Carbon reduction information management system based on chip-level encryption

Abstract

The present application relates to the technical field of carbon reduction information management, and specifically discloses a carbon reduction information management system based on chip-level encryption, which includes: a carbon reduction data acquisition module, a carbon reduction data preprocessing module, a data encryption module, a data transmission module, a data decryption module and a carbon reduction points calculation module. The carbon reduction information management system based on chip-level encryption aims to collect and analyze various indicators related to carbon reduction, such as hydrogen consumption, mileage, cargo load and road conditions, through advanced sensor technology and data processing methods, so as to provide operators with accurate carbon reduction effect evaluation, and support more efficient carbon reduction information management and optimization strategy formulation. In addition, the embedded security chip not only provides powerful encryption functions, but also enhances the overall stability of the system and reduces security risks caused by software vulnerabilities.

Description

Technical Field

[0001] The present application relates to the technical field of carbon reduction information management, and more specifically, to a carbon reduction information management system based on chip-level encryption. Background Art

[0002] As global climate change becomes increasingly severe, reducing greenhouse gas emissions has become a common concern of the international community. In the field of transportation, especially heavy-duty vehicle transportation, carbon emissions are an important source. In order to achieve carbon reduction goals, it has become a trend to replace traditional fossil fuels with clean energy such as hydrogen energy. Compared with traditional fossil fuels, hydrogen fuel cell vehicles can achieve zero tail gas emissions, thereby significantly reducing the impact on the environment. In order to ensure the effective operation of hydrogen vehicles and maximize the carbon reduction benefits they bring, it is necessary to establish an effective information management system to monitor relevant indicators and support optimization adjustments in the decision-making process based on the analysis of these data. However, since the collected information may contain sensitive business operation details, it is crucial to ensure that the data will not be illegally accessed or tampered with during the data transmission process in order to maintain the security of corporate and personal privacy and ensure the reliability and accuracy of decision-making.

[0003] Therefore, a carbon reduction information management system based on chip-level encryption is desired. Summary of the invention

[0004] The present application provides a carbon reduction information management system based on chip-level encryption, which uses advanced sensor technology and data processing methods to collect and analyze various indicators related to carbon reduction, such as hydrogen consumption, mileage, cargo load, and road conditions, so as to provide operators with accurate carbon reduction effect evaluation and support more efficient carbon reduction information management and optimization strategy formulation. In addition, the embedded security chip not only provides powerful encryption functions, but also enhances the overall stability of the system and reduces security risks caused by software vulnerabilities. Therefore, the carbon reduction information management system based on chip-level encryption can effectively monitor and manage the carbon reduction performance of the vehicle and ensure the security and authenticity of all data.

[0005] In a first aspect, a carbon reduction information management system based on chip-level encryption is provided, comprising:

[0006] A carbon reduction data collection module, used to collect carbon reduction related data through vehicle-mounted sensors, wherein the carbon reduction related data includes a time queue of hydrogen consumption, a time queue of mileage, a time queue of cargo load, and a time queue of road conditions;

[0007] A carbon reduction data preprocessing module, used to preprocess the carbon reduction related data to obtain preprocessed carbon reduction related data;

[0008] A data encryption module, used to encrypt the pre-processed carbon reduction related data using an embedded security chip to obtain ciphertext data;

[0009] A data transmission module, used to transmit the ciphertext data to a cloud server;

[0010] A data decryption module, used for decrypting the ciphertext data on the cloud server to obtain the pre-processed carbon reduction related data;

[0011] The carbon reduction credit calculation module is used to input the pre-processed carbon reduction related data into the carbon reduction credit calculation module based on the feedforward neural network model on the cloud server to obtain the carbon reduction credit value.

[0012] The present application provides a carbon reduction information management system based on chip-level encryption, which includes: a carbon reduction data acquisition module, a carbon reduction data preprocessing module, a data encryption module, a data transmission module, a data decryption module and a carbon reduction points calculation module. The carbon reduction information management system based on chip-level encryption aims to collect and analyze various indicators related to carbon reduction, such as hydrogen consumption, mileage, cargo load and road conditions, through advanced sensor technology and data processing methods, so as to provide operators with accurate carbon reduction effect evaluation and support more efficient carbon reduction information management and optimization strategy formulation. In addition, the embedded security chip not only provides powerful encryption functions, but also enhances the overall stability of the system and reduces security risks caused by software vulnerabilities. Therefore, the carbon reduction performance of the vehicle can be effectively monitored and managed through the carbon reduction information management system based on chip-level encryption, as well as the security and authenticity of all data can be guaranteed. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings of the embodiments of the present application are briefly introduced below. Obviously, the drawings described below only relate to some embodiments of the present application, and are not intended to limit the present application.

[0014] Figure 1 This is a schematic block diagram of a carbon reduction information management system based on chip-level encryption according to an embodiment of the present application.

[0015] Figure 2 This is a data flow diagram of a carbon reduction data preprocessing module in a carbon reduction information management system based on chip-level encryption in an embodiment of the present application.

[0016] Figure 3 This is a schematic block diagram of a carbon reduction data preprocessing module in a carbon reduction information management system based on chip-level encryption according to an embodiment of the present application.

[0017] Figure 4This is a schematic block diagram of a fine-grained interaction unit for hydrogen consumption characteristics in a carbon reduction information management system based on chip-level encryption according to an embodiment of the present application. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work also fall within the scope of protection of the present application.

[0019] Based on this, in the technical solution of this application, a carbon reduction information management system based on chip-level encryption is proposed, such as Figure 1 and Figure 2 As shown, the carbon reduction information management system based on chip-level encryption includes: a carbon reduction data acquisition module 10, which is used to collect carbon reduction related data through vehicle-mounted sensors, and the carbon reduction related data includes a time queue of hydrogen consumption, a time queue of mileage, a time queue of cargo load and a time queue of road conditions; a carbon reduction data preprocessing module 20, which is used to preprocess the carbon reduction related data to obtain preprocessed carbon reduction related data; a data encryption module 30, which is used to encrypt the preprocessed carbon reduction related data using an embedded security chip to obtain ciphertext data; a data transmission module 40, which is used to transmit the ciphertext data to a cloud server; a data decryption module 50, which is used to decrypt the ciphertext data on the cloud server to obtain the preprocessed carbon reduction related data; a carbon reduction credit calculation module 60, which is used to input the preprocessed carbon reduction related data into a carbon reduction credit calculation module based on a feedforward neural network model on the cloud server to obtain a carbon reduction credit value. The carbon reduction information management system based on chip-level encryption is designed to collect and analyze various indicators related to carbon reduction, such as hydrogen consumption, mileage, cargo load, and road conditions, through advanced sensor technology and data processing methods, so as to provide operators with accurate carbon reduction effect evaluation and support more efficient carbon reduction information management and optimization strategy formulation. In addition, the embedded security chip not only provides powerful encryption functions, but also enhances the overall stability of the system and reduces security risks caused by software vulnerabilities. Therefore, the carbon reduction information management system based on chip-level encryption can effectively monitor and manage the carbon reduction performance of vehicles and ensure the security and authenticity of all data.

[0020] Exemplarily, in the carbon reduction data collection module 10, carbon reduction related data are collected through vehicle-mounted sensors, and the carbon reduction related data include a time queue of hydrogen consumption, a time queue of mileage, a time queue of cargo load, and a time queue of road conditions. It should be understood that carbon reduction related data (such as a time queue of hydrogen consumption, a time queue of mileage, a time queue of cargo load, and a time queue of road conditions) are the key to evaluating the actual carbon emissions generated during vehicle operation. For example, hydrogen consumption is directly related to the emission level of hydrogen fuel cell vehicles; mileage affects total energy consumption; cargo load can reflect transportation efficiency, and road conditions affect the operating efficiency and energy consumption mode of the vehicle. Therefore, collecting these data is crucial to accurately evaluate the effectiveness of carbon reduction measures. Based on detailed operating parameters, operators can better understand which factors have a positive or negative impact on carbon reduction, and adjust the operating process accordingly to achieve higher emission reduction efficiency. For example, optimize route planning according to different road conditions, reduce unnecessary idling waiting time, etc.

[0021] In one embodiment, the carbon reduction related data is collected through vehicle-mounted sensors, and the carbon reduction related data includes the time queue of hydrogen consumption, the time queue of mileage, the time queue of cargo load and the time queue of road conditions, including: first, the necessary sensors need to be installed on the vehicle, such as flow meters for measuring hydrogen consumption, GPS receivers for recording driving paths and speeds, weight sensors for monitoring cargo weight changes, and cameras or radars for obtaining real-time road condition information. Next, a communication network is established to ensure that all sensors can stably send the collected data to the central processing unit, which involves the use of CAN bus, LIN bus or other dedicated automotive network protocols.

[0022] Exemplarily, in the carbon reduction data preprocessing module 20, the carbon reduction related data is preprocessed to obtain preprocessed carbon reduction related data. It should be understood that in actual application, the data collected by the sensor may be affected by various factors such as sensor failure or external interference and abnormal values ​​may appear. These abnormal values ​​will not only interfere with the normal carbon reduction performance analysis, but also may cause deviations in the final carbon reduction points calculation results and affect the reliability of decision-making. Therefore, it is very necessary to effectively identify and fill in the outliers of the original data to ensure the integrity of the data set. However, traditional data cleaning methods are often difficult to deal with missing or outlier problems in complex scenarios, especially in continuous time series data. How to accurately identify and reasonably fill missing values ​​is particularly important.

[0023] In view of the above technical problems, in the preprocessing step of outlier identification and filling of carbon reduction related data, taking the time queue data of hydrogen consumption as an example, the technical concept of the present application is to identify outliers of the time queue of hydrogen consumption through the missing value identification module, and then introduce data processing and analysis algorithms based on artificial intelligence and deep learning at the back end to perform forward hydrogen consumption time series features and backward hydrogen consumption time series features with the outlier as the identifier and time node, and use the forward and backward fine-grained time series interaction information of hydrogen consumption to perform missing data reasoning and estimation of the outlier, so as to achieve estimated value filling of abnormal missing values. In this way, outlier identification and missing value estimation filling of carbon reduction related data can be automatically performed, thereby realizing intelligent carbon reduction information management, which helps to support more efficient vehicle carbon reduction information management and optimization strategy formulation.

[0024] In one embodiment, Figure 3 As shown, the carbon reduction data preprocessing module 20 includes: a hydrogen consumption extraction unit 21, which is used to extract a time queue of hydrogen consumption from the carbon reduction related data; a hydrogen consumption missing value identification processing unit 22, which is used to input the hydrogen consumption time queue into a missing value identification module to obtain an identification result, and the identification result is used to indicate whether there is a missing value; a hydrogen consumption sequence segmentation unit 23, which is used to respond to the identification result that there is a missing value, and use the missing value as an identifier to perform sequence segmentation on the hydrogen consumption time queue to obtain a hydrogen consumption before missing subsequence and a hydrogen consumption after missing subsequence; a hydrogen consumption sequence encoding unit 24, which is used to sequence the hydrogen consumption before missing subsequence and the hydrogen consumption after missing subsequence, respectively. Column encoding is performed to obtain a forward implicit coding vector of the time series before the hydrogen consumption is missing and a backward implicit coding vector of the time series after the hydrogen consumption is missing; a hydrogen consumption feature fine-grained interaction unit 25 is used to perform feature fine-grained interaction on the forward implicit coding vector of the time series before the hydrogen consumption is missing and the backward implicit coding vector of the time series after the hydrogen consumption is missing to obtain a fine-grained interaction coding feature of the sequence before and after the hydrogen consumption is missing; a hydrogen consumption missing value inference unit 26 is used to perform missing value inference based on the fine-grained interaction coding feature of the sequence before and after the hydrogen consumption is missing to determine an estimated value of the missing value; a hydrogen consumption estimated value filling unit 27 is used to fill the estimated value of the missing value into the time queue of the hydrogen consumption to obtain a pre-processed time queue of the hydrogen consumption.

[0025] Exemplarily, in the hydrogen consumption extraction unit 21, a time queue of hydrogen consumption is extracted from the carbon reduction related data. It should be understood that hydrogen consumption directly reflects the energy efficiency and emission level of hydrogen fuel cell vehicles. By analyzing the change of hydrogen consumption over time, the carbon emissions of the vehicle under different operating conditions can be accurately measured, thereby providing solid data support for evaluating the effectiveness of carbon reduction measures. Understanding the time pattern of hydrogen consumption helps to identify which time periods or operating conditions are most efficient in the use of hydrogen fuel.

[0026] In one embodiment, extracting the time queue of hydrogen consumption from the carbon reduction related data includes: first, it is necessary to decode the received data packet according to the communication protocol (such as MQTT, HTTP, CoAP, etc.) used for data transmission. Different protocols have different formats and encoding rules, so these differences must be correctly identified and processed. For example, in the case of using JSON or XML format, the parser needs to be able to understand the corresponding grammatical structure and extract the part containing the actual measurement value. Once the data packet is successfully decoded, the next step is to determine which fields correspond to the information of hydrogen consumption. This is usually based on a pre-defined data model or schema that specifies the specific meaning of each sensor output. For example, there may be a specific field identified as "hydrogen_flow_rate", which indicates the instantaneous hydrogen flow rate; another field may be "timestamp", which is used to record the time point of each measurement. Through this field mapping, the required hydrogen consumption data can be accurately located. Sometimes, different devices may report the same type of measurement results in different units (such as liters / minute vs. kilograms / hour). In order to ensure consistency, necessary unit conversion operations need to be performed during the parsing process so that all data conform to a unified standard. After extracting the values ​​related to hydrogen consumption from the raw data, the next thing to do is to build an ordered time series. This means sorting all hydrogen consumption measurements by timestamp to form a coherent timeline. This process helps to maintain the temporal continuity of the data, making subsequent analysis more intuitive and easy to understand. If there are multiple records at the same point in time, you need to decide how to deal with this situation. You can choose to keep the last updated data or calculate the average as a representative. In addition, you should consider whether you need to add additional metadata to describe data overlap. In the process of building a time series, you may encounter situations where there is no corresponding hydrogen consumption data for certain time periods. At this time, these gaps should be appropriately marked to provide a basis for subsequent data filling or interpolation.

[0027] Exemplarily, in the hydrogen consumption missing value identification and processing unit 22, the time queue of the hydrogen consumption is input into the missing value identification module to obtain an identification result, and the identification result is used to indicate whether there are missing values. It should be understood that the data set of hydrogen consumption may have missing values ​​due to sensor failure, communication interruption or other unforeseen reasons. If these missing values ​​are not identified and processed, subsequent analysis may be biased, leading to wrong conclusions or decisions. Accurate data is the basis for effective analysis. By identifying and processing missing values, it can be ensured that the data set used to calculate key indicators such as carbon reduction credits is as complete and unbiased as possible, thereby improving the reliability and accuracy of the analysis results. In addition, after identifying the missing values, the missing data can be reasonably estimated and filled through advanced algorithms and techniques (such as machine learning models), so that reliable analysis results can be provided and support optimization strategy formulation even under the condition of incomplete data.

[0028] In one embodiment, the missing value identification module uses the Z-score method, that is, for each hydrogen consumption data point, its standardized score relative to the average value of the entire time series is calculated. A threshold (such as 3) is usually selected. If the Z-score of a data point exceeds this threshold, it is considered to be an outlier, which may be caused by sensor failure or transmission error, indicating that there may be a missing value at this location.

[0029] In another embodiment, the missing value identification module uses the IQR (Interquartile Range) method. First, the first quartile Q1 and the third quartile Q3 of the hydrogen consumption time series are calculated, and then To set the upper and lower bounds, Data points that fall outside this range are considered potential outliers, indicating the presence of missing values.

[0030] Exemplarily, in the hydrogen consumption sequence segmentation unit 23, in response to the identification result that there are missing values, the time queue of hydrogen consumption is segmented into a sequence using the missing values ​​as identifiers to obtain a subsequence before the hydrogen consumption is missing and a subsequence after the hydrogen consumption is missing. It should be understood that in response to the identification result that there are missing values, the time queue of hydrogen consumption is segmented into a sequence using the missing values ​​as identifiers to obtain a subsequence before the hydrogen consumption is missing and a subsequence after the hydrogen consumption is missing. The problem can be localized by dividing the time series into a part before the missing value (the subsequence before the hydrogen consumption is missing) and a part after the missing value (the subsequence after the hydrogen consumption is missing). In this way, without interfering with the entire sequence, we can focus on solving the problem of missing data in a specific time period. In addition, missing values ​​usually do not exist in isolation, and they often have some kind of temporal correlation with the previous and next data points. By segmenting the sequence, the characteristics of the data in the time period before and after the missing value can be better utilized for prediction or filling. For example, if hydrogen consumption shows a certain trend before the missing value, then this trend is likely to continue after the missing value, thereby providing clues for estimating the missing value.

[0031] Exemplarily, in the hydrogen consumption sequence encoding unit 24, the subsequence before the hydrogen consumption is missing and the subsequence after the hydrogen consumption is missing are sequence encoded to obtain a forward implicit encoding vector of the hydrogen consumption before the missing time series and a backward implicit encoding vector of the hydrogen consumption after the missing time series. It should be understood that the subsequence before the hydrogen consumption is missing contains historical information before the missing value. By using a long short-term memory network (LSTM), the historical time series feature information of hydrogen consumption in the past time period can be extracted. This information helps to understand the causes of the current state and provides background support for predicting missing values. Similarly, for the data after the missing value, the present application also applies the LSTM model to obtain the development trend of hydrogen consumption in the future period, that is, the future time series feature information. This not only reflects what may happen next, but also reveals some potential influencing factors or patterns that may indirectly affect the situation before the missing point. At the same time, considering that only relying on a single direction of encoding is not enough to fully describe all relevant information in a time series. Because each data point in the time series may be the result of the combined action of multiple factors in the past and the future. Therefore, by combining the latent state vectors in both the forward and backward directions, a more comprehensive context view can be constructed, thereby improving the accuracy of filling missing values.

[0032] In one embodiment, the hydrogen consumption sequence encoding unit is used to: input the subsequence before the hydrogen consumption is missing into the forward sequence encoder based on the forward LSTM model to obtain the forward implicit coding vector of the hydrogen consumption before the missing time series; input the subsequence after the hydrogen consumption is missing into the backward sequence encoder based on the backward LSTM model to obtain the backward implicit coding vector of the hydrogen consumption after the missing time series. Here, by processing the subsequence before the hydrogen consumption is missing through the forward sequence encoder based on the forward LSTM model, the time series feature information of the hydrogen consumption before the missing value can be extracted, which includes the time series change trend and dynamic pattern of the hydrogen consumption, etc., which helps to capture how the past information in the sequence affects the current state. And by processing the subsequence after the hydrogen consumption is missing through the backward sequence encoder based on the backward LSTM model, the time series feature information of the hydrogen consumption after the missing value and the dynamic change of the hydrogen consumption can be extracted, which helps to capture the impact of future information in the sequence on the current state.

[0033] Exemplarily, in the hydrogen consumption feature fine-grained interaction unit 25, the forward implicit coding vector of the hydrogen consumption before the missing time series and the backward implicit coding vector of the hydrogen consumption after the missing time series are subjected to feature fine-grained interaction to obtain the fine-grained interactive coding features of the hydrogen consumption before and after the missing time series. It should be understood that since the forward implicit coding vector of the hydrogen consumption before the missing time series and the backward implicit coding vector of the hydrogen consumption after the missing time series respectively contain the historical time series feature information about the hydrogen consumption before the missing value and the future time series feature information after the missing value, the time series semantics of the two can provide time series clues about the missing value of hydrogen consumption, which is helpful for the subsequent estimation and reasoning of the missing value. Therefore, in order to be able to construct a complete context view of the missing position, thereby obtaining a richer context representation to improve the accuracy of filling the missing value, it is necessary to combine the interactive information and dynamic relationship between the forward time series features before the missing hydrogen consumption and the backward time series features after the missing hydrogen consumption. Based on this, in order to be able to perform fine-grained interaction on the forward implicit coding vector of the time series before the hydrogen consumption is missing and the backward implicit coding vector of the time series after the hydrogen consumption is missing, thereby providing support for the subsequent estimation and filling of missing hydrogen consumption values, in the technical solution of the present application, the forward implicit coding vector of the time series before the hydrogen consumption is missing and the backward implicit coding vector of the time series after the hydrogen consumption is missing are further subjected to fine-grained feature interaction to obtain fine-grained interactive coding features of the sequence before and after the hydrogen consumption is missing.

[0034] In one embodiment, Figure 4As shown, the hydrogen consumption feature fine-grained interaction unit 25 includes: an interaction optimization analysis subunit 251, which is used to perform interaction optimization analysis on the forward implicit coding vector of the hydrogen consumption before the loss and the backward implicit coding vector of the hydrogen consumption after the loss to obtain an external knowledge optimized hydrogen consumption before-after loss sequence fine-grained feature interaction matrix; a fine-grained modulation optimization coding subunit 252, which is used to optimize the forward implicit coding vector of the hydrogen consumption before the loss and the hydrogen consumption before the loss based on the external knowledge optimized hydrogen consumption before-after loss sequence fine-grained feature interaction matrix. The backward implicit coding vector of the time sequence after the missing hydrogen consumption is fine-grained modulated and optimized to obtain the optimized forward implicit coding vector of the time sequence before the missing hydrogen consumption and the optimized backward implicit coding vector of the time sequence after the missing hydrogen consumption; the semantic association coding subunit 253 is used to perform semantic association coding on the optimized forward implicit coding vector of the time sequence before the missing hydrogen consumption and the optimized backward implicit coding vector of the time sequence after the missing hydrogen consumption to obtain the fine-grained interactive coding feature vector of the sequence before and after the missing hydrogen consumption as the fine-grained interactive coding feature of the sequence before and after the missing hydrogen consumption.

[0035] In one embodiment, the interaction optimization analysis subunit 251 is used to: input the forward implicit coding vector of the hydrogen consumption before the loss and the backward implicit coding vector of the hydrogen consumption after the loss into a fine-grained feature interaction network to obtain a fine-grained feature interaction matrix of the hydrogen consumption before and after the loss; and perform external knowledge attention modulation on the fine-grained feature interaction matrix of the hydrogen consumption before and after the loss to obtain the external knowledge optimized hydrogen consumption before and after the loss fine-grained feature interaction matrix. Specifically, the process can be expressed as:

[0036]

[0037]

[0038] in, and are respectively the forward implicit coding vector of the time sequence before the hydrogen consumption is missing and the backward implicit coding vector of the time sequence after the hydrogen consumption is missing, is matrix multiplication, is the matrix transpose, is the fine-grained feature interaction matrix of the sequence before and after the loss of hydrogen consumption, and denotes the first external knowledge attention learnable memory parameter matrix and the second external knowledge attention learnable memory parameter matrix, represents the normalization function, Optimizing the fine-grained feature interaction matrix of pre- and post-missing sequences of hydrogen consumption for external knowledge.

[0039] That is, in addition to the historical and future data itself, external knowledge such as experience from domain experts and historical records can also be used to optimize the interaction matrix. This additional knowledge can be effectively integrated into the interaction process of forward and backward implicit states through the attention mechanism. For example, in some cases, a certain type of vehicle operating conditions (such as the startup phase and the acceleration phase) may show a consistent hydrogen consumption pattern, and such a regularity can be used as external knowledge to guide the learning of the model.

[0040] In one embodiment, the fine-grained modulation optimization coding subunit 252 is used to: perform a linear transformation on the forward implicit coding vector of the hydrogen consumption before the loss to obtain a forward query feature vector of the hydrogen consumption before the loss and a forward value feature vector of the hydrogen consumption before the loss; use the external knowledge optimized hydrogen consumption before-after sequence fine-grained feature interaction matrix as a key matrix, input the forward query feature vector of the hydrogen consumption before the loss, the forward value feature vector of the hydrogen consumption before the loss and the key matrix into a fine-grained modulation module based on a Transformer structure to obtain the optimized hydrogen consumption The forward implicit coding vector of the time series before the missing; the backward implicit coding vector of the time series after the missing hydrogen consumption is linearly transformed to obtain the backward query feature vector of the time series after the missing hydrogen consumption and the backward value feature vector of the time series after the missing hydrogen consumption; the fine-grained feature interaction matrix of the sequence before and after the missing hydrogen consumption is optimized by the external knowledge as the key matrix, and the backward query feature vector of the time series after the missing hydrogen consumption, the backward value feature vector of the time series after the missing hydrogen consumption and the key matrix are input into the fine-grained modulation module based on the Transformer structure to obtain the optimized backward implicit coding vector of the time series after the missing hydrogen consumption. Specifically, the process can be expressed as follows:

[0041]

[0042]

[0043]

[0044]

[0045]

[0046]

[0047] in, is the temporal forward implicit coding vector before the hydrogen consumption is missing, Optimize the fine-grained feature interaction matrix of the pre- and post-sequences of hydrogen consumption loss for external knowledge, and They represent the weight matrix of the forward query before the hydrogen consumption is missing and the bias vector of the forward query before the hydrogen consumption is missing, and They represent the weight matrix of the forward value of the time series before the loss of hydrogen consumption and the bias vector of the forward value of the time series before the loss of hydrogen consumption, respectively. is matrix multiplication, and are respectively the time series forward query feature vector before the hydrogen consumption is missing and the time series forward value feature vector before the hydrogen consumption is missing, is the matrix transpose, for The scale is the matrix width multiplied by the matrix height, express function, is the optimized hydrogen consumption before loss time series forward implicit coding vector, and They represent the weight matrix of the backward query after the hydrogen consumption is missing and the bias vector of the backward query after the hydrogen consumption is missing, and They represent the weight matrix of the backward value of the time series after the hydrogen consumption is missing and the bias vector of the backward value of the time series after the hydrogen consumption is missing, and are the backward query feature vector of the time series after the hydrogen consumption is missing and the backward value feature vector of the time series after the hydrogen consumption is missing, It is the backward implicit coding vector of the optimized hydrogen consumption after missing time series.

[0048] That is, first, by linearly transforming the implicit encoding vectors before and after the loss of hydrogen consumption, the present application can enhance the expressive power, so that the model can focus more on specific time points or feature dimensions, thereby better capturing the dynamic change patterns in the sequence. This transformation not only increases learning flexibility, but also prepares key inputs for the subsequent attention mechanism, allowing the model to more accurately select important information when calculating the attention score. Next, the present application uses the fine-grained feature interaction matrix of the sequence before and after the loss of hydrogen consumption optimized by external knowledge as the key matrix. This matrix contains important interaction information between the forward and backward implicit states and has been optimized by external knowledge (such as expert experience, historical data, etc.). In this way, the present application incorporates prior knowledge in the field into the model, making it closer to the actual situation, especially when processing data in complex or special scenarios. In addition, this step strengthens the context association, so that the model considers the relationship between the context more when calculating the attention score. For example, there may be different hydrogen consumption patterns in the startup phase and the acceleration phase, and these patterns can be better understood and applied through external knowledge. Then, the application inputs the query feature vector, value feature vector, and key matrix together into a fine-grained modulation module based on the Transformer structure. In this process, the model automatically adjusts the focus according to the importance of each time step, giving more weight to those time points that are critical to filling missing values, and reducing attention to less relevant time points. The final output is the optimized implicit coding vectors, which not only contain all the information of the original sequence, but also integrate supplementary information from forward and backward feature interactions and external knowledge. Such an optimization process helps to improve the accuracy of filling missing values, because the model can now better understand the spatiotemporal environment around the missing location and make more reasonable predictions based on this.

[0049] In one embodiment, the semantic association coding subunit 253 is used to calculate the position point-by-position division between the optimized forward implicit coding vector of the hydrogen consumption before the missing sequence and the optimized backward implicit coding vector of the hydrogen consumption after the missing sequence to obtain the fine-grained interactive coding feature vector of the hydrogen consumption before and after the missing sequence. Specifically, the process can be expressed as follows:

[0050]

[0051] in, is the optimized hydrogen consumption before loss time series forward implicit coding vector, is the backward implicit coding vector of the optimized hydrogen consumption after missing time series, A fine-grained interactive encoding feature vector for the pre- and post-missing sequence of the hydrogen consumption is obtained.

[0052] That is, when processing the operating data of hydrogen fuel cell trucks, this application aims to further improve the accuracy and reliability of missing value estimation through the interaction analysis between the optimized forward implicit coding vector of the hydrogen consumption before missing time series and the optimized backward implicit coding vector of the hydrogen consumption after missing time series. Specifically, this application calculates the position point division between the two optimized implicit coding vectors to obtain the fine-grained interaction coding feature vector of the hydrogen consumption before and after missing time series. The effect of this process is that it can capture and strengthen the small but important relationship between the forward and backward implicit states.

[0053] In summary, the process of fine-grained feature interaction processing can further capture the small but important interactive information and subtle connections between the forward features of the time series before the missing value of hydrogen consumption and the backward features of the time series after the missing value of hydrogen consumption, which are not easily captured by simple linear combinations. This fine-grained interactive information helps to more accurately understand the dynamic relationship between the forward time series semantics of hydrogen consumption before the missing value and the backward time series semantics after the missing value, which is crucial for the subsequent estimation and filling of missing values ​​of hydrogen consumption. It is worth mentioning that in the process of fine-grained feature interaction processing, not only the interaction between the forward features of the time series before the missing value of hydrogen consumption and the backward features of the time series after the missing value of hydrogen consumption is strengthened, but also these interactive semantic information is optimized by integrating external knowledge. External knowledge can come from the experience of domain experts, historical data, etc. This knowledge is effectively integrated into the interaction process between the forward time series features and the backward time series features of hydrogen consumption before and after the missing value through the attention mechanism, which enables better understanding and processing of complex hydrogen consumption missing value filling tasks in different scenarios, helps to provide reliable prediction results even when the data is incomplete, and improves its performance in practical applications. In other words, this enhanced generalization capability enables the model to maintain the accuracy and reliability of missing value filling in different missing value filling scenarios, providing reliable data support for subsequent missing value estimation and optimization decision making of hydrogen consumption.

[0054] Exemplarily, in the missing value reasoning unit 26 of the hydrogen consumption, missing value reasoning is performed based on the fine-grained interactive coding features of the sequence before and after the missing hydrogen consumption to determine the estimated value of the missing value. In one embodiment, the missing value reasoning unit of the hydrogen consumption is used to: input the fine-grained interactive coding vector of the sequence before and after the missing hydrogen consumption into the missing value reasoning module based on the decoder to obtain the estimated value of the missing value. In other words, the decoding regression is performed using the fine-grained interactive semantics between the forward time series features and the backward time series features of the missing values ​​of the hydrogen consumption, so as to infer and estimate the missing value to obtain the estimated value of the missing value. Then, the estimated value of the missing value is filled into the time queue of the hydrogen consumption to obtain the time queue of the preprocessed hydrogen consumption. In this way, outlier identification and missing value estimation filling of carbon reduction related data can be realized in an automated and intelligent manner, thereby realizing intelligent carbon reduction information management, which helps to support more efficient vehicle carbon reduction information management and optimization strategy formulation.

[0055] In a specific embodiment, the fine-grained interactive coding vector of the hydrogen consumption before and after the missing sequence is input into the missing value inference module based on the decoder to obtain the estimated value of the missing value, including: determining the weight matrix of the decoder; based on the weight matrix of the decoder, linearly transforming and mapping the fine-grained interactive coding vector of the hydrogen consumption before and after the missing sequence to obtain the fine-grained decoding vector of the hydrogen consumption before and after the missing sequence; and accumulating and summing all eigenvalues ​​of the fine-grained decoding vector of the hydrogen consumption before and after the missing sequence to obtain the estimated value of the missing value. Specifically, the process can be expressed as follows: ,in, is the fine-grained interaction encoding vector of the hydrogen consumption before and after the loss, is the estimated value of the missing value, is the weight matrix of the decoder, Represents matrix multiplication.

[0056] In another specific embodiment, a decoder based on a long short-term memory network (LSTM) is used to decode the fine-grained interaction encoding vector of the before-after sequence of the missing hydrogen consumption to obtain an estimate of the missing value. When constructing the decoder model, the decoder first receives the fine-grained interaction encoding vector of the before-after sequence of the missing hydrogen consumption as input. These feature vectors contain complex interaction information between forward and backward implicit states, as well as context associations optimized by external knowledge. Next is the core LSTM layer. LSTM networks are good at processing time series data and capturing long-term dependencies, and are very suitable for inferring missing values. The LSTM unit predicts the hydrogen consumption of the next time step based on the input feature vector of the current time step and its internal state. Each LSTM unit has three gating mechanisms (forget gate, input gate, output gate) inside, which work together to determine which information should be remembered or forgotten. In order to improve the estimation accuracy, an attention mechanism can also be added after the LSTM layer. This allows the decoder to selectively focus on the most relevant part of the input feature vector when generating an estimate for each time step. For example, in some cases, features of the startup phase or acceleration phase may be particularly important to fill in missing values, and the attention mechanism can help the decoder focus more on these key features. Finally, there is a fully connected layer that converts the state of the LSTM output into the actual hydrogen consumption estimate. This layer is responsible for mapping the high-dimensional LSTM output to the actual low-dimensional numerical value.

[0057] At the beginning of the decoding process, the initial state of the LSTM is initialized with a fine-grained interaction encoding vector of the pre- and post-missing sequence of hydrogen consumption. This initialization method enables the decoder to carry rich contextual information from the beginning. The decoder then generates estimates of the missing values ​​time by time step. For each time step, the LSTM unit combines the input feature vector of the current time step and the state of the previous time step to predict the hydrogen consumption of the next time step. If the attention mechanism is enabled, the decoder also dynamically adjusts the degree of attention to different features based on the importance of the current time step. After each prediction, the LSTM unit updates its internal state to reflect the results of the latest prediction and provide context for the prediction of the next time step. This dynamic update mechanism enables the decoder to gradually approach the true hydrogen consumption pattern and maintain high accuracy even in the presence of multiple consecutive missing values. After a series of time step predictions, the decoder finally outputs a series of estimates of the missing values. These estimates are filled back into the corresponding positions in the original time series, completing the entire data filling process. In order to ensure that the decoder can accurately estimate the missing values, the present application will fully train the model. During the training process, a large amount of historical data with known true values ​​is used to adjust the model parameters so that the decoder learns to recognize typical hydrogen consumption patterns in different scenarios. In addition, techniques such as cross-validation can be used to evaluate model performance and continuously optimize to improve estimation accuracy.

[0058] Exemplarily, in the hydrogen consumption estimate filling unit 27, the estimated value of the missing value is filled into the time queue of the hydrogen consumption to obtain the time queue of the preprocessed hydrogen consumption. It should be understood that a complete data sequence is crucial for subsequent analysis and modeling. If there are a large number of missing values ​​and they are not processed, the results of the data analysis will be inaccurate, which will in turn affect the quality of the decision. By filling the missing values, the actual operating conditions of the vehicle can be more accurately reflected. For example, when calculating carbon reduction credits or other key indicators, complete and accurate data can provide a more reliable basis to help operators make scientific and reasonable decisions. That is, after obtaining the estimated values ​​of the missing values, the present application fills these estimated values ​​back to the corresponding positions in the original hydrogen consumption time series. Specifically, for each missing time point, the original vacant position is replaced with the estimated value generated by the decoder to ensure that the entire time series remains consistent.

[0059] Exemplarily, in the data encryption module 30, an embedded security chip is used to encrypt the pre-processed carbon reduction related data to obtain ciphertext data. It should be understood that when processing the operating data of hydrogen fuel cell trucks, the present application not only focuses on how to accurately fill missing values ​​and optimize data analysis, but also pays special attention to data security and privacy protection. Therefore, the present application chooses to use an embedded security chip to encrypt the pre-processed carbon reduction related data (such as hydrogen consumption, mileage, cargo load and road conditions, etc.) to obtain ciphertext data. Specifically, when the data is uploaded from the vehicle to the cloud server, there may be a risk of being eavesdropped or tampered with. By using an embedded security chip for encryption, unauthorized third parties can be effectively prevented from accessing sensitive information, ensuring the integrity and confidentiality of the data during transmission. Carbon reduction related data may contain the company's operational details and personal information, which may cause economic losses or reputation damage to the company if leaked. Encryption measures can ensure the security of these sensitive data and safeguard the privacy rights and interests of enterprises and individuals. The embedded security chip not only provides powerful encryption functions, but also enhances the overall stability of the system and reduces security risks caused by software vulnerabilities. Hardware-level encryption is more difficult to crack than pure software implementation, providing an extra layer of security for the entire system.

[0060] In one embodiment, first, this application will select an embedded security chip with efficient encryption and decryption capabilities, support multiple encryption algorithms (such as AES, RSA, etc.) and certification. Such chips usually integrate dedicated cryptographic processors and random number generators, can quickly perform complex encryption operations, and provide a high level of security. Next, the embedded security chip is installed in the vehicle's data acquisition module and integrated into the existing hardware architecture. Ensure that the chip can directly access the pre-processed carbon reduction-related data stream and can encrypt it in real time before the data is sent. According to actual needs, this application will configure appropriate encryption strategies, including selecting a suitable encryption algorithm, setting the key length, and defining a key management mechanism. For example, AES-256 can be selected as the encryption algorithm, and keys can be generated and stored in a secure manner to ensure that only authorized devices and services can decrypt the received data. Once the above preparations are completed, whenever new pre-processed carbon reduction-related data is generated, the embedded security chip will automatically encrypt the data. Specifically, the chip reads the data block to be encrypted, applies the selected encryption algorithm, and uses the pre-configured key to encrypt and convert it, and finally outputs the ciphertext data.

[0061] Exemplarily, in the data transmission module 40, the ciphertext data is transmitted to the cloud server. It should be understood that the transmission of encrypted carbon reduction related data (such as hydrogen consumption, mileage, cargo load and road conditions, etc.) to the cloud server can realize the centralized storage and management of data. This not only facilitates the unified processing and analysis of large-scale data, but also utilizes the powerful computing resources of cloud computing to execute complex algorithms, thereby more efficiently supporting decision-making. By uploading data to the cloud, operators can access the latest vehicle operation information at any time from any location. This is very important for real-time monitoring of vehicle status, timely response to abnormal situations, and optimization of scheduling arrangements. In addition, cloud platforms usually have high availability and fault tolerance, which can ensure the persistence and reliability of data. At the same time, cloud platforms provide convenient conditions for data exchange between different departments or partners. For example, logistics companies can share authorized data fragments with other relevant parties (such as suppliers, customers, etc.) through cloud platforms to promote collaboration and improve overall operational efficiency.

[0062] In one embodiment, the ciphertext data is transmitted to the cloud server, including: first, the present application will select a secure and reliable transmission protocol, such as transport layer security (TLS) or secure socket layer (SSL). These protocols can establish an encrypted connection between the client and the server to prevent man-in-the-middle attacks and other forms of network threats. Next, the corresponding network interface is configured in the vehicle's data acquisition module so that it can connect to the Internet through a mobile network (such as 4G / 5G), Wi-Fi or other suitable means, and communicate with the designated cloud server. Ensure that the interface supports the selected security protocol and can work stably under various network environments. Once the above preparations are completed, whenever new ciphertext data is generated, the embedded security chip will automatically send this data to the cloud server. Specifically, the vehicle end will initiate an HTTPS request or use other secure transmission methods to package the ciphertext data into an appropriate format (such as JSON, XML, etc.) and send it out through an encrypted channel. After arriving at the cloud server, the ciphertext data will be captured by the corresponding receiving module, and the legitimacy of the sender's identity will be confirmed through the identity verification mechanism. Only verified data packets will be accepted and stored in the database. In addition, technologies such as digital signatures or hash checks can be added to verify the authenticity and integrity of the data to ensure that the data has not been tampered with. In order to ensure the security and stability of the entire transmission process, this application will set up a monitoring system in the cloud to track data traffic and transmission status in real time. Any abnormal activity will trigger an alarm and record detailed log information for subsequent audits and troubleshooting. Finally, considering the risk of data loss, this application will also regularly back up the data stored in the cloud and develop a comprehensive disaster recovery plan. In this way, even if an unexpected situation occurs, normal service operations can be quickly restored to ensure business continuity.

[0063] Exemplarily, in the data decryption module 50, the cloud server decrypts the ciphertext data to obtain the pre-processed carbon reduction related data. It should be understood that although encryption ensures the security of the data, in practical applications, in order to perform analysis, modeling and other advanced processing tasks, the ciphertext data must be converted back to the original plaintext form. Only after decryption can the powerful computing resources of cloud computing be fully utilized to execute complex algorithms to support decision making. Many data analysis tools and techniques rely on specific data formats and content. If the ciphertext data is operated directly, the results may be distorted or cannot be correctly parsed. Through decryption, it can be guaranteed that the subsequent analysis process can accurately process the real data values ​​and avoid any interference caused by encryption.

[0064] In one embodiment, in the cloud server, the ciphertext data is decrypted to obtain the pre-processed carbon reduction related data, including: first, configuring a secure and reliable decryption environment on the cloud server. This includes installing the necessary software libraries and tools, and setting strict access control policies to ensure that only authorized users and services can access the decryption function. In addition, it is necessary to ensure that the key management mechanism used in the decryption process is secure and efficient. Since the decryption process depends on the correct key, the secure management of the key is crucial. This application will use a secure way to generate, distribute and store keys, such as using a hardware security module (HSM) or a key management system (KMS) provided by a cloud service provider. These measures can prevent key leakage and ensure that it is always under control. When the ciphertext data arrives at the cloud server, the decryption module will automatically receive and start the decryption process. Specifically, the decryption module will read the ciphertext data packet, extract the encrypted information therein, and decrypt it using the corresponding key and algorithm. Common encryption algorithms such as AES-256 can quickly complete the decryption task through a high-performance CPU or GPU accelerator to ensure the speed and efficiency of data processing. After decryption is completed, this application will also add technologies such as digital signatures or hash checks to verify the authenticity and integrity of the data. By comparing the hash values ​​or other feature codes of the data before and after decryption, it can be confirmed that the data has not been tampered with, further enhancing the security of the system. In order to ensure the security and stability of the entire decryption process, this application will set up a detailed logging system in the cloud to track the time, source, results and other information of each decryption activity in real time. Any abnormal situation will trigger an alarm and be recorded for subsequent auditing and troubleshooting.

[0065] Exemplarily, in the carbon reduction credit calculation module 60, in the cloud server, the pre-processed carbon reduction related data is input into the carbon reduction credit calculation module based on the feedforward neural network model to obtain the carbon reduction credit value. It should be understood that by using the feedforward neural network model to calculate the carbon reduction credit value, the carbon emissions reduced during vehicle operation can be quantified more accurately. This method can comprehensively consider a variety of factors, including but not limited to hydrogen consumption, driving distance, speed changes, and environmental conditions, so as to provide a comprehensive and accurate assessment. Carbon reduction credits, as a key indicator for measuring emission reduction effects, are essential for formulating scientific and reasonable operating strategies. By learning a large amount of historical data, the feedforward neural network model can capture the complex relationship between different variables and predict the future carbon reduction potential based on this. Using the powerful computing resources of cloud computing, large-scale data processing and analysis tasks can be completed in a short time. The feedforward neural network model can run efficiently in the cloud and quickly output carbon reduction credit results, which provides the possibility for real-time monitoring and dynamic adjustment. Compared with traditional statistical methods or simple rule engines, this method not only improves the calculation speed, but also obtains more refined results.

[0066] In one embodiment, in the cloud server, the pre-processed carbon reduction related data is input into the carbon reduction integral calculation module based on the feedforward neural network model to obtain the carbon reduction integral value, including: first, a feedforward neural network model suitable for carbon reduction integral calculation is constructed on the cloud server. The model is usually composed of multiple layers, including an input layer, several hidden layers and an output layer. Each neuron is connected by weights to form a complex nonlinear mapping relationship. Select appropriate activation functions (such as ReLU) and loss functions (such as mean square error), and train with a large amount of historical data so that the model learns to identify the impact of different variables on the carbon reduction effect. Next, the pre-processed and decrypted carbon reduction related data is organized into a form suitable for inputting into the neural network. This may involve operations such as standardization and normalization to ensure that all features are in the same numerical range and avoid some features dominating the learning process of the model due to excessive or too small values. In addition, the time queue of hydrogen consumption, the time queue of mileage, the time queue of cargo load and the time queue of road conditions are used as input features. Necessary computing resources, such as high-performance CPU / GPU clusters, are configured on the cloud server to support large-scale parallel computing of the feedforward neural network model. At the same time, install and configure deep learning frameworks (such as TensorFlow, PyTorch, etc.) to facilitate the deployment and calling of the model. Ensure that the entire computing environment has good stability and scalability to cope with data processing needs of different scales. Once the above preparations are completed, whenever new pre-processed carbon reduction-related data arrives at the cloud server, the carbon reduction credit calculation process will be automatically triggered. Specifically, the data will be sent to the input layer of the feedforward neural network model, and after being transmitted and transformed layer by layer, the corresponding carbon reduction credit value will be generated at the output layer. This process can be automated through API interfaces or other forms to ensure that each calculation can respond and return results in a timely manner.

[0067] In another embodiment of the present application, a carbon reduction information management system based on chip-level encryption is also provided, which includes: an Android mobile phone end (driver end), a mine site WEB end (enterprise end), and a platform management backend (government end), wherein, the Android mobile phone end (driver end): the driver obtains carbon reduction points by using hydrogen energy vehicles to work, and the hydrogen consumption, mileage, cargo load, road conditions and other data when the vehicle is used are integrated, and the carbon reduction points are obtained by calculating through methodology, and the carbon reduction points generated by the driver in actual work are updated in real time on the driver side APP. Hydrogen energy vehicles are used as data sources, and the vehicle's mileage, hydrogen consumption, cargo load, road conditions and other data are combined with location information and time, and uploaded in real time through the vehicle terminal acquisition device, and all data are processed on the chain through chip-level encryption security chips to ensure the security and authenticity of all data. Mine site WEB end (enterprise end): As an enterprise entity, the government has specific indicators for carbon reduction points for mine sites every year. Mine sites obtain carbon reduction points through the number of minerals transported by the driver end each time. The driver end obtains the opportunity to pull the mine by transferring its own carbon reduction points, thereby realizing the circulation of carbon reduction points. Platform management backend (government side): driver-side management, public account management, acquisition of carbon reduction credits, circulation and transaction of carbon reduction credits, statistics of carbon reduction data, and formulation of carbon reduction transactions will all be reflected on the government side.

[0068] In summary, according to the embodiment of the present application, the carbon reduction information management system based on chip-level encryption is explained, which collects and analyzes various indicators related to carbon reduction, such as hydrogen consumption, mileage, cargo load, and road conditions, through advanced sensor technology and data processing methods, so as to provide operators with accurate carbon reduction effect evaluation, and support more efficient carbon reduction information management and optimization strategy formulation. In addition, the embedded security chip not only provides powerful encryption functions, but also enhances the overall stability of the system and reduces security risks caused by software vulnerabilities. Therefore, the carbon reduction information management system based on chip-level encryption can effectively monitor and manage the carbon reduction performance of the vehicle and ensure the security and authenticity of all data.

[0069] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0070] It should be understood that the specific examples in this article are only intended to help those skilled in the art better understand the embodiments of the present application, rather than to limit the scope of the embodiments of the present application.

[0071] It should also be understood that in the various embodiments of the present application, the size of the serial number of each process does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0072] It should also be understood that the various implementation methods described in this specification can be implemented individually or in combination, and the embodiments of the present application are not limited to this.

[0073] Unless otherwise stated, all technical and scientific terms used in the embodiments of the present application are the same as the meanings generally understood by those skilled in the art of the technical field of the present application. The terms used in the present application are only for the purpose of describing specific embodiments and are not intended to limit the scope of the present application. The term "and / or" used in the present application includes any and all combinations of one or more related listed items. The singular forms of "a kind of", "above" and "the" used in the embodiments of the present application and the appended claims are also intended to include majority forms, unless the context clearly indicates other meanings. In addition, the terms "first", "second" etc. are only used for description purposes and cannot be understood as indicating or suggesting relative importance.

[0074] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0075] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0076] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0077] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0078] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A carbon reduction information management system based on chip-level encryption, characterized in that: include: A carbon reduction data collection module, used to collect carbon reduction related data through vehicle-mounted sensors, wherein the carbon reduction related data includes a time queue of hydrogen consumption, a time queue of mileage, a time queue of cargo load, and a time queue of road conditions; A carbon reduction data preprocessing module, used to preprocess the carbon reduction related data to obtain preprocessed carbon reduction related data; A data encryption module, used to encrypt the pre-processed carbon reduction related data using an embedded security chip to obtain ciphertext data; A data transmission module, used to transmit the ciphertext data to a cloud server; A data decryption module, used for decrypting the ciphertext data on the cloud server to obtain the pre-processed carbon reduction related data; A carbon reduction credit calculation module, used to input the pre-processed carbon reduction related data into the carbon reduction credit calculation module based on the feedforward neural network model on the cloud server to obtain a carbon reduction credit value; Wherein, the carbon reduction data preprocessing module includes: a hydrogen consumption extraction unit, configured to extract a time queue of hydrogen consumption from the carbon reduction related data; A hydrogen consumption missing value identification processing unit, used for inputting the hydrogen consumption time queue into a missing value identification module to obtain an identification result, wherein the identification result is used to indicate whether there is a missing value; A hydrogen consumption sequence segmentation unit, configured to, in response to the identification result that there is a missing value, perform sequence segmentation on the hydrogen consumption time queue using the missing value as an identifier to obtain a subsequence before the hydrogen consumption is missing and a subsequence after the hydrogen consumption is missing; A hydrogen consumption sequence encoding unit, used for performing sequence encoding on the subsequence before the hydrogen consumption is missing and the subsequence after the hydrogen consumption is missing, respectively, to obtain a forward implicit encoding vector of the hydrogen consumption before the hydrogen consumption is missing and a backward implicit encoding vector of the hydrogen consumption after the hydrogen consumption is missing; A hydrogen consumption feature fine-grained interaction unit, used for performing feature fine-grained interaction on the forward implicit coding vector of the hydrogen consumption before the loss and the backward implicit coding vector of the hydrogen consumption after the loss to obtain a fine-grained interactive coding feature of the sequence before and after the loss of hydrogen consumption; A missing value inference unit for hydrogen consumption, used for performing missing value inference based on fine-grained interactive coding features of a sequence before and after the hydrogen consumption is missing to determine an estimated value of the missing value; The hydrogen consumption estimated value filling unit is used to fill the estimated value of the missing value into the time queue of the hydrogen consumption to obtain the preprocessed time queue of the hydrogen consumption.

2. The carbon reduction information management system based on chip-level encryption according to claim 1 is characterized in that: The hydrogen consumption sequence encoding unit is used to: Input the subsequence before the hydrogen consumption is missing into a forward sequence encoder based on a forward LSTM model to obtain a forward implicit coding vector of the time sequence before the hydrogen consumption is missing; The subsequence after the hydrogen consumption is missing is input into the backward sequence encoder based on the backward LSTM model to obtain the backward implicit coding vector of the time series after the hydrogen consumption is missing.

3. The carbon reduction information management system based on chip-level encryption according to claim 2 is characterized in that: The hydrogen consumption feature fine-grained interaction unit includes: An interactive optimization analysis subunit is used to perform interactive optimization analysis on the forward implicit coding vector of the hydrogen consumption before the loss and the backward implicit coding vector of the hydrogen consumption after the loss to obtain an external knowledge optimized hydrogen consumption before-after loss sequence fine-grained feature interaction matrix; A fine-grained modulation optimization coding subunit is used to optimize the fine-grained feature interaction matrix of the sequence before and after the hydrogen consumption is missing based on the external knowledge, and perform fine-grained modulation optimization coding on the forward implicit coding vector of the time sequence before the hydrogen consumption is missing and the backward implicit coding vector of the time sequence after the hydrogen consumption is missing, respectively, to obtain the optimized forward implicit coding vector of the time sequence before the hydrogen consumption is missing and the optimized backward implicit coding vector of the time sequence after the hydrogen consumption is missing; The semantic association coding subunit is used to perform semantic association coding on the optimized forward implicit coding vector of the hydrogen consumption before the loss and the optimized backward implicit coding vector of the hydrogen consumption after the loss to obtain a fine-grained interactive coding feature vector of the sequence before and after the loss of hydrogen consumption as the fine-grained interactive coding feature of the sequence before and after the loss of hydrogen consumption.

4. The carbon reduction information management system based on chip-level encryption according to claim 3 is characterized in that: The interactive optimization analysis subunit is used to: Inputting the forward implicit coding vector of the hydrogen consumption before the loss and the backward implicit coding vector of the hydrogen consumption after the loss into a fine-grained feature interaction network to obtain a fine-grained feature interaction matrix of the hydrogen consumption before and after the loss; The fine-grained feature interaction matrix of the sequence before and after the hydrogen consumption loss is modulated by external knowledge attention to obtain the external knowledge optimized fine-grained feature interaction matrix of the sequence before and after the hydrogen consumption loss.

5. The carbon reduction information management system based on chip-level encryption according to claim 4 is characterized in that: The fine-grained modulation optimization coding subunit is used to: Performing a linear transformation on the forward implicit coding vector of the hydrogen consumption before missing to obtain a forward query feature vector of the hydrogen consumption before missing and a forward value feature vector of the hydrogen consumption before missing; The external knowledge-optimized hydrogen consumption before-after missing sequence fine-grained feature interaction matrix is ​​used as a key matrix, and the forward query feature vector of the hydrogen consumption before missing sequence, the forward value feature vector of the hydrogen consumption before missing sequence and the key matrix are input into a fine-grained modulation module based on a Transformer structure to obtain the optimized forward implicit coding vector of the hydrogen consumption before missing sequence; Performing a linear transformation on the backward implicit coding vector of the time series after the hydrogen consumption is missing to obtain a backward query feature vector of the time series after the hydrogen consumption is missing and a backward value feature vector of the time series after the hydrogen consumption is missing; The external knowledge-optimized fine-grained feature interaction matrix of the sequence before and after the hydrogen consumption is missing is used as a key matrix, and the backward query feature vector of the time series after the hydrogen consumption is missing, the backward value feature vector of the time series after the hydrogen consumption is missing and the key matrix are input into the fine-grained modulation module based on the Transformer structure to obtain the optimized backward implicit coding vector of the time series after the hydrogen consumption is missing.

6. The carbon reduction information management system based on chip-level encryption according to claim 5 is characterized in that: The semantic association coding subunit is used to calculate the position point-by-position point division between the optimized forward implicit coding vector of the hydrogen consumption before the loss and the optimized backward implicit coding vector of the hydrogen consumption after the loss to obtain the fine-grained interactive coding feature vector of the sequence before and after the loss of hydrogen consumption.

7. The carbon reduction information management system based on chip-level encryption according to claim 6 is characterized in that: The hydrogen consumption missing value inference unit is used to: input the fine-grained interactive coding vector of the hydrogen consumption before and after missing sequence into the decoder-based missing value inference module to obtain the estimated value of the missing value.

8. The carbon reduction information management system based on chip-level encryption according to claim 7 is characterized in that: The hydrogen consumption missing value inference unit is used to: determine the weight matrix of the decoder; based on the weight matrix of the decoder, linearly transform and map the fine-grained interactive coding vector of the hydrogen consumption before and after missing sequence to obtain the fine-grained decoding vector of the hydrogen consumption before and after missing sequence; and accumulate and sum all eigenvalues ​​of the fine-grained decoding vector of the hydrogen consumption before and after missing sequence to obtain the estimated value of the missing value.

9. The carbon reduction information management system based on chip-level encryption according to claim 7, characterized in that: The hydrogen consumption missing value inference unit is used to: use a decoder based on a long short-term memory network to decode the fine-grained interactive coding vector of the hydrogen consumption before and after the missing sequence to obtain an estimated value of the missing value.

Images