PC component full life cycle carbon emission accounting method, medium and system

Abstract

The invention provides a PC component full-life-cycle carbon emission accounting method, medium and system, and belongs to the technical field of PC component full-life-cycle carbon emission, and the method comprises the steps: firstly, obtaining ultrasonic data, vibration transmission data and a high-definition image of a PC component, and carrying out the Fourier transform analysis of the data, so as to form three sine curve sets; and fusing the three sets, extracting fusion features, and inputting the fusion features into a pre-trained PC component production carbon emission model to obtain production stage carbon emission. And then, calculating carbon emissions in the stages of transportation, construction, use and dismantling recovery, and finally, obtaining a full-life-cycle carbon emission value of the PC component by adopting an addition mode. According to the scheme provided by the invention, multi-source monitoring data are fully fused, advanced signal processing and machine learning algorithms are adopted, accurate evaluation of carbon emission in each life cycle stage of the PC component is realized, and the technical problems of complex statistical process and low accuracy in the existing method are solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of carbon emissions during the entire life cycle of PC components, and specifically, relates to a method, medium and system for calculating carbon emissions during the entire life cycle of PC components. Background Art

[0002] The construction industry is one of the important pillars of current social and economic development, and precast concrete components (PC components) as one of the key elements of the construction industry play an important role in the design and construction of building structures. PC components are widely used in construction projects in various fields such as residential, commercial, and infrastructure due to their many advantages such as high production efficiency, convenient construction, and stable structure. However, PC components will generate a large amount of carbon emissions throughout their life cycle, which will bring certain pressure to the environment. Therefore, it is crucial to accurately assess the carbon emissions of PC components throughout their life cycle and take targeted carbon reduction measures accordingly to promote the sustainable development of the construction industry. At present, the carbon emission accounting methods commonly used in the industry for PC components mainly include two categories: one is a static calculation method based on the material list and corresponding emission factors of PC components, which has the defect of ignoring emissions in stages such as construction and use; the other is a dynamic simulation method based on life cycle assessment (LCA), which requires a large amount of production, transportation, construction and other process data, and the data collection is difficult, the calculation process is complicated, and it is difficult to quickly obtain accurate carbon emission results.

[0003] Although these methods provide certain support for carbon emission accounting of PC components, there are still technical problems such as complex statistical process and low accuracy. Summary of the invention

[0004] In view of this, the present invention provides a method, medium and system for calculating carbon emissions during the entire life cycle of PC components, which can solve the technical problems of the existing methods, such as complex statistical process and low accuracy.

[0005] The present invention is achieved in that:

[0006] A first aspect of the present invention provides a method for calculating carbon emissions during the entire life cycle of a PC component, comprising the following steps:

[0007] S10, obtaining ultrasonic data, vibration transmission data and high-definition images of the entire surface of the PC component;

[0008] S20, after preprocessing the ultrasonic data, a set of sinusoidal waveforms is obtained by Fourier transform, which is recorded as the first set; after preprocessing the vibration transmission data, a set of sinusoidal waveforms is obtained by Fourier transform, which is recorded as the second set; after preprocessing the high-definition image, a set of sinusoidal waveforms representing the high-definition image is obtained by two-dimensional fast Fourier transform, which is recorded as the third set;

[0009] S30, merging the first set, the second set and the third set into a fused set, and extracting features of the fused set to obtain fused features;

[0010] S40, inputting the fusion feature into a pre-trained PC component production carbon emission model, and outputting the carbon emission value of the PC component in the production stage;

[0011] S50, calculating the carbon emission values ​​of the PC component during the transportation stage, construction stage, use stage, and dismantling and recycling stage;

[0012] S60: Calculate and obtain the carbon emission value of the PC component throughout its life cycle by adding the carbon emission value and output it.

[0013] The ultrasonic data is obtained by measuring multiple points on the surface of the PC component using an ultrasonic detector.

[0014] Furthermore, the vibration transmission data is obtained by installing an acceleration sensor on the surface of the PC component and obtaining the data through a knocking test.

[0015] Furthermore, the high-definition image is obtained by photographing the entire surface of the PC component using a high-resolution camera.

[0016] Furthermore, the production carbon emission model adopts a multi-branch neural network structure, including a feature separation module, a macro feature sub-network, a micro feature sub-network, an outlier feature sub-network and a fusion sub-network.

[0017] The training steps of the PC component production carbon emission model specifically include:

[0018] Step 1: Obtain production data during the production process of PC components, including carbon emission values ​​obtained based on production raw materials and production steps, and fusion features obtained by using steps S10-S30;

[0019] Step 2: Establish a training data set, wherein the training input is the fusion feature and the training output is the carbon emission value;

[0020] Step 3: Use the training data set to train the model of the multi-branch neural network structure to obtain a carbon emission model for PC component production.

[0021] Furthermore, the feature separation module is used to separate the input fusion features into macro fusion features, micro fusion features and outlier fusion features.

[0022] The macroscopic fusion characteristics are low-frequency sinusoidal waveform characteristics that reflect the overall properties of the PC component, including:

[0023] The curve characteristics of the lowest 25% frequency in a set of sine waveforms;

[0024] The first 30% of the sine wave characteristics with the largest amplitude;

[0025] The first 20% of the longest wavelengths are sinusoidal waveform characteristics;

[0026] The first 35% of the energy is the sinusoidal waveform characteristic.

[0027] Furthermore, the macro feature sub-network is used to extract the overall features of the PC component, the input is the first set and the second set, the output is the macro feature vector, and the structure is a convolutional neural network plus an attention mechanism;

[0028] The micro-feature sub-network is used to extract PC component detail features, the input is the third set, the output is a micro-feature vector, and the structure is U-Net plus a spatial attention mechanism;

[0029] The outlier feature subnetwork is used to detect abnormal features, the input is a fusion set, the output is an outlier feature vector, and the structure is an autoencoder plus an adversarial training mechanism;

[0030] The fusion subnetwork is used to integrate the features of each subnetwork. The input is the macro feature vector, the micro feature vector and the outlier feature vector. The output is the final fusion feature vector. The structure is a multi-layer perceptron plus a self-attention mechanism.

[0031] Furthermore, the microscopic fusion feature is a medium-frequency sinusoidal waveform feature reflecting the local details of the PC component, including:

[0032] The curve characteristics of the frequency between 25% and 75% in the sine waveform set;

[0033] The amplitude is in the medium range, i.e. 30% to 70% of the sine wave characteristics;

[0034] The wavelength is in the medium range, i.e. 20% to 80% of the sinusoidal waveform characteristics;

[0035] The energy ratio is in the medium range, i.e. 35% to 85%, i.e. the sinusoidal waveform characteristics;

[0036] The outlier fusion features are high-frequency or abnormal sinusoidal waveform features that may indicate abnormalities or special situations, including:

[0037] The curve characteristics of the 25% with the highest frequencies in the set of sine waveforms;

[0038] Sinusoidal waveform characteristics with abnormally small amplitude, that is, below the 5% percentile of the population, or abnormally large amplitude, that is, above the 95% percentile of the population;

[0039] Sinusoidal waveform characteristics with abnormally short wavelength, i.e. below the 10% percentile of the population, or abnormally long wavelength, i.e. above the 90% percentile of the population;

[0040] The energy proportion is abnormally low, that is, below the 15% percentile of the whole, or abnormally high, that is, above the 85% percentile of the whole, which is a sinusoidal waveform characteristic.

[0041] Optionally, isolated sinusoidal waveform features that differ significantly from the overall trend are also included.

[0042] Wherein, the step S10 specifically includes: obtaining ultrasonic data, vibration transmission data and high-definition images of the entire surface of the PC component. First, an ultrasonic detector is used to perform multi-point measurements on the surface of the PC component to obtain ultrasonic data of the PC component, and these ultrasonic data can reflect the overall characteristics of the internal structure of the PC component. Secondly, an acceleration sensor is installed on the surface of the PC component, and vibration transmission data of the PC component is obtained by tapping test. These vibration transmission data can reveal the detailed characteristics of the internal microstructure of the PC component. Finally, a high-resolution camera is used to shoot the entire surface of the PC component to obtain high-definition image data, which contains rich information on the surface morphology and texture of the PC component.

[0043] Wherein, the step S20 specifically includes: preprocessing and Fourier transforming the obtained data. First, the ultrasonic data and the vibration transmission data are preprocessed respectively, including filtering, normalization and other operations to remove noise and improve data quality. Then, the two types of time domain data are converted into the frequency domain by Fourier transform technology, and a series of sinusoidal waveform sets are obtained, which are respectively recorded as the first set and the second set. At the same time, the high-definition image data is preprocessed, such as denoising, adjusting brightness and contrast, etc., and then a two-dimensional fast Fourier transform is used to obtain a sinusoidal waveform set describing the image texture characteristics, which is recorded as the third set. Through these preprocessing and Fourier transform steps, three sinusoidal waveform sets are obtained, which respectively reflect the overall, microscopic and surface characteristics of the PC component.

[0044] Wherein, the step S30 specifically includes: merging the above three sinusoidal waveform sets into a fusion set and extracting its features. First, the first set, the second set and the third set are merged into a fusion set. Then, a feature separation module is designed to extract features of different levels from the fusion set, which are divided into three categories: macroscopic fusion features, microscopic fusion features and outlier fusion features: macroscopic fusion features reflect the overall properties of PC components, including the 25% curves with the lowest frequency, the 30% curves with the largest amplitude, the 20% curves with the longest wavelength and the 35% curves with the highest energy share; microscopic fusion features reflect the local details of PC components, including curves with frequencies between 25% and 75%, curves with amplitudes in the range of 30% to 70%, curves with wavelengths in the range of 20% to 80%, and curves with energy share in the range of 35% to 85%; outlier fusion features may indicate abnormalities or special cases, including the 25% curves with the highest frequency, curves with abnormal amplitudes or wavelengths, and curves with abnormal energy share. Through the above feature separation, a fusion feature vector that fully reflects the characteristics of PC components is obtained.

[0045] Wherein, the step S40 specifically includes: inputting the fusion feature into the pre-trained PC component production carbon emission model to obtain the carbon emission value of the production stage. This production carbon emission model adopts a multi-branch neural network structure, including a macro feature subnetwork, a micro feature subnetwork, an outlier feature subnetwork and a fusion subnetwork. The macro feature subnetwork uses a convolutional neural network plus an attention mechanism to extract the overall features of the PC component, the input is the first set and the second set, and the output is a macro feature vector; the micro feature subnetwork uses U-Net plus a spatial attention mechanism to extract the detailed features of the PC component, the input is the third set, and the output is a micro feature vector; the outlier feature subnetwork uses an autoencoder plus an adversarial training mechanism to detect abnormal features, the input is the fusion set, and the output is an outlier feature vector; the fusion subnetwork uses a multi-layer perceptron plus a self-attention mechanism to integrate the features of each subnetwork, the input is the macro, micro and outlier feature vectors, and the output is the final fusion feature vector. By inputting this fusion feature vector into the production carbon emission model, the carbon emission value of the PC component production stage can be obtained.

[0046] Wherein, the step S50 specifically includes: calculating the carbon emission values ​​of each stage of transportation, construction, use and dismantling and recycling of PC components. The calculation formula of the carbon emission value in the transportation stage is: the carbon emission value in the transportation stage is equal to the weight of the PC component multiplied by the transportation distance, and then multiplied by the carbon emission factor of the transportation mode, for example, truck transportation is 0.1 kg of carbon dioxide / ton·kilometer. The calculation formula of the carbon emission value in the construction stage is: the carbon emission value in the construction stage is equal to the volume or weight of the PC component multiplied by the predetermined carbon emission factor in the construction stage, for example, 0.05 kg of carbon dioxide / kg component. The calculation formula of the carbon emission value in the use stage is: the carbon emission value in the use stage is equal to the weight of the PC component multiplied by the service life, and then multiplied by the annual average maintenance carbon emission factor, for example, 0.01 kg of carbon dioxide / kg component / year. The calculation formula of the carbon emission value in the dismantling and recycling stage is: the carbon emission value in the dismantling and recycling stage is equal to the weight of the PC component multiplied by the comprehensive carbon emission factor of dismantling and recycling, for example, 0.03 kg of carbon dioxide / kg component.

[0047] Wherein, the step S60 specifically includes: adding the carbon emission values ​​of each stage to obtain the carbon emission value of the PC component throughout its life cycle. The specific calculation formula is: the carbon emission value of the PC component throughout its life cycle is equal to the sum of the carbon emission value of the production stage, the carbon emission value of the transportation stage, the carbon emission value of the construction stage, the carbon emission value of the use stage, and the carbon emission value of the dismantling and recycling stage. Finally, the calculation result is output.

[0048] A second aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores program instructions, and when the program instructions are run, they are used to execute the above-mentioned method for calculating carbon emissions during the entire life cycle of a PC component.

[0049] A third aspect of the present invention provides a PC component full life cycle carbon emission accounting system, which includes the above-mentioned computer-readable storage medium.

[0050] Compared with the prior art, the beneficial effects of the PC component life cycle carbon emission accounting method, medium and system provided by the present invention are:

[0051] 1. Strong data comprehensiveness: This method makes full use of multi-source data of PC components, including ultrasonic data, vibration transmission data and high-definition image data. Through signal processing, these data are converted into a set of sinusoidal waveforms that reflect the multi-scale characteristics of PC components, providing comprehensive data support for carbon emission assessment, which is more comprehensive than the existing static material list and dynamic LCA methods.

[0052] 2. High evaluation accuracy: This method constructs a carbon emission prediction model for PC component production based on deep learning, which can make full use of the fused multimodal features to achieve accurate prediction of carbon emissions in the production stage. At the same time, for other life cycle stages such as transportation, construction, use, and dismantling and recycling, this method also proposes a scientific carbon emission calculation formula, which greatly improves the accuracy of the overall evaluation.

[0053] 3. Strong practicability: The data acquisition methods used in this method, such as ultrasonic detectors, acceleration sensors and high-resolution cameras, are commonly used monitoring methods in the construction industry and are easy to promote and apply in engineering practice. At the same time, the calculation process of this method is relatively simple, does not require a large amount of production process data, has low technical requirements for users, and has good practicability.

[0054] 4. Broad analysis depth: In addition to providing the total carbon emissions of PC components throughout their life cycle, this method can also conduct detailed analysis of the characteristics of PC components from three levels: macro, micro, and outlier, providing richer data support for the low-carbon design and optimization of components.

[0055] In general, the carbon emission accounting method for the entire life cycle of PC components proposed in the present invention fully integrates multi-source monitoring data, adopts advanced signal processing and machine learning algorithms, and realizes accurate assessment of carbon emissions at each life cycle stage of PC components, solving the technical problems of complex statistical process and low accuracy in existing methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 A flow chart of the method provided by the present invention. DETAILED DESCRIPTION

[0057] In order to make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0058] like Figure 1 As shown, it is a flow chart of a method for calculating carbon emissions of PC components throughout their life cycle provided by the present invention. The method comprises the following steps:

[0059] S10, obtaining ultrasonic data, vibration transmission data and high-definition images of the entire surface of the PC component;

[0060] S20, after preprocessing the ultrasonic data, a set of sinusoidal waveforms is obtained by Fourier transform, which is recorded as the first set; after preprocessing the high-definition image, a set of sinusoidal waveforms representing the high-definition image is obtained by Fourier transform, which is recorded as the second set; after preprocessing the high-definition image, a two-dimensional fast Fourier transform is used to obtain a set of sinusoidal waveforms representing the high-definition image, which is recorded as the third set;

[0061] S30, merging the first set, the second set and the third set into a fused set, and extracting features of the fused set to obtain fused features;

[0062] S40, inputting the fusion features into a pre-trained PC component production carbon emission model, and outputting the carbon emission value of the PC component production stage;

[0063] S50, calculating the carbon emission values ​​of the PC component transportation stage, construction stage, use stage and dismantling and recycling stage;

[0064] S60. Calculate and obtain the carbon emission value of the PC component throughout its life cycle by adding up the values ​​and output them.

[0065] The specific implementation methods of the above steps are described in detail below:

[0066] The specific implementation method of step S10 is: obtaining ultrasonic data, vibration transmission data and high-definition images of the entire surface of the PC component. First, an ultrasonic detector is used to perform multi-point measurements on the surface of the PC component to obtain ultrasonic data of the PC component. These ultrasonic data can reflect the overall characteristics of the internal structure of the PC component. Secondly, an acceleration sensor is installed on the surface of the PC component, and vibration transmission data of the PC component is obtained by tapping test. These vibration transmission data can reveal the detailed characteristics of the internal microstructure of the PC component. Finally, a high-resolution camera is used to shoot the entire surface of the PC component to obtain high-definition image data. These high-definition image data contain rich information on the surface morphology and texture of the PC component.

[0067] The specific implementation method of step S20 is: preprocessing and Fourier transforming the obtained data. First, the ultrasonic data and the vibration transmission data are preprocessed respectively, including filtering, normalization and other operations to remove noise and improve data quality. Then, the two types of time domain data are converted into frequency domain by Fourier transform technology to obtain a series of sinusoidal waveform sets. These two sinusoidal waveform sets are respectively recorded as the first set and the second set. At the same time, the high-definition image data is preprocessed, such as denoising, adjusting brightness and contrast, etc., and then a two-dimensional fast Fourier transform is used to obtain a sinusoidal waveform set describing the image texture characteristics, which is recorded as the third set. Through these preprocessing and Fourier transform steps, three sinusoidal waveform sets will be obtained, which respectively reflect the overall, microscopic and surface characteristics of the PC component.

[0068] The specific implementation method of step S30 is: merge the above three sinusoidal waveform sets into a fusion set and extract its features. First, merge the first set, the second set and the third set into a fusion set. This fusion set contains multi-scale feature information of PC components. Then, design a feature separation module to extract features of different levels from the fusion set. Specifically, the fusion features are divided into three categories: macro fusion features, micro fusion features and outlier fusion features:

[0069] The macroscopic fusion characteristics reflect the overall properties of the PC components, including the 25% curve with the lowest frequency, the 30% curve with the largest amplitude, the 20% curve with the longest wavelength, and the 35% curve with the highest energy share;

[0070] The microscopic fusion features reflect the local details of PC components, including curves with frequencies ranging from 25% to 75%, curves with amplitudes ranging from 30% to 70%, curves with wavelengths ranging from 20% to 80%, and curves with energy proportions ranging from 35% to 85%;

[0071] Outlier fusion features may indicate abnormalities or special cases, including the 25% curves with the highest frequency, curves with abnormal amplitude or wavelength, and curves with abnormal energy proportion. These threshold reference values ​​are determined based on relevant research and engineering practice experience. Through the above feature separation, a fusion feature vector that fully reflects the characteristics of PC components is obtained.

[0072] The specific implementation of step S40 is: input the fusion features into the pre-trained PC component production carbon emission model to obtain the carbon emission value of the production stage. This production carbon emission model adopts a multi-branch neural network structure, including the following sub-networks:

[0073] Macro feature sub-network: Convolutional neural network (CNN) plus attention mechanism is used to extract the overall features of PC components. The input is the first set and the second set, and the output is the macro feature vector.

[0074] Micro-feature sub-network: U-Net plus spatial attention mechanism is used to extract PC component detail features. The input is the third set and the output is the micro-feature vector.

[0075] Outlier feature sub-network: uses autoencoder plus adversarial training mechanism to detect abnormal features. The input is the fusion set and the output is the outlier feature vector.

[0076] Fusion sub-network: A multi-layer perceptron (MLP) plus a self-attention mechanism is used to integrate the features of each sub-network. The input is the macro, micro and outlier feature vectors, and the output is the final fused feature vector.

[0077] By inputting this fused feature vector into the production carbon emission model, the carbon emission value of the PC component production stage can be obtained. This model is pre-trained based on a large amount of PC component production data.

[0078] The specific implementation of step S50 is: calculate the carbon emission values ​​of each stage of PC component transportation, construction, use and dismantling and recycling. Among them, the carbon emission calculation of the transportation stage is: according to the weight and transportation distance of the PC component, multiply the carbon emission factor of the corresponding transportation mode (such as the carbon emission factor of truck transportation is 0.1kg CO2 / ton-kilometer) to obtain the carbon emission of the transportation stage. If there is no specific emission factor, it can be estimated by multiplying the fuel consumption by the carbon content of the fuel.

[0079] Calculation of carbon emissions during the construction phase: Use the volume or weight of the PC component and multiply it by a pre-determined carbon emission factor for the construction phase (e.g. 0.05kg CO2 / kg component). This factor should include emissions from the use of construction equipment, auxiliary materials, and personnel activities. If there is no ready-made factor, it can be estimated based on the main energy consumption (e.g. electricity, diesel) during the construction process.

[0080] Calculation of carbon emissions in the use phase: Consider the service life of the PC component (e.g. 50 years) and multiply it by the annual maintenance carbon emission factor (e.g. 0.01kg CO2 / kg component / year). This factor should take into account daily maintenance, repair and replacement activities. If there is no specific factor, a rough estimate can be made based on the expected maintenance frequency and intensity.

[0081] Calculation of carbon emissions during the dismantling and recycling phase: Use the weight of the PC component and multiply it by the comprehensive carbon emission factor for dismantling and recycling (e.g. 0.03kg CO2 / kg component). This factor should include emissions from the dismantling, transportation and handling processes, while also taking into account the carbon reduction brought about by material recycling. If there is no ready-made factor, it can be estimated based on the energy consumption of the dismantling equipment and the waste treatment method.

[0082] Specifically:

[0083] The calculation formula for the carbon emission value in the transportation stage is: E trans =W×D×f trans

[0084] Among them, E trans is the carbon emission value in the transportation stage, W is the weight of PC components, D is the transportation distance, and f trans is the carbon emission factor of the mode of transport, for example, 0.1kg CO2 / ton·km for truck transport. If there is no specific emission factor, it can also be estimated based on fuel consumption and carbon content.

[0085] The calculation formula for the carbon emission value during the construction phase is: const =V×fconst

[0086] Among them, E const is the carbon emission value in the construction phase, V is the volume or weight of the PC component, and f const It is a pre-determined carbon emission factor for the construction phase, such as 0.05kg CO2 / kg component. If there is no ready-made factor, it can be estimated based on the main energy consumption of the construction process.

[0087] The calculation formula for the carbon emission value in the use phase is: use =W×L×f use

[0088] Among them, E use is the carbon emission value in the use phase, W is the weight of the PC component, L is the service life, f use is the annual maintenance carbon emission factor, for example 0.01kg CO2 / kg component / year. If there is no specific factor, it can be estimated based on the expected maintenance frequency and intensity.

[0089] The calculation formula for the carbon emission value in the demolition and recycling stage is: E demo =W×f demo

[0090] Among them, E demo is the carbon emission value of the dismantling and recycling stage, W is the weight of the PC component, and f demo The combined carbon emission factor for dismantling and recycling is, for example, 0.03kg CO2 / kg component. If there is no ready-made factor, it can be estimated based on the energy consumption of the dismantling equipment and the waste treatment method.

[0091] The specific implementation of step S60 is: add the carbon emission values ​​of each stage to obtain the carbon emission value of the PC component throughout its life cycle. The calculation formula is:

[0092] E total =E prod +E trans +E const +E use +E demo

[0093] Among them, E total is the carbon emission value of the PC component throughout its life cycle, E prod is the carbon emission value in the production stage (obtained from step S40), E trans is the carbon emission value in the transportation stage, E const is the carbon emission value in the construction phase, E use is the carbon emission value in the use phase, E demois the carbon emission value of the dismantling and recycling stage. By adding these values, we can get the total carbon emission of the PC component throughout its life cycle. Finally, the calculation result is output.

[0094] In general, this PC component life cycle carbon emission accounting method makes full use of multi-source data such as ultrasound, vibration and images, extracts multi-scale features through signal processing techniques such as Fourier transform, uses advanced deep learning models for carbon emission prediction, and combines carbon emission calculations at each life cycle stage to ultimately produce a comprehensive and accurate carbon emission assessment result. This method can provide strong support for environmental impact assessment and low-carbon design of PC components.

[0095] A second aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores program instructions, and when the program instructions are run, they are used to execute the above-mentioned method for calculating carbon emissions during the entire life cycle of a PC component.

[0096] A third aspect of the present invention provides a PC component full life cycle carbon emission accounting system, which includes the above-mentioned computer-readable storage medium.

[0097] Specifically, the principle of the present invention is:

[0098] 1. Data acquisition: This method uses commonly used monitoring methods in the construction industry, such as ultrasonic detectors, acceleration sensors, and high-resolution cameras, to obtain ultrasonic data, vibration transmission data, and high-definition image data of PC components. These data can fully reflect the overall structural characteristics, local detail characteristics, and surface morphology characteristics of PC components.

[0099] 2. Data preprocessing and feature extraction: The acquired data is preprocessed, including filtering, normalization and other operations, and then the time domain data is converted into the frequency domain using Fourier transform technology to obtain a series of sinusoidal waveform sets. These three sinusoidal waveform sets represent the overall, microscopic and surface characteristics of PC components respectively.

[0100] 3. Feature fusion and separation: The three sinusoidal waveform sets mentioned above are merged into a fusion set, and then a feature separation module is designed to divide the fusion features into three categories: macro fusion features, micro fusion features, and outlier fusion features. This multi-scale feature decomposition helps to more comprehensively characterize the structural characteristics of PC components.

[0101] 4. Carbon emission prediction in the production stage: The fusion features are input into the pre-trained PC component production carbon emission prediction model. The model adopts a multi-branch neural network structure, which can make full use of feature information at different scales to achieve accurate prediction of carbon emissions in the production stage.

[0102] 5. Calculation of carbon emissions in other stages: For other life cycle stages of PC components, such as transportation, construction, use, and dismantling and recycling, this method proposes a scientific carbon emission calculation formula, taking into account factors such as component weight, transportation distance, and service life, to ensure the accuracy of the overall assessment.

[0103] 6. Full life cycle carbon emission assessment: By adding up the carbon emission values ​​of each life cycle stage, the total carbon emission of the PC component during its entire life cycle can be obtained. At the same time, this method can also conduct in-depth analysis of the characteristics of PC components from three levels: macro, micro and outlier, providing a basis for low-carbon design and optimization.

[0104] The reason why the solution of the present invention can solve the problems in the prior art is mainly due to the following reasons:

[0105] First, make full use of multi-source data, which not only includes conventional material lists, but also integrates structural health data and surface defect data obtained by advanced monitoring technology. It can more comprehensively reflect the actual situation of PC components during production and use, thereby improving the accuracy of carbon emission assessment.

[0106] Secondly, the use of deep learning technology to build a production carbon emission prediction model can fully explore the inherent connections between multimodal features, and has stronger prediction capabilities than the static emission factor method and complex LCA simulation. At the same time, scientific calculation formulas are also proposed for other life cycle stages to ensure the accuracy of the overall assessment.

[0107] Thirdly, the data acquisition methods used are all commonly used monitoring technologies in the construction industry, which are easy to promote and apply in actual projects. At the same time, the calculation process is relatively simple, does not require a large amount of production process data, and has low technical requirements for users, making it highly practical.

[0108] Finally, this method can conduct in-depth analysis of the characteristics of PC components from three levels: macro, micro, and outlier. It can not only obtain the overall carbon emissions, but also discover abnormal conditions inside the components, providing a richer basis for low-carbon design and optimization.

[0109] In summary, the carbon emission accounting method for the entire life cycle of PC components proposed in the present invention fully integrates multi-source data, adopts advanced signal processing and machine learning techniques, realizes accurate assessment of carbon emissions at each life cycle stage, provides effective support for the green transformation of the construction industry, and has important application value.

[0110] In order to better understand and implement the present invention, a specific embodiment 1 of the present invention is provided below. The steps of this embodiment 1 are specifically described as follows:

[0111] First, in step S10, three types of data of the PC component are acquired: ultrasonic data, vibration transmission data and high-definition image data.

[0112] The acquisition of ultrasound data can be expressed as:

[0113]

[0114] Among them, X u represents the ultrasonic data set obtained by multi-point measurement on the surface of PC components, represents the ultrasonic data of the i-th measurement point. These ultrasonic data can reflect the overall characteristics of the internal structure of the PC component.

[0115] The acquisition of vibration transmission data can be expressed as:

[0116]

[0117] Among them, X v represents the vibration transmission data set obtained by the accelerometer installed on the surface of the PC component, Represents the vibration transmission data of the jth measurement point. These vibration transmission data can reveal the detailed characteristics of the internal microstructure of the PC component.

[0118] The acquisition of high-definition image data can be expressed as:

[0119]

[0120] Among them, X i It represents a collection of high-definition images of the entire surface of a PC component taken with a high-resolution camera. represents the kth high-definition image. These high-definition image data contain rich information on the surface morphology and texture of PC components.

[0121] In step S20, the above data is preprocessed and Fourier transformed. The preprocessing of ultrasonic data and vibration transmission data can be expressed as:

[0122]

[0123]

[0124] Preprocess (·) represents the preprocessing function, including filtering, normalization and other operations, to obtain the processed ultrasound data. and vibration transmission data

[0125] Then, the two types of data are Fourier transformed to obtain a set of sinusoidal waveforms describing the overall and microscopic characteristics of PC components:

[0126]

[0127]

[0128] Where FFT(·) represents the Fourier transform function, F u and F v They are respectively the sets of sinusoidal waveforms obtained by Fourier transform of ultrasonic data and vibration transmission data, recorded as the first set and the second set.

[0129] At the same time, the high-definition image data is preprocessed and two-dimensional fast Fourier transformed to obtain a set of sinusoidal waveforms describing the surface characteristics of PC components:

[0130]

[0131]

[0132] Where 2DFFT(·) represents the two-dimensional fast Fourier transform function, F i It is a set of sinusoidal waveforms obtained by Fourier transform of high-definition image data, recorded as the third set.

[0133] In step S30, the three sinusoidal waveform sets are merged into a fusion set F, and its features are extracted. Specifically, a feature separation module is designed to divide the fusion features into macro fusion features F m , Micro-fusion feature F μ and outlier fusion feature F o Three categories:

[0134] Macro fusion feature F m It reflects the overall properties of the PC component and can be expressed as:

[0135]

[0136] Among them, s m is the number of macro-fusion features, Represents the i-th macroscopic fusion feature, including the 25% curve with the lowest frequency, the 30% curve with the largest amplitude, the 20% curve with the longest wavelength, and the 35% curve with the highest energy proportion.

[0137] Micro-fusion feature F μ It reflects the local details of PC components and can be expressed as:

[0138]

[0139] Among them, s μ is the number of micro-fusion features, Represents the i-th microscopic fusion feature, including curves with frequencies ranging from 25% to 75%, curves with amplitudes ranging from 30% to 70%, curves with wavelengths ranging from 20% to 80%, and curves with energy proportions ranging from 35% to 85%.

[0140] Outlier fusion feature F o May indicate an abnormal or special situation, which can be expressed as:

[0141]

[0142] Among them, s o is the number of outlier fusion features, Represents the fusion features of the i-th outlier, including the 25% curves with the highest frequency, curves with abnormal amplitude or wavelength, and curves with abnormal energy proportion.

[0143] Through the above feature separation, the fusion feature vector F = {F m ,F μ ,F o}.

[0144] In step S40, the fusion feature F is input into the pre-trained PC component production carbon emission model to obtain the carbon emission value E of the production stage. prod This production carbon emission model adopts a multi-branch neural network structure and can be expressed as:

[0145]

[0146] in, Represents the carbon emission model of PC component production, which includes the following sub-networks:

[0147] Macro feature subnetwork

[0148]

[0149] in, represents the macro feature extraction subnetwork, and the input is the first set F u and the second set F v , the output is the macroscopic feature vector F m .

[0150] Micro-feature sub-network

[0151]

[0152] in, represents the micro-feature extraction subnetwork, and the input is the third set F i , the output is the microscopic eigenvector Fμ .

[0153] Outlier Feature Subnetwork

[0154]

[0155] in, Represents the outlier feature detection subnetwork, the input is the fusion set F, and the output is the outlier feature vector F o .

[0156] Fusion sub-network

[0157]

[0158] in, Represents the feature fusion subnetwork, and the input is the macro feature vector F m , microscopic eigenvector F μ and the outlier feature vector F o , the output is the final fusion feature vector F ′ .

[0159] This fused feature vector F ′ Input to the production carbon emissions model The carbon emission value E of the PC component production stage can be obtained prod .

[0160] In step S50, the carbon emission values ​​of each life cycle stage of the PC component are calculated.

[0161] Carbon emission value E in the transportation stage trans The calculation formula is:

[0162] E trans =W×D×f trans

[0163] Where W is the weight of the PC component, D is the transportation distance, and f trans is the carbon emission factor of the mode of transportation, for example, truck transportation is 0.1kg CO2 / ton·km.

[0164] Carbon emission value E during construction phase const The calculation formula is:

[0165] E const =V×f const

[0166] Where V is the volume or weight of the PC component, f const It is a predetermined carbon emission factor for the construction phase, for example, 0.05kg CO2 / kg component.

[0167] Carbon emission value E in the use phase use The calculation formula is:

[0168] E use =W×L×f use

[0169] Where, L is the service life, f use The annual average maintenance carbon emission factor is, for example, 0.01 kg CO2 / kg component / year.

[0170] Carbon emission value E in the demolition and recycling stage demo The calculation formula is:

[0171] E demo =W×f demo

[0172] Among them, f demo The combined carbon emission factor for dismantling and recycling is, for example, 0.03kg CO2 / kg component.

[0173] In step S60, the carbon emission values ​​of each stage are added together to obtain the carbon emission value E of the PC component throughout its life cycle. total :

[0174] E total =E prod +E trans +E const +E use +E demo

[0175] In summary, this PC component full life cycle carbon emission accounting method makes full use of multi-source data, and through advanced signal processing and deep learning models, it achieves accurate prediction of carbon emissions in the production stage, and combines the carbon emission calculations of each life cycle stage to ultimately obtain comprehensive and accurate carbon emission assessment results.

[0176] In order to further better understand and implement the present invention, the following provides Example 2 of a specific application scenario of the present invention: A construction company is carrying out a large-scale commercial complex construction project in a certain city, and the project uses a large number of precast concrete component (PC component) technologies. In order to evaluate the overall carbon emissions of the project and provide a basis for future green construction, the construction company decided to use the PC component full life cycle carbon emission accounting method proposed in the present invention for detailed analysis.

[0177] First, in the production workshop of PC components, the company staff used an ultrasonic detector to measure 10 different positions on the surface of a column component and obtained the ultrasonic data shown in Table 1:

[0178] Table 1 Ultrasound data table

[0179] Measuring point number Ultrasonic data (μs) 1 221.3 2 218.7 3 223.1 4 219.5 5 222.4 6 220.6 7 221.9 8 219.2 9 222.0 10 220.8

[0180] At the same time, three acceleration sensors were installed on the surface of the column component, and the vibration transmission data shown in Table 2 were obtained through the knocking test:

[0181] Table 2 Vibration transmission data table

[0182] Sensor No. <![CDATA[Vibration transmission data (m / s 2 )]]> 1 0.58 2 0.63 3 0.61

[0183] In addition, a high-resolution camera was used to photograph the entire surface of the column component, obtaining 16 high-definition images.

[0184] Next, these data are preprocessed and Fourier transformed:

[0185] After preprocessing the ultrasonic data, Fourier transform was used to obtain a set of sinusoidal waveforms, which is recorded as the first set F u :

[0186] The vibration transmission data is transformed by Fourier transform to obtain a set of sinusoidal waveforms, which is recorded as the second set F v :

[0187] After high-definition image data preprocessing, a two-dimensional fast Fourier transform is used to obtain a set of sinusoidal waveforms, which is recorded as the third set F i :

[0188] Then, the three sinusoidal waveform sets are merged into a fusion set F, and their features are extracted. Specifically, a feature separation module is designed to divide the fusion features into macro fusion features F m , Micro-fusion feature F μ and outlier fusion feature F o Three categories;

[0189] These feature information are aggregated to form a fusion feature vector F = {F m ,F μ ,F o}.

[0190] Next, this fused feature vector is input into the pre-trained PC component production carbon emission prediction model to obtain the carbon emission value E of the column component production stage. prod =127.3kg CO2.

[0191] For the transportation stage, the weight of the column component is 2.5 tons, the transportation distance is 120 kilometers, and the transportation method is truck transportation. According to the carbon emission factor of 0.1 kg CO2 / ton·km, the carbon emission value of the transportation stage is E trans =2.5×120×0.1=30kgCO2.

[0192] The carbon emissions during the construction phase mainly come from the use of construction equipment, consumption of auxiliary materials and personnel activities. Referring to industry experience, the carbon emission factor of 0.05kg CO2 / kg component can be used. The carbon emission value of the column component during the construction phase is E const =2.5×0.05=0.125kg CO2.

[0193] The carbon emissions in the use phase mainly come from daily maintenance, repair and replacement activities. Assuming that the service life of the column component is 50 years and the annual maintenance carbon emission factor is 0.01 kg CO2 / kg component / year, the carbon emission value in the use phase is E use =2.5×50×0.01=1.25kg CO2.

[0194] Finally, for the dismantling and recycling stage, based on the weight of the column component, 2.5 tons, and the comprehensive carbon emission factor, 0.03 kg CO2 / kg component, the carbon emission value E of the dismantling and recycling stage is calculated. demo =2.5×0.03=0.075kg CO2.

[0195] In summary, by adding up the carbon emission values ​​at each life cycle stage, we can get the total carbon emission of the column PC component over its entire life cycle:

[0196] E total =E prod +E trans +E const +E use +E demo =127.3+30+0.125+1.25+0.075=158.75kgCO2

[0197] The characteristics of the column component are analyzed from three levels: macroscopic, microscopic and outlier: Macroscopic characteristics F m This shows that the overall properties of the column component are good and the structure is stable, and there may not be large deformation or damage. Microscopic characteristics F μ It shows that the internal details of the component are stable and there are no obvious abnormal points. o Only a small number of curves with higher frequencies and abnormal energy proportions appear in the figure, indicating that there are some small defects or irregularities on the surface of the component, but they will not affect the overall performance.

[0198] Overall, the PC component of this column generated a total of 158.75kg CO2 in carbon emissions during the entire life cycle of production, transportation, construction, use and disassembly. From the perspective of emissions in each stage, the production stage is the main source of emissions, accounting for about 80%, followed by the transportation stage, accounting for about 19%, and emissions in other stages are relatively low. From the analysis of component characteristics, the overall performance of the column component is good, the local details are relatively stable, and there are only a few surface defects, which will not have a significant impact on carbon emissions.

[0199] This comprehensive and systematic carbon emission assessment result provides an important basis for construction companies to carry out green construction in the future. On the one hand, the company can optimize the production process and choose low-carbon transportation methods to further reduce carbon emissions in the two main emission links of production and transportation; on the other hand, the company can also refer to the characteristic analysis results and focus on the prevention and control of surface defects in the component design and production process to improve the overall performance, thereby achieving a lower carbon construction.

[0200] In general, the carbon emission accounting method for the entire life cycle of PC components proposed in this paper provides strong support for the green development of construction companies. Through advanced technical means such as multi-source data fusion and deep learning modeling, an accurate assessment of the carbon emissions of PC components at each stage is achieved, and the component characteristics are deeply analyzed from multiple dimensions, providing an important decision-making basis for low-carbon design and construction optimization.

[0201] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for calculating carbon emissions during the entire life cycle of PC components. Features , including the following steps: S10, obtaining ultrasonic data, vibration transmission data and high-definition images of the entire surface of the PC component; S20, after preprocessing the ultrasonic data, a set of sinusoidal waveforms is obtained by Fourier transform, which is recorded as the first set; after preprocessing the vibration transmission data, a set of sinusoidal waveforms is obtained by Fourier transform, which is recorded as the second set; after preprocessing the high-definition image, a set of sinusoidal waveforms representing the high-definition image is obtained by two-dimensional fast Fourier transform, which is recorded as the third set; S30, merging the first set, the second set and the third set into a fused set, and extracting features of the fused set to obtain fused features; S40, inputting the fusion feature into a pre-trained PC component production carbon emission model, and outputting the carbon emission value of the PC component in the production stage; S50, calculating the carbon emission values ​​of the PC component during the transportation stage, construction stage, use stage, and dismantling and recycling stage; S60: Calculate and obtain the carbon emission value of the PC component throughout its life cycle by adding the carbon emission value and output it.

2. According to claim 1, a method for calculating carbon emissions during the entire life cycle of a PC component, It is characterized in that The ultrasonic data is obtained by measuring multiple points on the surface of the PC component using an ultrasonic detector.

3. According to claim 2, a method for calculating carbon emissions during the entire life cycle of a PC component, It is characterized in that The vibration transmission data is obtained by installing an acceleration sensor on the surface of the PC component and obtaining the data through a knocking test.

4. According to claim 3, a method for calculating carbon emissions during the entire life cycle of a PC component, It is characterized in that The high-definition image is obtained by using a high-resolution camera to photograph the entire surface of the PC component.

5. According to claim 4, a method for calculating carbon emissions during the entire life cycle of a PC component, It is characterized in that The production carbon emission model adopts a multi-branch neural network structure, including a feature separation module, a macro feature sub-network, a micro feature sub-network, an outlier feature sub-network and a fusion sub-network.

6. A method for calculating carbon emissions during the entire life cycle of a PC component according to claim 5, It is characterized in that The feature separation module is used to divide the input fusion features into macro fusion features, micro fusion features and outlier fusion features. The macroscopic fusion characteristics are low-frequency sinusoidal waveform characteristics that reflect the overall properties of the PC component, including: The curve characteristics of the lowest 25% frequency in a set of sine waveforms; The first 30% of the sine wave characteristics with the largest amplitude; The first 20% of the longest wavelengths are sinusoidal waveform characteristics; The first 35% of the energy is the sinusoidal waveform characteristic.

7. A method for calculating carbon emissions during the entire life cycle of a PC component according to claim 6, It is characterized in that The macro feature sub-network is used to extract the overall features of the PC component, the input is the first set and the second set, the output is the macro feature vector, and the structure is a convolutional neural network plus an attention mechanism; The micro-feature sub-network is used to extract PC component detail features, the input is the third set, the output is a micro-feature vector, and the structure is U-Net plus a spatial attention mechanism; The outlier feature subnetwork is used to detect abnormal features, the input is a fusion set, the output is an outlier feature vector, and the structure is an autoencoder plus an adversarial training mechanism; The fusion subnetwork is used to integrate the features of each subnetwork. The input is the macro feature vector, the micro feature vector and the outlier feature vector. The output is the final fusion feature vector. The structure is a multi-layer perceptron plus a self-attention mechanism.

8. A method for calculating carbon emissions during the entire life cycle of a PC component according to claim 7, It is characterized in that The microscopic fusion features are medium-frequency sinusoidal waveform features that reflect the local details of PC components, including: The curve characteristics of the frequency between 25% and 75% in the sine waveform set; The amplitude is in the medium range, i.e. 30% to 70% of the sine wave characteristics; The wavelength is in the medium range, i.e. 20% to 80% of the sinusoidal waveform characteristics; The energy ratio is in the medium range, i.e. 35% to 85%, i.e. the sinusoidal waveform characteristics; The outlier fusion features are high-frequency or abnormal sinusoidal waveform features that may indicate abnormalities or special situations, including: The curve characteristics of the 25% with the highest frequencies in the set of sine waveforms; Sinusoidal waveform characteristics with abnormally small amplitude, that is, below the 5% percentile of the population, or abnormally large amplitude, that is, above the 95% percentile of the population; Sinusoidal waveform characteristics with abnormally short wavelength, i.e. below the 10% percentile of the population, or abnormally long wavelength, i.e. above the 90% percentile of the population; The energy proportion is abnormally low, that is, below the 15% percentile of the whole, or abnormally high, that is, above the 85% percentile of the whole, which is a sinusoidal waveform characteristic.

9. A computer-readable storage medium, It is characterized in that The computer-readable storage medium stores program instructions, which, when executed, are used to execute a method for calculating carbon emissions throughout the life cycle of a PC component as described in any one of claims 1 to 8.

10. A PC component full life cycle carbon emission accounting system, It is characterized in that A computer-readable storage medium comprising the computer-readable storage medium of claim 9.

Images