Carbon emission calculation system
The carbon emission calculation system uses wearable devices and a non-autoregressive Transformer model to predict transportation modes and calculate emissions, addressing the challenge of inconsistent data collection and ensuring compliance with carbon emission standards.
Patent Information
- Application Number
- TW115201872
- Authority / Receiving Office
- TW · TW
- Patent Type
- Utility models
- Current Assignee / Owner
- Filing Date
- 2026-03-04
- Publication Date
- 2026-07-11
- Estimated Expiration
- 2036-03-03
AI Technical Summary
Existing systems struggle to accurately and consistently collect carbon emission data from employee commutes due to unpredictable changes in transportation modes caused by weather and traffic conditions, making it difficult for companies to comply with carbon emission standards.
A carbon emission calculation system utilizing wearable devices with sensing units and machine learning models to identify transportation modes by analyzing environmental data, employing a non-autoregressive Transformer model to predict probabilities and calculate emissions based on carbon emission coefficients.
Accurately calculates carbon emissions from employee commutes, enabling companies to meet ISO certification requirements by providing precise sustainability information.
Smart Images
Figure IMG-2_DRAW_115201872-A0305-14-0001-1 
Figure IMG-2_DRAW_04_A0101_DRAWINGS_1
Abstract
Description
Carbon emission calculation system Technical Field
[0001] This invention relates to a carbon emission calculation system, and more particularly to a carbon emission calculation system capable of identifying the means of transportation used by a user during commuting and calculating its carbon emissions. Prior Technology
[0002] In response to the net-zero emissions trend, Taiwan enacted the Greenhouse Gas Reduction and Management Act in 2015 to assist companies in calculating their carbon footprint through various auditing methods to reduce carbon emissions year by year and comply with relevant carbon emission standards. Currently, different ISO environmental sustainability standards have varying requirements for the sustainability information that companies must disclose, with carbon emissions being a key audit item for measuring a company's net-zero carbon emissions. However, while internal company activities can be systematically audited using various data, employee commutes or business trips may involve temporary changes in transportation due to weather and traffic conditions, making it difficult for companies to consistently and accurately collect carbon emission data during employee movement. Summary of the Invention
[0003] Therefore, the purpose of this invention is to provide a carbon emission calculation system that uses wearable devices to detect environmental data and machine learning models to identify the means of transportation used by employees, thereby accurately calculating the carbon emissions generated when employees move.
[0004] Therefore, this novel carbon emission calculation system can communicate with a wearable device containing multiple sensing units and includes a communication unit, a storage unit and a processing unit.
[0005] This communication unit can establish a communication channel with the wearable device.
[0006] This storage unit stores multiple carbon emission coefficients corresponding to multiple modes of transportation.
[0007] The processing unit is electrically connected to the communication unit and the storage unit to communicate with the wearable device via the communication unit, and includes an artificial intelligence computing model.
[0008] When the processing unit receives multiple sensing data generated by the sensing units included in the wearable device, the processing unit uses feature extraction technology to obtain a movement distance and a sensing data from the sensing data, and inputs the sensing data and the movement distance into the artificial intelligence computing model, so that the artificial intelligence computing model predicts a probability value associated with each of the vehicles based on the sensing data and the movement distance.
[0009] The processing unit retrieves the carbon emission coefficient corresponding to the vehicle associated with the highest probability value from the storage unit, calculates a carbon emission amount based on the travel distance and the carbon emission coefficient, and stores the sensing data, the travel distance, and the carbon emission amount in the storage unit.
[0010] In some embodiments, the processing unit further includes a data processing module. The processing unit inputs the sensing data into the data processing module, which then performs data cleaning and standardization on the sensing data. Using feature extraction technology, the processing unit extracts the distance traveled and sensing data related to time, speed, acceleration, angular velocity, vibration, audio, and air pressure from the sensing data. The sensing data and the distance traveled are then input into the artificial intelligence computing model.
[0011] In some embodiments, the storage unit also stores a carbon emission calculation formula. The processing unit calculates the carbon emission amount based on the carbon emission calculation formula. The carbon emission calculation formula is: distance traveled * carbon emission coefficient. The processing unit then transmits the vehicle, the distance traveled, and the carbon emission amount to the wearable device.
[0012] In some implementations, the artificial intelligence computing model includes a non-autoregressive transformer pre-trained by the processing unit based on complex sensing data and complex movement distances. Each piece of sensing data relates to time, velocity, acceleration, angular velocity, vibration, audio, and air pressure.
[0013] The advantages of this invention are as follows: When the processing unit receives the sensing data generated by the sensing units included in the wearable device, the processing unit obtains the travel distance and the sensing data from the sensing data using feature extraction technology, and predicts the probability value associated with each vehicle based on the sensing data and the travel distance. It then retrieves the carbon emission coefficient corresponding to the vehicle associated with the highest probability value from the storage unit, and calculates the carbon emission amount based on the travel distance and the carbon emission coefficient. In this way, the carbon emission amount generated by the user's movement can be accurately calculated, enabling enterprises to disclose the sustainability information required by ISO certification documents to comply with relevant carbon emission standards. Simple Explanation of the Diagram
[0014] Other features and effects of this invention will be clearly presented in the embodiments with reference to the drawings, wherein: Figure 1 is a block diagram illustrating an embodiment of the novel carbon emission calculation system. Implementation
[0015] Before this invention is described in detail, it should be noted that similar elements are represented by the same reference numerals in the following description.
[0016] Referring to Figure 1, one embodiment of the novel carbon emission calculation system 100 is shown, which can communicate with a wearable device 200 comprising multiple sensing units 300, and includes a communication unit 1, a storage unit 2, and a processing unit 3. In this embodiment, the carbon emission calculation system 100 is, for example, a server set up by a company consisting of one or more computer devices; the wearable device 200 is, for example, a watch or a bracelet, and has a display screen, and can be worn by a user; the sensing units 300 are, for example, but not limited to, sensing devices such as Global Positioning System (GPS), gyroscope, microphone, barometer, accelerometer, etc., that can be used to sense values of the surrounding environment.
[0017] In detail, the Global Positioning System (GPS) is used to sense the user's distance and path of movement; the gyroscope and accelerometer are used to sense the user's rotation and speed of movement to determine the mode of transportation the user is using. For example, when the speed is 80-120 km / hr, it indicates that the user may be taking a bus or car; when the speed is below 10 km / hr, it indicates that the user may be walking or cycling. Furthermore, the acceleration value can be used to further distinguish the mode of transportation the user is using. For example, when the acceleration value changes more frequently, it indicates that the user may be taking a subway with more frequent starts and stops, rather than a car with more stable acceleration. The microphone is used to sense the sound waves of the surrounding environment. When the user is in a subway station, the microphone will sense low-frequency rumbling sounds; when the user is driving, the microphone will sense the fixed frequency of the car engine. The barometer is used to sense changes in air pressure. When the air pressure value changes rapidly in a short period of time (i.e., high-speed vertical movement), it indicates that the user may be taking an elevator.
[0018] The communication unit 1 is, for example but not limited to, a network card, network controller, or other network device that supports wireless network technologies (such as Wi-Fi, LTE-M, LTE Advanced, or 5G modules) and is used to provide networking functionality and establish a communication channel with the wearable device 200.
[0019] The storage unit 2 is, for example, but not limited to, a computer-readable recording medium, which stores a carbon emission calculation formula and multiple carbon emission coefficients 21 corresponding to various modes of transportation (e.g., MRT, buses, cars). The carbon emission calculation formula is: distance traveled * carbon emission coefficient. In this embodiment, these carbon emission coefficients 21 can be queried on the publicly available website Climatiq's Data Explorer page. For example, the global average carbon emission coefficient for MRT is 0.04 kg CO2e / km per person, while the global average carbon emission coefficient for buses is 0.08 kg CO2e / km per person. Furthermore, when selecting the carbon emission coefficient for one mode of transportation, different carbon emission coefficients can be selected based on the region (e.g., Asia, America) and calculated accordingly, but this is not a limitation.
[0020] The processing unit 3 is, for example, but not limited to, a central processing unit, a microprocessor, or a single-chip microcontroller. It is electrically connected to the communication unit 1 and the storage unit 2 to communicate with the wearable device 200 via the communication unit 1, and includes an artificial intelligence computing model 31 composed of software programs and a data processing module 32.
[0021] Therefore, when the user wears the wearable device 200 while commuting or going out, the sensing units 300 of the wearable device 200 continuously sense the surrounding environment and continuously generate and transmit (through the communication unit 1) multiple sensing data to the processing unit 3. When the processing unit 3 continuously receives the sensing data generated by the sensing units 300 included in the wearable device 200, the processing unit 3 inputs the sensing data into the data processing module 32. In this embodiment, the sensing data is positioning data generated by the Global Positioning System, angular velocity data generated by the gyroscope, sound wave data generated by the microphone, air pressure data generated by the barometer, and acceleration data generated by the accelerometer, but is not limited to these.
[0022] Next, the data processing module 32 cleans and standardizes the sensing data, and uses feature extraction technology to obtain the user's movement distance and sensing data related to time, speed, acceleration, angular velocity, vibration, audio, and air pressure from the sensing data, and inputs the sensing data and the movement distance into the artificial intelligence computing model 31.
[0023] In detail, data cleaning involves removing or replacing outliers and missing values in the sensing data, and then aligning them according to time. For example, when the temperature exceeds a reasonable range (e.g., 50°C), the abnormal temperature needs to be deleted. Furthermore, since each sensing data point represents a different physical meaning, the numerical range of the data is also inconsistent. Therefore, the MinMaxScaler technique is used to scale each sensing data point to a value between 0 and 1.
[0024] Then, the artificial intelligence computing model 31 predicts a probability value associated with each of the vehicles based on the sensing data and the distance traveled.
[0025] In this embodiment, the artificial intelligence computing model 31 includes a Transformer model, such as a non-autoregressive Transformer, which is pre-trained by the processing unit 3 based on complex sensing data and complex movement distances. Furthermore, compared to a traditional autoregressive Transformer, which can only output the probability value related to one mode of transportation at a time, the non-autoregressive Transformer has the advantage of efficiently outputting the probability values related to multiple modes of transportation simultaneously.
[0026] In detail, the core technology of the Transformer model is the self-attention mechanism, which can simultaneously focus on the order of all input data to generate output. Compared with neural network models (such as RNN and LSTM) that need to process the input data sequence item by item and are difficult to parallelize, it is more efficient. In this embodiment, the non-autoregressive transformer (Transformer model) uses positional encoding to enable the model to determine the numerical sequence of the sensed data. The cross attention in the Transformer model connects the encoder and decoder in a single layer, that is, the decoder uses the last layer of the encoder as input. The encoder and decoder calculate the mean and standard deviation of the same layer in the sensed data through layer normalization, and add the sensed data to the output generated by the self-attention mechanism through the residual neural network, and extract deeper features through feedforward. Finally, the probability value associated with the sensed data and the travel distance with each mode of transportation is obtained through the softmax function. The sum of these probability values is 1.
[0027] Then, the processing unit 3 retrieves the carbon emission coefficient 21 corresponding to the vehicle with the highest probability value from the storage unit 2 based on the probability value output by the artificial intelligence computing model 31 for each vehicle, and calculates a carbon emission amount based on the travel distance and the carbon emission coefficient 21 using the carbon emission calculation formula stored in the storage unit 2. The processing unit 3 then stores the sensing data, the travel distance and the carbon emission amount in the storage unit 2, and transmits the vehicle, the travel distance and the carbon emission amount to the wearable device 200.
[0028] For example, if a user's travel distance this morning is 15 kilometers, and the mode of transportation associated with the highest probability value is the MRT, the processing unit 3 retrieves the corresponding MRT carbon emission coefficient 21 "0.04 kg CO2e / km" from the storage unit 2. Based on the carbon emission calculation formula 22, it multiplies the travel distance by the carbon emission coefficient 21 to calculate the user's carbon emissions this morning as 0.6 kg CO2e. The user then sends the mode of transportation, travel distance, and carbon emissions to the wearable device 200, allowing the user to view the mode of transportation used this morning and the carbon emissions generated. The user can also provide feedback on whether the mode of transportation associated with the highest probability value output by the AI computing model 31 matches the user's actual mode of transportation, assisting in model adjustments. The company can also reward users based on their carbon emissions when they choose to reduce carbon emissions by taking public transportation, cycling, climbing stairs, or walking, thereby encouraging employees to conserve energy and reduce carbon emissions.
[0029] In summary, when the processing unit 3 of this novel carbon emission calculation system 100 receives sensing data generated by the sensing units 300 included in the wearable device 200, the processing unit 3 uses feature extraction technology to obtain the travel distance and sensing data from the sensing data, and predicts the probability value associated with each mode of transportation based on the sensing data and the travel distance. It then retrieves the carbon emission coefficient 21 corresponding to the mode of transportation associated with the highest probability value from the storage unit 2, and calculates the carbon emission amount based on the travel distance and the carbon emission coefficient 21. This accurately calculates the carbon emissions generated by the user's movement, enabling companies to disclose the sustainability information required by ISO certification documents to comply with relevant carbon emission standards. Therefore, this novel system effectively achieves its purpose and efficacy.
[0030] However, the above description is merely an embodiment of this invention and should not be construed as limiting the scope of this invention. Any simple equivalent changes and modifications made in accordance with the scope of the patent application and the contents of the patent specification shall still fall within the scope of this invention.
[0031] 100: Carbon Emission Calculation System 200: Wearable devices 300: Sensing Unit 1: Communication Unit 2: Storage Unit 21: Carbon Emission Coefficient 22: Carbon emission calculation formula 3: Processing Unit 31: Artificial Intelligence Computational Model 32: Data Processing Module
Claims
1. A carbon emission calculation system capable of communicating with a wearable device including multiple sensing units, and comprising: a communication unit capable of establishing a communication channel with the wearable device; a storage unit storing multiple carbon emission coefficients corresponding to multiple modes of transportation; and a processing unit electrically connected to the communication unit and the storage unit for communication with the wearable device via the communication unit, and including an artificial intelligence computing model; wherein, When the processing unit receives multiple sensing data from the sensing units included in the wearable device, the processing unit uses feature extraction technology to obtain a travel distance and a sensing data from the sensing data, and inputs the sensing data and the travel distance into the artificial intelligence computing model, so that the artificial intelligence computing model predicts a probability value associated with each vehicle based on the sensing data and the travel distance; the processing unit retrieves the carbon emission coefficient corresponding to the vehicle associated with the highest probability value from the storage unit, calculates a carbon emission amount based on the travel distance and the carbon emission coefficient, and stores the sensing data, the travel distance and the carbon emission amount in the storage unit.
2. The carbon emission calculation system as described in claim 1, wherein, The processing unit also includes a data processing module, and the processing unit inputs the sensing data into the data processing module, so that the data processing module performs data cleaning and data standardization on the sensing data, and obtains the moving distance and the sensing data related to time, speed, acceleration, angular velocity, vibration, audio and air pressure from the sensing data through feature extraction technology, and inputs the sensing data and the moving distance into the artificial intelligence computing model.
3. The carbon emission calculation system as described in claim 1, wherein, The storage unit also stores a carbon emission calculation formula, and the processing unit calculates the carbon emission amount according to the carbon emission calculation formula, whereby the carbon emission calculation formula = distance traveled * carbon emission coefficient, and transmits the vehicle, the distance traveled, and the carbon emission amount to the wearable device.
4. The carbon emission calculation system as described in claim 1, wherein, The artificial intelligence computing model includes a non-autoregressive transformer that has been pre-trained by the processing unit based on complex sensing data and complex movement distances, each of which is related to time, velocity, acceleration, angular velocity, vibration, audio, and air pressure.