) to drill or saw:

[0110] Low load (idling or light load operation): When the power tool is idling or light load operation, the vibration signal amplitude is small, the power consumption is low, and the mapping coefficient is small, then κ = 0.5;

[0111] High load (high load work), when the power tool works at high load, the vibration signal increases, and the power consumption increases accordingly, so the mapping coefficient will also increase, then κ = 1.0;

[0112] S5, the analysis module analyzes the vibration signal through fractal dimension detection to dynamically capture the working condition characteristics of the equipment;

[0113] Optionally, the carbon emission monitoring method further includes: in step S5, using a fractal dimension algorithm to process the vibration signal to determine a current fractal dimension D of the vibration signal. f , and compare D f and the reference fractal dimension D ref , calculate the fractal dimension deviation ΔD;

[0114] Wherein, the reference fractal dimension D ref Obtained through experiment or calibration, equivalent to a known value;

[0115] If D f If the system exceeds the preset threshold range, an alarm will be issued to prompt that the equipment may have abnormal operating conditions.

[0116] In this embodiment, the vibration signal includes multi-axis data or single-axis data;

[0117] The fractal dimension D of the vibration signal is calculated by using the fractal dimension f , whose value is calculated according to the following formula:

[0118]

[0119] Where k is the time step, L(k) is the curve length of the signal at the set time interval k, and its value is determined by the following steps:

[0120] The normalized vibration signal S(t) = [S1, S2, ..., SN] is divided into subsequences of length k. Specifically, the signal is divided into multiple sub-signal segments of length k:

[0121] S(1),S(2),...,S(k);

[0122] S(2),S(3),...,S(k+1);

[0123] …

[0124] S(N-k+1),S(N-k+2),...,S(N);

[0125] For each subsequence, the total deviation L'(k) of each subsequence is calculated, and its value is calculated according to the following formula:

[0126]

[0127] Where n is the total length of the signal sequence, S(i) is the value of the signal at time point i, and S(i+k) is the value of the signal at time point i+k.

[0128] Calculate the average curve length L(k) of all subsequences of length k:

[0129]

[0130] Where n is the total length of the signal sequence, S(i) is the value of the signal at time point i, and S(i+k) is the value of the signal at time point i+k.

[0131] The current fractal dimension D f With reference value D ref Compare and calculate the fractal dimension deviation △D:

[0132] ΔD=D f -D ref ;

[0133] If ΔD exceeds the preset threshold range, the system will consider that the equipment may have an abnormal operating condition;

[0134] Optionally, the carbon emission monitoring method further comprises dynamically adjusting P according to the fractal dimension deviation ΔD, skewness, kurtosis, and the mapping relationship between the vibration signal and the power. corrected .

[0135] Among them, the corrected instantaneous power consumption value P corrected (t) is calculated according to the following formula:

[0136] P corrected =P raw ×(1+α·Skew+β·Kurt+δ·ΔD);

[0137] Where P raw is the initial power estimate, ΔD=D f -D refrepresents the fractal dimension adjustment value, Skew is the skewness of the original vibration signal, Kurt is the kurtosis of the original vibration signal, α, β, δ are correction factors, which respectively control the influence weights of skewness, kurtosis and fractal dimension on the correction power.

[0138] The skew of the original vibration signal is calculated according to the following formula:

[0139]

[0140] Where Sclent(t) is the intensity of the vibration signal, μ is the mean of all sampling points in the vibration signal, and σ is the standard deviation of all sampling points in the vibration signal. In the above formula, E[·] is the expectation operator, which means taking the average of all signal values.

[0141] The kurtosis Kurt of the original vibration signal is calculated according to the following formula:

[0142]

[0143] Where Sclent(t) is the intensity of the vibration signal, μ is the mean of all sampling points in the vibration signal, and σ is the standard deviation of all sampling points in the vibration signal. In the above formula, E[·] is the expectation operator, which means taking the average of all signal values.

[0144] Among them, the mean μ of all sampling points in the vibration signal is calculated according to the following formula:

[0145]

[0146] Where S(i) is the vibration signal value of the i-th sampling point, and N is the total number of signal sampling points, that is, N discrete time series data are collected in the vibration signal.

[0147] The standard deviation σ of all sampling points in the vibration signal is calculated according to the following formula:

[0148]

[0149] Where S(i) is the vibration signal value of the i-th sampling point, and μ is the mean value of all sampling points in the vibration signal.

[0150] In this embodiment, the skewness reflects the symmetry of the signal, while the kurtosis reflects the degree of peaks. A skewness of zero indicates a symmetrical distribution, indicating that the left and right distributions are uniform. A positive skewness indicates that the distribution is elongated on the right side, and there are abnormally large values ​​on the right side, which may indicate that the device vibrates strongly under high load and the power consumption increases suddenly. A negative skewness indicates that the distribution is elongated on the left side, and there are abnormally large values ​​on the left side, which may indicate that the device operates under low load and has low power consumption. In the algorithm, if the absolute value of the skewness is large, the benchmark of power consumption can be corrected. For example, by adjusting the predicted power to make it more sensitive to high load and low load conditions. Kurtosis is a statistic that measures the steepness of the distribution shape, and is particularly sensitive to extreme values. The higher the kurtosis value, the sharper the distribution, which means that more extreme values ​​appear in the data. Specifically: A high kurtosis indicates that the data distribution is steep and has more extreme values, indicating that there may be sudden changes in the device vibration process, reflecting possible abnormal behavior. Low kurtosis indicates that the data distribution is relatively flat, indicating that the device vibration process is relatively stable. High kurtosis is usually associated with high volatility and abnormal device conditions. The system can correct the error tolerance of the model prediction through kurtosis, making it more robust in identifying sudden changes. Through non-Gaussian statistical feature analysis, the system can effectively identify the high-load working state of the equipment, thereby correcting the calculation of the power consumption value.

[0151] Through the coordination of fractal dimension analysis and vibration signal feature extraction, the complexity and irregularity of the equipment's operating status can be captured in real time, ensuring the rapid identification of equipment failures and abnormal operating conditions, and ensuring that the entire system has the advantages of high sensitivity and real-time anomaly detection.

[0152] Through the cooperation of the data acquisition module and the data processing module, the data on equipment operating status and environmental conditions collected in real time can be efficiently cleaned, preprocessed and feature extracted, ensuring data quality, providing reliable input data for subsequent analysis, and ensuring that the entire system has the advantages of high-quality data processing and accurate analysis.

[0153] S6. The analysis module uses the LSTM long short-term memory network to perform time series analysis on the historical power data and real-time monitoring data of the equipment to predict the future power consumption value P. predicted (t+1), the memory of the time series data is updated through the forget gate, input gate and output gate, and the prediction model is dynamically adjusted using the historical prediction error e(t);

[0154] Through the cooperation of the data processing module and the analysis module, the processed high-quality data can be used to accurately calculate the carbon emissions of the equipment and predict the carbon emission trends, provide accurate analysis results, and ensure that the entire system has the advantages of accurate carbon emission monitoring and optimization suggestions. Optionally, the carbon emission monitoring method also includes: in step S6, collecting historical data to extract vibration-related data within a set time window, the vibration-related data including the mapped energy value Emapped(t), the fractal dimension adjustment value ΔD, the skewness Skew, the kurtosis Kurt, the short-time mean and short-time standard deviation, the vibration frequency distribution f(t), and the power consumption estimation initial value Praw(t), to form a time series data set Xltms(t):

[0155] X ltms (t) = {x t-W+1 , x t-W+2 , ..., x t};

[0156] Among them, x t-W+1 to x t is the vibration signal data of the past W time steps.

[0157] Through this window data, the data window X ltms (t) is a time series set composed of multidimensional features within the past W time steps, wherein the multidimensional features include the mapped energy value Emapped(t), the fractal dimension adjustment value ΔD, the skewness Skew, the kurtosis Kurt, the short-time mean, the short-time standard deviation, the vibration frequency distribution f(t) and the power consumption estimation initial value Praw(t).

[0158] After the data window is extracted, the LTMS algorithm models the time series data and uses the long short-term memory (LSTM) neural network to predict the power consumption value at the next moment (i.e., the time series data related to power consumption). By memorizing the past data state, LSTM can more accurately predict future trend changes. The core of the LSTM network lies in the update of the hidden layer state, including the calculation of the forget gate, input gate, and output gate.

[0159] Assume h t is the hidden state of STM, C t is the unit state, which is predicted by the following formula:

[0160] Forget gate: determine the previous memory state C t-1 The part that should be retained.

[0161] f t =σ(W fg[h t-1 ,x t ]+b f );

[0162] In the formula, f t is the output of the forget gate, σ represents the activation function, which is generally a sigmoid function, and W f is the weight matrix, b f is the bias term.

[0163] Input gate: calculate the current input x t The importance of the data determines the memory that needs to be stored. In this embodiment, the current input is equivalent to the time series data set Xltms(t) described above;

[0164] i t =σ(W i g[h t-1 ,x t ]+b i );

[0165] In the formula, i t is the output of the input gate, W i is the weight matrix of the input gate, b i is the bias term of the input gate.

[0166] Candidate memory state: Generate candidate memory using current input and previous hidden state

[0167]

[0168] Where, tanh: hyperbolic tangent activation function, restricting the candidate cell state to the range of [-1,1], W c is the weight matrix of the cell state, b c is the bias term of the cell state.

[0169] Update memory state: Generate a new memory state C by combining the forget gate and the input gate t :

[0170]

[0171] Output gate: finally outputs the current hidden state h t , used to generate power prediction results.

[0172] h t =o t gtanh(C t );

[0173] In the formula, tanh(C t) is to map the cell state to the range of [-1,1] through the hyperbolic tangent activation function.

[0174] Among them, the output gate o t =σ(W o ·[h t-1 ,X t ]+b o ) controls the output ratio of the hidden state.

[0175] Among them, t is the output of the output gate, σ is the activation function, generally the sigmoid function, W0 is the weight matrix of the output gate, h t-1 is the hidden state of the previous time step t-1, containing the information of the previous time step. Initially, h0 (initial hidden state) is usually initialized to a zero vector, h0 = 0. o is the bias term, usually initialized to a zero vector

[0176] In this embodiment, the weight matrix (Wf, Wi, Wc, Wo) is used to weight the input and hidden state, and determine how to combine the output of the previous time step and the current input to calculate the output of each gate. Usually, the weight matrix is ​​obtained by random initialization. For example, using Xavier initialization or He initialization.

[0177] For each weight matrix W, its initial value is sampled from a normal distribution or uniform distribution with mean 0 and variance 1 / fan_in, where fan_in is the number of neurons in the previous layer.

[0178] During the training process of the LSTM network, the weight matrix is ​​updated according to the gradient of the loss function through the backpropagation algorithm. The optimization algorithm usually uses gradient descent or its variants (such as Adam, RMSprop).

[0179] In addition, the bias term (b f ,b i ,b c ,b o ) is used to control the output of each gate, ensuring that the output of the gate can be adjusted to the appropriate value even when the input is zero. The bias term is usually initialized to a zero vector, that is, b f =b i =b c =b o =0 to ensure that there is no deviation in the network status in the initial stage.

[0180] At the same time, the bias term is updated via back-propagation and optimized together with the weight matrix.

[0181] Based on this structure, the LSTM network calculates and outputs the predicted power value P through the input of time series data Xltms(t). ltms (t+1).

[0182] The short-term predicted power value P is calculated ltms After (t+1), in order to reduce the prediction error, the actual measurement value P actual (t) is fed back to the model and the predicted value is adaptively corrected.

[0183] P actual (t) represents the actual power consumption value of the device at time t. This value is directly used by the system to correct the predicted power consumption P predicted The reference value of (t) should directly reflect the energy consumption of the equipment under operating conditions. actual (t) is usually obtained by external energy metering devices (such as power sensors, current sensors or other energy consumption measurement modules). These devices are installed on the power supply line of the monitored equipment and directly measure the real-time power consumption of the equipment.

[0184] In this embodiment, specific actual measurement values ​​P are calculated for different electrical or mechanical equipment. actual (t);

[0185] Actual measured value P of electrical equipment actual (t) Calculate power through real-time measurement of current and voltage:

[0186] P actual (t) = V(t)·I(t)·cos(φ);

[0187] Wherein, V(t) is the voltage, whose value is obtained by the voltage value of the actual connection circuit of the electrical equipment, that is, the voltage sensor; I(t) is the current, whose value is obtained by the current value of the actual connection circuit of the electrical equipment, that is, the current sensor; cos(φ) is the power factor, whose value is input according to the actual connection circuit of the electrical equipment and obtained by the smart meter.

[0188] Mechanical equipment: By monitoring the fuel consumption F(t) converted into power:

[0189] P actual (t) = η·F(t)·LHV;

[0190] Where η is the equipment efficiency, which is obtained by querying the nameplate of the specific equipment used, and LHV is the lower heating value of the fuel, which is measured by a constant volume combustion calorimeter (Bomb Calorimeter).

[0191] The power-related data is collected in real time, and the actual measured value P is communicated with the core module (ESP32) of the system through the interface. actual (t) is passed to the model for correction.

[0192] The ESP32 board receives the measurement data from external sensors and uses it as input to correct the model's predictions.

[0193] The correction process itself does not directly use LSTM. LSTM is mainly used to predict the power value P at the next moment. predicted (t+1), and the correction process is based on the predicted value P at the current time t predicted (t) and the actual measured value P actual (t) is a feedback mechanism.

[0194] The purpose of correction is to adjust the model output by feedback error so that the model's prediction ability is gradually improved. The specific steps are as follows:

[0195] 1) Calculation error:

[0196] By comparing the predicted value P predicted (t) and the actual value P actual (t), calculate the correction error e(t)):

[0197] e(t)=P actual (t)-P predicted (t);

[0198] According to the error e(t), the feedback correction factor K(t) is calculated to adjust the output of the model. The correction factor can be fixed or dynamically changed, for example:

[0199] Where P predicted (t) is the predicted value, α is the learning rate, and ∈ is a small value to avoid the denominator being zero, and its value is 10 -6 .

[0200] Use the correction factor to adjust the predicted value P predicted (t), and obtain the corrected power consumption value P corrected (t):

[0201] P corrected (t) = P predicted (t)+K(t)·e(t);

[0202] It can also be corrected by the following formula:

[0203] P corrected (t+1)=P ltms (t+1)+α×(P actual (t)-Pltms (t));

[0204] Where P corrected (t+1) is the predicted value of power consumption after correction, α is the feedback correction factor, and its value range is 0<α<1, which is used to control the feedback amount to make the correction process smooth.

[0205] In this embodiment, the feedback correction factor α adopts a dynamic update mechanism: in order to make the correction factor α adapt to changes in complex working conditions, it is dynamically adjusted through the following formula:

[0206]

[0207] In the formula, δ is a small constant used to avoid the situation where the denominator is zero, and its value is 10 -6 , P ltms (t) is the output predicted power value, P actual (t) represents the actual power consumption value of the device at time t.

[0208] After multiple corrections and training, an accurate LSTM long short-term memory network is obtained;

[0209] When the time series data Xltms(t) collected for a certain device is input into the calibrated and trained LSTM long short-term memory network, the final calibrated power consumption value P is obtained. corrected (t), the value is input into the analysis module and step S7 is executed;

[0210] Through the coordination of fractal dimension and carbon emission correction, the equipment can adjust the carbon emission prediction in real time according to the changes in its vibration signal during operation, ensuring the accuracy and dynamic adjustment capability of carbon emissions, and ensuring that the entire system has the advantages of accurate carbon emission monitoring and dynamic optimization.

[0211] Through the cooperation of fractal dimensionality analysis and LSTM network, the system can more efficiently perform trend prediction and carbon emission optimization, and use the historical vibration signals and power data of the equipment to generate more accurate future carbon emission forecasts, helping users to formulate long-term energy-saving plans and ensure that the entire system has the advantages of long-term energy-saving prediction and optimization.

[0212] S7, the analysis module calculates the carbon emissions based on the power consumption value;

[0213] Optionally, the carbon emission monitoring method further includes: in step S6, the carbon emission per unit time CO2(t) is calculated according to the following formula:

[0214] CO2(t)=β·P corrected (t);

[0215] Where P corrected (t) is the corrected instantaneous power consumption value, β is the carbon emission coefficient, and its value is a function of the type of energy used when the equipment is running;

[0216] Calculate the final carbon emissions based on the carbon emissions per unit time:

[0217]

[0218] Where CO2(t) is the carbon emission per unit time.

[0219] Among them, β is a dynamic parameter set according to the equipment type. Different equipment types have different carbon emission coefficients. The carbon emission coefficient (β) refers to the carbon dioxide emissions generated per unit power consumption, usually expressed in kgCO 2 The unit is / kWh. The carbon emission coefficient of different equipment depends on its energy type (such as electricity, diesel, natural gas) and operating efficiency.

[0220] The following is an example of typical carbon emission factor data associated with equipment types:

[0221] 1) Carbon emission coefficient β for tower crane:

[0222] The energy consumption of tower cranes mainly comes from electric drive, and the carbon emission coefficient is closely related to the carbon intensity of local electricity production.

[0223] Assuming the average carbon emission intensity of the local power grid is 0.6kgCO 2 / kWh (International Energy Agency IEA global grid average). The power consumption PPP of a tower crane is usually in the range of 30-50kW, depending on the workload and operation mode. The carbon emission factor β of a tower crane is calculated as: β tower crane = 0.6kgCO 2 / kWh;

[0224] 2) In the diesel generator (diesel driven) scenario, β 柴油发电机 =2.68kgCO 2 / kWh;

[0225] Value range: 2.5kgCO 2 / kWh to 3.0kgCO 2 / kWh;

[0226] Specifically, the carbon emission coefficient of diesel-driven equipment is relatively high, and its value is closely related to the fuel type and equipment efficiency.

[0227] 3) In the scenario of natural gas generator (natural gas driven), β 天然气发电机 =1.92kgCO 2 / kWh;

[0228] The value range of natural gas generator: 1.8kgCO 2 / kWh to 2.0kgCO 2 / kWh;

[0229] Specifically, natural gas burns cleaner than diesel and therefore has a lower carbon emission factor.

[0230] 4) In the wind power (green power) scenario, β 风电 =0.05kgCO 2 / kWh;

[0231] The value range of natural gas generator: less than 0.1kgCO 2 / kWh;

[0232] Specifically, the carbon emission coefficient of clean energy such as wind power and solar energy is very low, usually close to 0. In this embodiment, the carbon emission coefficient β needs to be set according to the type of equipment to be monitored and the actual situation of the equipment. This is a technical means well known to those skilled in the art, and those skilled in the art can refer to relevant technical manuals to learn about the technology, so it is not described in detail in this embodiment.

[0233] S8, the communication module uploads the monitoring result to the cloud storage and feeds back to the user terminal. In this embodiment, the monitoring result is set to the calculated carbon emission.

[0234] In addition, the present invention also provides a fully automatic carbon emission monitoring system, which includes a data acquisition module, a data processing module, an analysis module, and a communication module. The data acquisition module collects data on equipment operating status and environmental conditions in real time. The data processing module cleans, preprocesses and extracts features from the collected data. The analysis module analyzes the processed data to form an analysis result. The communication module feeds back the carbon emission historical trend chart and energy-saving optimization suggestions to the user terminal.

[0235] Optionally, the data processing module processes the device operation related data collected and received based on the ESP32 development board.

[0236] The fully automatic carbon emission monitoring system also includes a central processing unit and a server, which are respectively connected to the data acquisition module, data processing module, analysis module, and communication module, and store the intermediate data and process data of the data acquisition module, data processing module, analysis module, and communication module in the server's database (cloud) for query and call.

[0237] The central processing unit is respectively controlled and connected with the data acquisition module, data processing module, analysis module, and communication module, and centrally controls the data acquisition module, data processing module, analysis module, and communication module based on the central processing unit, and stores the control data in the server, thereby improving the accuracy and reliability of carbon emission monitoring of the entire system.

[0238] The data acquisition module includes a sensor component and a data storage device. The sensor component collects relevant data of the monitoring equipment. The data storage device stores the relevant data collected by the sensor component. The relevant data includes power value, voltage value, temperature value, humidity value, current value, vibration signal data and noise signal data.

[0239] The sensor assembly includes an image sensor, a vibration sensor, a noise sensor, a temperature sensor, a humidity sensor, a power sensor, a voltage sensor, and a current sensor. The image sensor collects image data of the monitoring device. The vibration sensor is used to monitor the vibration signal data generated during the operation of the device. The noise sensor is used to monitor the noise signal data generated during the operation of the device. The temperature sensor is used to monitor the temperature data inside the device. The humidity sensor is used to monitor the humidity data inside the device. The power sensor is used to monitor the power consumption of the device. The voltage sensor is used to monitor the voltage data of the device. The current sensor is used to monitor the current data of the device.

[0240] The data processing module includes an ESP32 development board and a data processing unit. The data processing unit is loaded on the ESP32 development board and processes the relevant data collected by the data acquisition module.

[0241] The processing operations of the data processing unit include digital-to-analog / analog-to-digital format conversion, cleaning, preprocessing and feature extraction to obtain input data with higher quality.

[0242] At the same time, in this embodiment, the data processing unit also performs normalization processing on the relevant data.

[0243] The communication module includes a communicator and a data receiver. The data receiver receives the carbon emission results calculated by the analysis module and transmits them to the communicator. The communicator stores the carbon emission results collected by the data receiver in the cloud and feeds back to the user terminal. Through the cooperation of the analysis module and the communication module, the calculated carbon emissions and energy-saving optimization suggestions can be uploaded to the cloud storage in real time and fed back to the user terminal, ensuring the real-time nature of the data and the timeliness of the feedback, and ensuring that the entire system has the advantages of real-time monitoring and rapid response. Through the cooperation of the data acquisition module and the data processing module, and the analysis module and the communication module, the carbon emission historical trend chart and energy-saving optimization suggestions can be transmitted from the device end to the cloud, and then to the user terminal, forming a closed-loop feedback, ensuring the continuous optimization of equipment energy efficiency and carbon emissions, and ensuring that the entire system has the advantages of intelligent optimization and decision support.

[0244] In this embodiment, by analyzing the nonlinear characteristics of the equipment vibration signal, a new dynamic monitoring and anomaly detection method is provided, which not only improves the sensitivity and real-time performance of the system, but also enhances the system's carbon emission correction and energy-saving optimization capabilities. By combining with other data analysis and LSTM networks, fractal dimension analysis enables the system to more accurately predict future carbon emissions and optimize equipment operation modes in real time, thereby achieving high-precision carbon emission monitoring and dynamic adjustment.

[0245] Embodiment 2: This embodiment should be understood to include all the features of any of the above embodiments, and further improve upon them. Figure 1 , Figure 2 and Figure 3 As shown, the automatic emission monitoring system also includes an evaluation module, which is connected to the server and stores the intermediate data and process data of the evaluation module in the database of the server for query and call.

[0246] At the same time, the evaluation module is also connected to the central processing unit control, and the evaluation module is centrally controlled based on the central processing unit, so that the evaluation module can evaluate the carbon emissions.

[0247] The evaluation module obtains the carbon emissions in a monitoring period T and calculates the growth index G(t) according to the following formula:

[0248] G(t)=|R(t)|·e λ·|A(t)| ;

[0249] Where R(t) is the growth rate at the current moment, A(t) is the change acceleration at the current moment, and λ is the adaptive attenuation factor, which ranges from [0 to 1].

[0250] Among them, the growth rate R(t) at the current moment is calculated according to the following formula:

[0251] Where CO2(t) is the carbon emissions at time t, and CO2(t-1) is the carbon emissions at time t-1.

[0252] The current changing acceleration A(t) is calculated according to the following formula:

[0253] A(t) = R(t) - R(t-1);

[0254] In the formula, R(t) is the growth rate at the current moment, and R(t-1) is the growth rate at the previous moment t-1. Its value can be calculated by referring to the growth rate at the current moment;

[0255] When λ is large (close to 1), the impact of acceleration is stronger, and the sharp increase in carbon emissions will lead to a rapid increase in the index, triggering a more sensitive warning.

[0256] When λ is small (close to 0), the impact of acceleration is weak, and the system responds more gently to changes in carbon emissions. The warning is triggered only when the changes are significant.

[0257] In this embodiment, an example of a value of the adaptive attenuation factor λ is provided, specifically:

[0258] 1) In the scenario of high-load and large-fluctuation equipment (such as construction tower cranes), the tower cranes usually have large load fluctuations. Especially when the load increases rapidly or the operation is interrupted, the carbon emissions of the equipment may increase sharply, then λ = 0.7.

[0259] 2) In the scenario of medium load and relatively stable equipment (such as power generators), the load fluctuation of the power generator is relatively small, the equipment status is relatively stable, and the main focus is on the gradual change of carbon emissions, then λ = 0.3.

[0260] 3) In the scenario of low load and very stable equipment (such as household appliances and low-power equipment), for small equipment or household appliances, the load fluctuation is very small and the carbon emissions are relatively stable, then λ = 0.1.

[0261] When the equipment load fluctuates greatly or is in an unstable operating state, a larger λ should be selected to enhance the sensitive response to changes in carbon emissions; when the equipment is relatively stable, a smaller λ can be selected to avoid overly drastic responses.

[0262] In short, the value of the element-adaptive attenuation factor needs to be set according to the specific situation of the monitored equipment. This is a technical means well known to technicians in this field. Technicians in this field can refer to relevant technical manuals to learn about this technology, so it will not be described one by one in this embodiment.

[0263] If the growth index G(t) exceeds the monitoring threshold range M set by the system, an early warning is triggered and feedback is given to the user terminal;

[0264] If the growth index G(t) is lower than or equal to the monitoring threshold range M set by the system, it means that the system is in a normal state and no warning needs to be triggered.

[0265] Among them, the monitoring threshold range M set by the system is set by the administrator or the system according to the actual situation. This is a technical means well known to technicians in this field. Technicians in this field can refer to relevant technical manuals to learn about the technology, so it will not be described one by one in this embodiment.

[0266] Specifically, when G(t)>1.2M, the system issues a low-level warning, indicating that carbon emissions have changed significantly and the administrator needs to pay attention to the operation of the equipment.

[0267] When G(t)>1.5M, the system issues an advanced warning, warning that the equipment may fail, be overloaded, or operate abnormally, and the administrator needs to take immediate action.

[0268] The above-mentioned early warning mechanism can effectively avoid abnormal carbon emissions caused by excessive energy consumption or failure of equipment, provide users with timely response and adjustment measures, and ensure efficient operation and energy-saving management of equipment.

[0269] In this embodiment, through the cooperation of the evaluation module and the analysis module, the system can efficiently realize carbon emission monitoring, real-time data analysis, trend prediction and fault warning, and provide a comprehensive view and intelligent decision support for equipment operation. This cooperation not only improves the accuracy and real-time performance of carbon emission monitoring, but also enhances the system's prediction ability and fault response ability, and ultimately improves the energy saving effect and stability of the equipment.

[0270] The contents disclosed above are only preferred feasible embodiments of the present invention, and do not limit the protection scope of the present invention. Therefore, all equivalent technical changes made using the contents of the present invention specification and drawings are included in the protection scope of the present invention. In addition, the elements therein can be updated as technology develops.

Claims

1. A carbon emission monitoring method, characterized in that: The carbon emission monitoring method comprises the following steps: S1. Collect images of the monitored equipment through the data acquisition module, determine the equipment type, and dynamically initialize the corresponding carbon emission algorithm parameters; S2. Use the data acquisition module to collect equipment operation related data in real time; S3, the data processing module normalizes the collected equipment operation related data; S4, the analysis module uses a similar energy state mapping model to convert the noise-reduced vibration signal into a device power consumption value; S5, the analysis module analyzes the vibration signal through fractal dimension detection to dynamically capture the working condition characteristics of the equipment; S6. The analysis module uses the LSTM long short-term memory network to perform time series analysis on the historical power data and real-time monitoring data of the equipment to predict the future power consumption value P. predicted (t+1), the memory of the time series data is updated through the forget gate, input gate and output gate, and the prediction model is dynamically adjusted using the historical prediction error e(t); S7, the analysis module calculates the carbon emissions based on the power consumption value; S8. The communication module uploads the monitoring results to the cloud storage and feeds them back to the user terminal.

2. The carbon emission monitoring method according to claim 1, characterized in that: The carbon emission monitoring method further includes: collecting image data of at least three angles of the detected device in step S1.

3. The carbon emission monitoring method according to claim 2, characterized in that: The carbon emission monitoring method also includes: the equipment operation related data includes vibration signals, noise signals and temperature and humidity data.

4. The carbon emission monitoring method according to claim 3, characterized in that: The carbon emission monitoring method further includes: in step S4, a mapping relationship between the vibration signal characteristics and the power consumption is established through the historical operation data of the equipment to obtain the reference power value E0, the maximum signal strength, and the capacity state mapping coefficient κ of the equipment, and the noise-reduced vibration signal is converted into the equipment power consumption value E according to the following formula: total : In the formula, E mapped (t) is the instantaneous energy consumption, and T is the time window length.

5. The carbon emission monitoring method according to claim 4, characterized in that: The carbon emission monitoring method further includes: in step S6, the carbon emission per unit time CO2(t) is calculated according to the following formula: CO2(t)=β·P corrected (t); Where P corrected (t) is the corrected instantaneous power consumption value, β is the carbon emission coefficient, and its value is a function of the type of energy used when the equipment is running; Calculate the final carbon emissions based on the carbon emissions per unit time: Where CO2(t) is the carbon emission per unit time.

6. The carbon emission monitoring method according to claim 5, characterized in that: The carbon emission monitoring method further includes: in step S7, collecting historical data to extract vibration-related data within a set time window, wherein the vibration-related data includes a mapped energy value Emapped(t), a fractal dimension adjustment value ΔD, a skewness Skew, a kurtosis Kurt, a short-time mean and a short-time standard deviation, a vibration frequency distribution f(t), and a power consumption estimation initial value Praw(t), to form a time series data set Xltms: X ltms (t)={x t-W+1 ,x t-W+2 ,...,x t }; Among them, x t-W+1 to x t is the vibration signal data of the past W time steps.

7. The carbon emission monitoring method according to claim 6, characterized in that: The carbon emission monitoring method further comprises: in step S5, using a fractal dimension algorithm to process the vibration signal to determine a current fractal dimension D of the vibration signal. f , and compare D f and the reference fractal dimension D ref , calculate the fractal dimension deviation ΔD; If D f If the system exceeds the preset threshold range, an alarm will be issued to prompt that the equipment may have abnormal operating conditions.

8. The carbon emission monitoring method according to claim 7, characterized in that: The carbon emission monitoring method further includes dynamically adjusting P according to the fractal dimension deviation ΔD, skewness, kurtosis, and the mapping relationship between the vibration signal and the power. corrected .

9. A fully automatic carbon emission monitoring system, using the carbon emission monitoring method according to claim 8, characterized in that: The fully automatic carbon emission monitoring system includes a data acquisition module, a data processing module, an analysis module, and a communication module. The data acquisition module collects data on the equipment operating status and environmental conditions in real time. The data processing module cleans, preprocesses and extracts features from the collected data. The analysis module analyzes the processed data to form an analysis result. The communication module feeds back the carbon emission historical trend chart and energy-saving optimization suggestions to the user terminal.

10. The fully automatic carbon emission monitoring system according to claim 9, characterized in that: The data processing module receives and processes the collected device operation related data based on the ESP32 development board.

Images