Power carbon emission near real-time data acquisition method based on cloud edge end cooperation

Abstract

The application discloses a kind of power carbon emission near real-time data acquisition method based on cloud edge end cooperation, comprising: terminal layer-carbon emission source data acquisition;Deployment terminal acquisition node, power key node deployment intelligent electric meter;The carbon emission raw data of original data that intelligent electric meter is collected is encrypted using secondary Sine chaos mapping;According to the network connection topology of intelligent electric meter, the minimum hop number of each node to other nodes is calculated, and a hop matrix is established;Set dynamic adjustment acquisition frequency;Intelligent electric meter minimization delay overhead is obtained;Edge layer-frequency regulation, hierarchical transmission;Determine optimal data packet size, spectrum sensing times;Data priority sorting;Cloud layer-edge load perception;Verify data legality, decrypt intelligent electric meter data, reverse restore the interference value of secondary Sine chaos mapping, and execute inverse operation on encrypted data;Real-time monitoring is realized to widely distributed power equipment, to ensure that data can be quickly and accurately collected and transmitted.

Description

Technical Field

[0001] This invention relates to the field of electricity carbon emission data acquisition technology, specifically to a near real-time data acquisition method for electricity carbon emissions based on cloud-edge-device collaboration. Background Technology

[0002] Under the dual carbon targets, the dynamic changes in carbon emissions from the power sector pose significant pressure for emission reduction. Traditional carbon emission monitoring methods largely rely on periodic manual data collection, which suffers from issues such as time lag, poor accuracy, and weak protection of carbon data privacy and security. These methods cannot meet the current needs for refined carbon emission management, exhibit resource misallocation, and fail to adapt to the dynamically changing load characteristics during carbon data collection, resulting in both resource waste and a decline in service quality.

[0003] Cloud-edge-device technology provides a reliable technical framework for near real-time data acquisition of electricity carbon emissions. Cloud computing enables centralized management and in-depth analysis of massive amounts of electricity carbon emission data; edge computing reduces data transmission latency and improves processing efficiency; and terminal devices collect raw data from all aspects of the power system in real time. Through cloud-edge-device collaboration, real-time monitoring of widely distributed power equipment can be achieved, ensuring rapid and accurate data collection and transmission. In this invention, the smart meter serves as the terminal data acquisition device. Based on the "Q / GDW 376.1-2019 Technical Specification for Smart Meters," it can achieve real-time collection and transmission of data such as electricity consumption and energy loss, providing fundamental support for obtaining raw carbon emission data (such as power generation, energy consumption, and losses). It can not only directly and accurately record electricity consumption but also communicate bidirectionally with the power company, realizing real-time collection, transmission, and management of electricity information. Summary of the Invention

[0004] The purpose of this invention is to address the problems in existing carbon emission monitoring methods, which mostly rely on manual periodic data collection. These methods suffer from time lag, poor accuracy, and weak protection of carbon data privacy and security. They cannot meet the current needs for refined carbon emission management, exhibit resource misallocation, and fail to adapt to the dynamically changing load characteristics during carbon data collection, resulting in both resource waste and a decline in service quality.

[0005] To address the aforementioned problems, this invention provides a near real-time data acquisition method for electricity carbon emissions based on cloud-edge-device collaboration, comprising:

[0006] Step 1: Terminal Layer - Carbon Emission Source Data Acquisition;

[0007] Step 1-1: Deploy terminal data acquisition nodes and smart meters i at key power nodes such as power plants and user meter boxes;

[0008] Step 1-2: Encrypt the raw carbon emission data collected by smart meter i using a quadratic Sine chaotic mapping;

[0009] Steps 1-3: Based on the network connection topology of smart meter i, calculate the minimum number of hops from each node to other nodes and establish the hop count matrix M;

[0010] Steps 1-4: Set the dynamic adjustment of the sampling frequency;

[0011] Steps 1-5: Obtain the minimum latency overhead of the smart meter;

[0012] Step 2: Edge layer - frequency regulation and hierarchical transmission;

[0013] Step 2-1: Determine the optimal data packet size, the number of spectrum sensing operations, and the time for each spectrum sensing operation;

[0014] Step 2-2: Prioritize and sort the data;

[0015] Step 3: Cloud Layer - Edge Load Awareness;

[0016] Step 3-1: Verify the legality of the data, decrypt the data of smart meter i, reverse the interference value of the quadratic Sine chaotic mapping, and perform the inverse operation on the encrypted data.

[0017] In the preferred approach, data collection at the terminal layer—carbon emission source—includes:

[0018] Step 1-1: Deploy terminal data acquisition nodes and deploy smart meters at key power nodes such as power plants and user meter boxes;

[0019] Step 1-2: Encrypt the raw carbon emission data collected by smart meter i using a quadratic Sine chaotic mapping, with the following formula:

[0020]

[0021] In the formula, the state variable of smart meter i is set as follows: , The initial value is used as a marker for the iteration count. ; Let i be the power of smart meter i at a certain moment. For the corresponding upper and lower power limits, perform a second-order Sine mapping iteration, as shown in the formula:

[0022]

[0023] In the formula, Given the control parameters of smart meter i, calculate the interference value of smart meter i in the first iteration using the following formula:

[0024]

[0025] Raw data related to carbon emissions of smart meters The formula after encryption is:

[0026]

[0027] In the formula, The encrypted carbon emission data for smart meters. This is the scaling factor;

[0028] Steps 1-3: Based on the network connection topology of smart meter i, calculate the minimum hop count from each node to other nodes and establish a hop count matrix M; to filter candidate data collection terminals, define a vector L to store the maximum hop count of each row in M, using the following formula:

[0029]

[0030] In the formula, M is the hop count matrix from each node to other nodes, L is a variable storing the maximum hop count in each row of the hop count matrix M, and suitable candidate points for data collection terminals are initially screened by comparing the values ​​in L, and n is the number of smart meters; if there are several points in L with the same minimum value, the point with the minimum average hop count is used as the final selection value as the data collection terminal, and the formula is:

[0031] Terminal location = argmin(L) (6)

[0032] Steps 1-4: Dynamically adjust the data collection frequency. Utilize a high-frequency data collection mechanism to dynamically adjust the data collection frequency based on the importance of the carbon emission sources. Define the data collection frequency selection variable using the following formula:

[0033] The sum of the selected variables for each sampling frequency is ≤1(7)

[0034] Using a dynamic programming algorithm, the optimal data packet size and the number of spectrum sensing operations are determined. The probability that a smart meter node has data to transmit is: The formula is:

[0035]

[0036] In the formula, is the base of the natural logarithm. The average arrival rate of smart meter i. For the data packet transmission time of the smart meter, The number of spectrum sensing operations for smart meter i;

[0037] Steps 1-5: Obtain the minimum delay overhead of the smart meter, using the following formula:

[0038]

[0039] In the formula, The latency overhead of data packets for smart meters; This represents the transmission delay time of the data packets in the smart meter.

[0040] In the preferred approach, edge layer frequency modulation and hierarchical transmission include:

[0041] Step 2-1: Define the optimization objective function f as the ratio of the delay overhead of the data packet of smart meter i to the average effective transmission time. Determine the optimal data packet size by maximizing the value of f, using the following formula:

[0042]

[0043] In the formula, f is the optimization objective function. The average effective transmission time of smart meter i;

[0044] The number of spectrum sensing operations, the size of the data packets, and the time of each spectrum sensing operation satisfy the constraints of equations (11) and (12). Under the premise of satisfying the maximum tolerable latency of the terminal, the optimal number of spectrum sensing operations is calculated using the following formula:

[0045]

[0046] In the formula, The data transmission rate on the i-channel of the smart meter. This represents the total data volume of the smart meter's i-task. The time for spectrum sensing of smart meter i. The maximum tolerable latency for the smart meter i-terminal;

[0047] Step 2-2: Data priority sorting, based on the Weighted Average Queue (WFQ) algorithm, prioritizes data by binding weights to virtual completion times, and enqueues data according to queue categories. The formula is:

[0048]

[0049] In the formula, The classification and queuing determination flag is 1, which means that the kth carbon data packet in the i-th queue is selected for priority queuing and forwarding, and 0 means that the packet belongs to other packets (others), and the packet does not meet the queuing priority and enters the waiting queue. The virtual completion time is the time for the k-th carbon data group in the i-th queue. The edge layer is divided into carbon data queues based on data service attributes; m is the number of carbon data packets to be processed in the i-th queue.

[0050] Steps 2-3: Initialize virtual time in the edge layer When the j-th event occurs, the virtual time is updated using the following formula:

[0051]

[0052] In the formula, The time when the event occurred. Time interval Internal active queue set, For time intervals, This represents a two-dimensional matching relationship; when queue i has data transmission, its virtual receiving rate is given by equation (15), and its actual allocation rate is given by equation (16). The higher weight queue has a higher bandwidth allocation ratio, as shown in the formula:

[0053]

[0054] In the formula, Let i be the virtual receiving rate. The actual allocation rate of queue i This is the sum of the weights of all queues in the active queue set. Available bandwidth for the edge layer;

[0055] Group the k-th carbon data in queue i, complete the calculation formula (17) for the virtual start time, solve the time coefficient formula (18) for queue i, and finally calculate the virtual completion time formula (19). The formula is as follows:

[0056]

[0057] In the formula, This is the virtual start time for the k-th carbon data group in queue i; Let i be the virtual completion time of the group preceding queue i; The system virtual time when the packet arrives; B is the queue time coefficient; B is the reference bandwidth. Let be the virtual completion time of the k-th carbon data group in queue i; This is the virtual start time for the k-th carbon data group in queue i; Define the carbon data packet length; Steps 2-4: Encrypt data forwarding, receive the final encrypted data and chaos control parameters transmitted by the terminal layer, and upload them to the cloud layer.

[0058] In the preferred approach, cloud-edge load awareness includes:

[0059] Step 3-1: Verify the legality of the data. Verify the legality of the data source using the binary hypothesis testing mechanism. The formula is:

[0060]

[0061] In the formula, The reference key for recording the chaotic control parameters a of smart meter i in the cloud. The reference key for recording the chaotic control parameter b of smart meter i in the cloud is: unmatched data means the data is not encrypted or comes from an illegal node, and matched data means the data is encrypted and matches a legal node. If the chaotic control parameters corresponding to the two are the same, the encryption value matches and the data is determined to come from a legal node. If the chaotic control parameters corresponding to the two are different, the encryption value does not match and the data is determined to come from an illegal node, and the data is discarded directly.

[0062] Step 3-2: Decrypt the data of smart meter i, reverse the reconstruction of the interference value of the quadratic Sine chaotic mapping, and perform the inverse operation on the encrypted data to calculate the initial value of smart meter i. The formula is:

[0063]

[0064] In the formula, Let i be the power of smart meter i at a certain moment. The upper and lower limits are defined as power limits; a quadratic Sine mapping iteration is performed on smart meter i, with the formula as follows:

[0065]

[0066] In the formula, Let be the chaotic control parameters of smart meter i; calculate the disturbance value of smart meter i using the following formula:

[0067]

[0068] The formula for decrypting data from smart meter i is:

[0069]

[0070] In the formula, Carbon emission data after decryption of smart meter i; The encrypted carbon emission data for smart meters. This is the scaling factor;

[0071] Step 3-3: Build a load awareness mechanism in the cloud layer;

[0072] The formula for constructing a cloud-based task state vector is as follows:

[0073]

[0074] In the formula, This indicates whether the task has been scheduled; 1 represents that it has been scheduled, and 0 represents that it has not been scheduled. This represents the task load requirement at time t; The deadline for the task; The security level of the task corresponds to different encryption and decryption complexities: Level 3 is highly sensitive, Level 2 is moderately sensitive, and Level 1 is low sensitive.

[0075] The CPU core operates in two modes: bound cores and shared cores. Bound cores are used for decrypting highly sensitive data, while shared cores are used for general data. The CPU core state vector is extended to adapt to carbon data processing requirements. The formula is as follows:

[0076]

[0077] In the formula, Displays the CPU core's operating mode; Let be the utilization rate of kernel j at time t; The operating frequency of core j; The temperature of nucleus j; The average latency of the currently processed task in kernel j;

[0078] Steps 3-4: Real-time load assessment is achieved through a state-action-excitation loop;

[0079] The state input is the task load vector. With system state vector The combination of these factors results in an action output that, based on the load level determination, corresponds to different resource allocation strategies. The incentive function formula is as follows:

[0080]

[0081] In the formula, For average nuclear utilization rate; The average latency of the task; This represents the average memory usage.

[0082] When initializing the CPU resource table and performing core binding operations, system environment modeling needs to consider the current core type. High-sensitivity tasks are bound to marked cores, while ordinary tasks are bound to shared cores. When receiving encrypted data uploaded from the edge, the security level of the task should be considered. Determine whether core binding is needed. When the value is 3, forced core binding is performed; the optimal core binding is selected based on the Q-Learning algorithm to minimize the number of bound cores and reduce power consumption. The excitation function formula is:

[0083]

[0084] In the formula, Bind the core percentage at time t-1. The percentage of kernels after binding at time t;

[0085] Steps 3-5: Dynamic resource allocation. While meeting the near real-time carbon data processing requirements, the cloud layer improves CPU resource utilization, reduces system power consumption, reduces frequency in low-load scenarios, increases frequency in high-load scenarios to ensure carbon data processing latency meets standards, and balances performance and power consumption in medium-load scenarios.

[0086] The beneficial effects of this invention are as follows: It constructs a three-tiered collaborative architecture of cloud, edge, and terminal, where the terminal layer determines the data collection frequency based on collection requirements, meeting near real-time monitoring needs. It employs a data privacy protection mechanism where the terminal layer encrypts the data and the cloud layer decrypts it, enhancing the security of collected carbon data and ensuring the privacy of electricity carbon emission data. It adopts a load-aware resource scheduling strategy, where the edge layer constructs a sorting and scheduling mechanism based on data priority, and the cloud layer performs resource integration operations, achieving an optimal balance between resource utilization and energy consumption, and optimizing collection frequency and data priority. Attached Figure Description

[0087] Figure 1 This is a diagram of a carbon emission data collection architecture based on cloud-edge-device collaboration;

[0088] Figure 2 It is a multi-hop wireless network data acquisition structure;

[0089] Figure 3 This is a flowchart of the encryption process for a 2D-ECs chaotic system;

[0090] Figure 4 This is a framework diagram of the cloud-based load awareness mechanism. Detailed Implementation

[0091] like Figure 1 As shown, a near real-time data acquisition method for electricity carbon emissions based on cloud-edge-device collaboration includes:

[0092] Step 1: As Figure 2 As shown, data collection at the terminal layer - carbon emission sources;

[0093] Step 1-1: Deploy terminal data acquisition nodes. Smart meters are deployed at key power nodes such as power plants (e.g., the No. 1 steam turbine generator plant), 220kV substations (for line losses), and user meter boxes (e.g., the meter box in Unit 1 of Building 3 in a residential community). As terminal data acquisition devices, smart meters, in accordance with the "Q / GDW 376.1-2019 Technical Specification for Smart Meters," can realize real-time acquisition and transmission of data such as energy consumption and energy loss. This provides fundamental support for obtaining raw carbon emission data (such as power generation, energy consumption, and losses). They can not only directly and accurately record energy consumption but also communicate bidirectionally with the power company, enabling real-time acquisition, transmission, and management of electricity information.

[0094] Steps 1-2: Construct a multi-hop wireless network spanning tree, establishing the structure based on the distance L between all smart meters, such as... Figure 2 As shown, it includes a data acquisition terminal and several smart meters. The data acquisition terminal acts as an edge-side control processor, responsible for processing the transmitted data.

[0095] Step 1-2-1: Based on the network connection topology of the smart meter, calculate the minimum hop count from each node to other nodes and establish a hop count matrix M. To filter candidate data collection terminals, define a vector L to store the maximum hop count of each row in M, using the following formula:

[0096]

[0097] In the formula, M is the hop count matrix from each node to other nodes, used to measure the transmission distance relationship between nodes; L is a variable that stores the maximum hop count of each row in the hop count matrix M. Suitable candidate collection terminals are initially screened by comparing the values ​​in L; and n is the number of smart meters.

[0098] Step 1-2-2: If L has several identical minimum values, then the minimum average hop count point is used as the final selected value for the data acquisition terminal. The formula is:

[0099] Terminal location = argmin(L)

[0100] Steps 1-3: Dynamically adjust the data acquisition frequency. For each time slot and terminal in the power grid business data acquisition process, only one frequency can be selected. Utilizing a high-frequency acquisition mechanism, the acquisition frequency is dynamically adjusted based on the importance of carbon emission sources. Define the acquisition frequency selection variable using the following formula:

[0101] The sum of the selected variables for each sampling frequency is ≤1

[0102] By employing a dynamic programming algorithm, the optimal data packet size and the number of spectrum sensing operations are determined. Within a given time period Dm, transmission overhead is minimized, reducing interference and latency in multi-hop transmission. This ensures a high data acquisition scheduling success rate while minimizing the distance from all smart meters to the acquisition terminal, achieving local data aggregation and reducing single-point transmission pressure. The probability that a smart meter node has data to transmit is... The formula is:

[0103]

[0104] In the formula, e is the base of the natural logarithm. The average arrival rate of smart meter i. For the data packet transmission time of the smart meter i, Let i be the number of spectrum sensing operations for smart meter i; Steps 1-4: Optimize transmission parameters. Through data transmission model optimization, combining the probability that the smart meter node needs to transmit data and the transmission delay time of the data packet, obtain the minimum delay overhead of the smart meter, thereby improving the efficiency of carbon emission source data collection. The formula is:

[0105]

[0106] In the formula, The latency overhead of data packets for smart meters; This represents the transmission delay time of the data packets in the smart meter.

[0107] Steps 1-5: Perform quadratic Sine chaotic encryption. To improve the complexity of traditional chaotic systems, a quadratic Sine chaotic mapping is used to encrypt the raw data related to the carbon emissions of smart meter i. The state variables of smart meter i are set as follows: initial value The formula is:

[0108]

[0109] In the formula, Let i be the power of smart meter i at a certain moment. The upper and lower limits are defined as power limits; a quadratic Sine mapping iteration is performed on smart meter i, with the formula as follows:

[0110]

[0111] In the formula, Given the control parameters of smart meter i, calculate the interference value of smart meter i in the first iteration using the following formula:

[0112]

[0113] Raw data related to carbon emissions of smart meters The formula after encryption is:

[0114]

[0115] In the formula, The encrypted carbon emission data for smart meters. This is the scaling factor;

[0116] Step 2: Edge layer - frequency regulation and hierarchical transmission;

[0117] Step 2-1: Optimize data packets and spectrum sensing parameters. Edge nodes are deployed near substations, regional power dispatch center edge equipment rooms, etc. The edge gateway receives data from the acquisition terminal. To minimize transmission latency, the edge layer optimizes the data packet size and the number of spectrum sensing operations through a dynamic programming algorithm.

[0118] Under channel bandwidth constraints, to balance data packet length and retransmission probability, the optimization objective function f is defined as the ratio of the delay overhead of smart meter i's data packets to the average effective transmission time. The optimal data packet size is determined by maximizing the value of f, as shown in the formula:

[0119]

[0120] In the formula, f is the optimization objective function. Let i be the average effective transmission time of smart meter i; the number of spectrum sensing operations, the size of the data packets, and the time of each spectrum sensing operation satisfy the constraints of the following formula. Under the premise of satisfying the maximum tolerable delay of the terminal, calculate the optimal number of spectrum sensing operations to reduce the channel idle waiting time. The formula is:

[0121]

[0122] In the formula, The data transmission rate on the i-channel of the smart meter. This represents the total data volume of the smart meter's i-task. The time for spectrum sensing of smart meter i. The maximum tolerable latency for the smart meter i-terminal;

[0123] Step 2-2: Data Prioritization. The edge layer divides the carbon data into queues based on the business attributes of the carbon emission data. For example, with n=3 (control queue i=1, monitoring queue i=2, and statistics queue i=3), after receiving encrypted data from the terminal, the data identifier is parsed. Based on the Weighted Average Queuing (WFQ) algorithm, weights are bound to virtual completion times. Higher-weighted queues (such as control data) have shorter virtual completion times for grouping. Priority is then assigned, and data is enqueued according to queue category to ensure priority transmission of critical data. The formula is:

[0124]

[0125] In the formula, The classification and queuing determination flag is 1, which means that the kth carbon data packet in the i-th queue is selected for priority queuing and forwarding, and 0 means that the packet belongs to other packets (others), and the packet does not meet the queuing priority and enters the waiting queue. The virtual completion time is the time for the k-th carbon data group in the i-th queue. The edge layer is divided into carbon data queues based on data service attributes; m is the number of carbon data packets to be processed in the i-th queue.

[0126] Steps 2-3: Initialize virtual time in the edge layer When the j-th event (data packet arrival or forwarding completion) occurs, the virtual time is updated using the following formula:

[0127]

[0128] In the formula, The time when the event occurred. Time interval Internal active queue set, For time intervals, This represents a two-dimensional matching relationship. When queue i has data transmission, its virtual receiving rate and actual allocation rate are calculated using the following formulas. Higher-weight queues are allocated a higher proportion of bandwidth, ensuring rapid forwarding of high-priority data.

[0129]

[0130] In the formula, Let i be the virtual receiving rate. The actual allocation rate of queue i This is the sum of the weights of all queues in the active queue set. Given the available bandwidth for the edge layer; group the k-th carbon data in queue i, calculate the virtual start time, solve for the time coefficient of queue i, and finally calculate the virtual completion time using the following formula:

[0131]

[0132] In the formula, This is the virtual start time for the k-th carbon data group in queue i; Let i be the virtual completion time of the group preceding queue i; The system virtual time when the packet arrives; B is the queue time coefficient; B is the reference bandwidth. Let be the virtual completion time of the k-th carbon data group in queue i; This is the virtual start time for the k-th carbon data group in queue i; The carbon data packet length is defined in steps 2-4: Encrypted data forwarding. The receiving terminal layer transmits the final encrypted data and chaotic control parameters, and uploads them to the cloud layer. Step 3: Cloud layer - edge load awareness; Step 3-1: Verifying data legitimacy. After receiving the chaotic encrypted data uploaded by the edge layer, the cloud layer introduces a binary hypothesis testing mechanism to verify the legitimacy of the data source. If the chaotic control parameters corresponding to both are consistent, it indicates that the terminal is a registered node within the system and the data source is trustworthy; if they are inconsistent, it indicates that the data has been tampered with or the source is illegal. The formula is:

[0133]

[0134] In the formula, The reference key for recording the chaotic control parameters a of smart meter i in the cloud. The reference key for recording the chaotic control parameter b of smart meter i in the cloud is defined as follows: unmatched data indicates unencrypted data or data originating from an illegal node; matched data indicates encrypted data matching a legitimate node. If the chaotic control parameters corresponding to both are consistent, the encrypted values ​​match, and the data is determined to originate from a legitimate node. If the chaotic control parameters corresponding to both are inconsistent, the encrypted values ​​do not match, and the data is determined to originate from an illegal node and is discarded. Step 3-2: Figure 3 As shown, the data of smart meter i is decrypted, the interference value of the quadratic Sine chaotic mapping is reversed, and the inverse operation is performed on the encrypted data. Step 3-2-1: Calculate the initial value of smart meter i:

[0135]

[0136] In the formula, Let i be the power of smart meter i at a certain moment. The upper and lower limits are the power limits; Step 3-2-2: Perform a second-order Sine mapping iteration for smart meter i, with the formula as follows:

[0137]

[0138] In the formula, Let i be the chaotic control parameters for smart meter i; Step 3-2-3: Calculate the interference value of smart meter i:

[0139]

[0140] Step 3-2-4: Decrypt the data from smart meter i:

[0141]

[0142] In the formula, Carbon emission data after decryption of smart meter i; The encrypted carbon emission data for smart meters. This is the scaling factor;

[0143] Step 3-3: Build a load-aware mechanism in the cloud layer, such as... Figure 4 .

[0144] The carbon data task load model is established by combining the specific characteristics of electricity carbon data and defining a cloud task state vector, the formula of which is:

[0145]

[0146] In the formula, This indicates whether the task has been scheduled; 1 represents that it has been scheduled, and 0 represents that it has not been scheduled. This represents the task load requirement at time t; The deadline for the task; The security levels correspond to different encryption and decryption complexities: Level 3 is high sensitivity, Level 2 is medium sensitivity, and Level 1 is low sensitivity. The CPU cores operate in two modes: bound cores and shared cores. Bound cores are used for decrypting highly sensitive data, while shared cores are used for ordinary data. An extended CPU core state vector is used to adapt to carbon data processing requirements; the formula is:

[0147]

[0148] In the formula, Displays the CPU core's operating mode; Let be the utilization rate of kernel j at time t; The operating frequency of core j; The temperature of nucleus j; The average latency of the currently processed task in kernel j;

[0149] Steps 3-4: Real-time load assessment is achieved through a state-action-excitation loop. The state input is the task load vector. With system state vector The combination of these factors results in an action output that, based on the load level determination, corresponds to different resource allocation strategies. The incentive function formula is as follows:

[0150]

[0151] In the formula, For average nuclear utilization rate; The average latency of the task; This represents the average memory usage.

[0152] When initializing the CPU resource table and performing core binding operations, system environment modeling needs to consider the current core type: highly sensitive tasks are bound to marked cores, and ordinary tasks are bound to shared cores. When receiving encrypted data uploaded from the edge side, the task security level should be considered. Determine whether core binding is needed. Forced core binding is executed when the value is 3.

[0153] The optimal core is selected based on the Q-Learning algorithm, minimizing the number of cores and reducing power consumption. The excitation function formula is as follows:

[0154]

[0155] In the formula, Bind the core percentage at time t-1. The percentage of kernels after binding at time t;

[0156] Steps 3-5: Dynamic resource allocation. While meeting the near real-time carbon data processing requirements, the cloud layer improves CPU resource utilization, reduces system power consumption, reduces frequency in low-load scenarios, increases frequency in high-load scenarios to ensure carbon data processing latency meets standards, and balances performance and power consumption in medium-load scenarios.

[0157] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the present invention. Various changes and modifications can be made to the present invention without departing from its spirit and scope. All such changes and modifications fall within the scope of the present invention as claimed, which is defined by the appended claims and their equivalents.

Claims

1. A cloud edge end collaborative based power carbon emission near real-time data collection method, characterized in that, Comprise: Step 1: terminal layer-carbon emission source data collection; Step 1-1: deploy terminal collection nodes, deploy intelligent electric meters i at power key nodes of power plant rooms and user electric meter boxes; Step 1-2: encrypt the carbon emission raw data collected by the intelligent electric meter i using secondary Sine chaos mapping; Step 1-3: calculate the minimum hop count of each node to other nodes according to the network connection topology of the intelligent electric meter i, and establish a hop count matrix M; Step 1-4: set dynamic adjustment of collection frequency; Step 1-5: obtain the minimum delay overhead of the intelligent electric meter; Step 2: edge layer-frequency regulation, hierarchical transmission; Step 2-1: determine the optimal data packet size, spectrum sensing frequency and time of each spectrum sensing; Step 2-2: data priority sorting; Step 3: cloud layer-edge load sensing; Step 3-1: verify data legitimacy, decrypt intelligent electric meter i data, reverse the interference value of secondary Sine chaos mapping, and perform inverse operation on encrypted data; Terminal layer-carbon emission source data collection includes: Step 1-1: deploy terminal collection nodes, deploy intelligent electric meters i at power key nodes of power plant rooms and user electric meter boxes; Step 1-2: encrypt the carbon emission raw data collected by the intelligent electric meter i using secondary Sine chaos mapping, the formula is: In the formula, the state variable of the smart meter i is set as , is the iteration number mark, the initial value ; is the power of the smart meter i at a certain time, is the corresponding power upper and lower limit, and the quadratic Sine mapping iteration is performed, and the formula is: In the formula, is the control parameter of the smart meter i, and the interference value of the first iteration of the smart meter i is calculated according to the formula: Raw data related to carbon emissions of smart meter i The formula after encryption is: In the formula, encrypted carbon emission data for the smart meter i, is a scaling factor; Step 1-3: calculate the minimum hop count of each node to other nodes according to the network connection topology of the intelligent electric meter i, and establish a hop count matrix M; to screen the collection terminal candidate points, define a vector L to store the maximum hop count of each row in M, the formula is: In the formula, M is the hop count matrix of each node to other nodes, L is a variable that stores the maximum hop count of each row in the hop count matrix M, the value in L is compared to preliminarily screen the appropriate collection terminal candidate points, and n is the number of intelligent electric meters; if there are several minimum values of L, the minimum average hop count point is selected as the final value as the collection terminal, the formula is: Collection terminal position=argmin(L)(6) Step 1-4: dynamically adjust the collection frequency, use the high-frequency collection mechanism to dynamically adjust the collection frequency according to the importance of the carbon emission source, define the collection frequency selection variable, the formula is: The sum of each collection frequency selection variable is less than or equal to 1(7) The optimal packet size and the number of spectrum sensing are solved by dynamic programming algorithm. The probability that the smart meter node has data to transmit is , and the formula is wherein, is the base of the natural logarithm, is the average arrival rate of smart meters i, is the transmission time of smart meter i data packets, is the number of spectrum sensing of smart meter i; Step 1-5: obtain the minimum delay overhead of the intelligent electric meter, the formula is: In the formula, is the delay overhead of the smart meter i data packet; is the transmission delay time of the smart meter i data packet.

2. The cloud edge-end collaborative based power carbon emission near real-time data collection method according to claim 1, characterized in that, Edge layer-frequency regulation, hierarchical transmission includes: Step 2-1: define the optimization objective function f as the ratio of the delay overhead and the average effective transmission time of the data packet of the intelligent electric meter i, determine the optimal data packet size by the maximum f value, the formula is: In the formula, f is an optimization objective function, is the average effective transmission time of the smart meter i. The spectrum sensing frequency and the data packet size, and the time of each spectrum sensing, satisfy the constraint conditions as formula (11) and formula (12), calculate the optimal spectrum sensing frequency under the premise of satisfying the maximum tolerable delay of the terminal, the formula is: wherein, is the transmission rate of data on the smart meter i channel, is the total amount of data for the smart meter i task, is the time for spectrum sensing of the smart meter i, is the maximum tolerated latency of the smart meter i terminal; Step 2-2: data priority sorting, based on the weighted average queue WFQ algorithm, the priority is sorted by binding the weight and the virtual completion time, and the formula is: In the formula, is the classification entry determination mark, 1 represents that the kth carbon data packet of the ith queue is selected for entry and forwarding, 0 represents that the packet belongs to other packets for others, and the packet does not temporarily meet the entry priority and enters the waiting queue; is the virtual completion time of the kth carbon data packet of the ith queue; is the total number of carbon data queues divided by the edge layer according to the data service attribute; m is the number of carbon data packets to be processed in the ith queue; Step 2-3: Edge layer initialization virtual time When the jth event occurs, update the virtual time, the formula is: wherein, is the time of event occurrence, is the time interval is the set of active queues in the inner, is the time interval, is the two-dimensional matching relationship; when the queue i has data transmission, its virtual acceptance rate is formula (15), and the actual allocation rate is formula (16), and the high-weight queue has a higher bandwidth allocation proportion, and the formula is: wherein is the virtual acceptance rate for queue i, is the actual allocation rate for queue i, is the sum of weights for all queues in the active queue set, is the available bandwidth for the edge tier; For the kth carbon data packet in the queue i, the virtual start time is calculated by formula (17), the time coefficient of the queue i is solved by formula (18), and the virtual completion time is calculated by formula (19). wherein, is the virtual start time of the kth carbon data packet in queue i; is the virtual finish time of the previous packet in queue i; is the system virtual time at which the packet arrived; is the queue time coefficient; B is the reference bandwidth is the virtual finish time of the kth carbon data packet in queue i; is the virtual start time of the kth carbon data packet in queue i; is the carbon data packet length; Step 2-4: Encrypted data forwarding, receiving terminal layer transmission final encrypted data and chaos control parameters, and uploading to the cloud layer.

3. The cloud edge-end collaborative based power carbon emission near real-time data collection method according to claim 1, characterized in that, Cloud tier-edge load awareness includes: Step 3-1: Verify the legitimacy of the data, verify the legitimacy of the data source according to the binary hypothesis testing mechanism, and the formula is: In the formula, Reference key for chaos control parameter a of smart meter i recorded in the cloud, Reference key for chaos control parameter b of smart meter i recorded in the cloud, unmatched data is data not encrypted or from illegal nodes, matched data is data encrypted and matched with legal nodes; if the corresponding chaos control parameters of the two are consistent, the encrypted processing values match, it is determined that the data comes from a legal node; if the corresponding chaos control parameters of the two are inconsistent, the encrypted processing values do not match, it is determined that the data comes from an illegal node, and the data is directly discarded. Step 3-2: Smart meter i data decryption, reverse the interference value of the secondary Sine chaos mapping, and perform inverse operation on the encrypted data to calculate the initial value of the smart meter i, and the formula is: In the formula, Pi is the power of the smart meter i at a certain time, Pi is the power of the smart meter i at a certain time, The formula for the second Sine mapping iteration of the smart meter i is: In the formula, is the chaos control parameter of the smart meter i; the interference value of the smart meter i is calculated according to the formula: Step 3-2: Smart meter i data decryption, formula: In the formula, is the decrypted carbon emission data of the smart meter i; is the encrypted carbon emission data of the smart meter i, is a scaling factor; Step 3-3: Build a load awareness mechanism in the cloud tier; Build a cloud task state vector, formula: In the formula, 1 represents that the task has been scheduled, and 0 represents that the task has not been scheduled; represents the task load demand at t time; is the task deadline; is the task security level, corresponding to different encryption and decryption complexity, 3 is high sensitive, 2 is medium sensitive, and 1 is low sensitive; The working mode of CPU core includes binding core and shared core, binding core is used for high sensitive data decryption, shared core is used for ordinary data, expand CPU core state vector, adapt to carbon data processing demand, formula: wherein displaying a working mode of the CPU core; is a utilization of the core j at time t; is a running frequency of the core j; is a temperature of the core j; is an average latency of the core j currently processing a task; Step 3-4: Real-time load assessment through state-action-incentive cycle; State input is a task load vector combined with system state vector Action output is a stimulus function formula corresponding to different resource allocation strategies through the determination of load level In the formula, is the average utilization of the core; is the average latency of the task; is the average memory occupancy; Initialize the CPU resource table, and consider the type of the current core when modeling the system environment during the binding operation. High-sensitivity tasks are bound to marked cores, and ordinary tasks are bound to shared cores. When receiving encrypted data uploaded from the edge side, the task security level is determined determine whether the core needs to be bound, when 3, execute forced binding; select the optimal binding core based on the Q-Learning algorithm, minimize the number of binding cores, reduce power consumption, and the incentive function formula is: In the formula, is the binding core occupancy ratio at time t-1, is the occupancy ratio after binding the core at time t; Step 3-5: Dynamic resource allocation, cloud tier meets the near real-time processing requirements of carbon data, improves CPU resource utilization, reduces system power consumption, low load scenario reduces frequency; High load scenario frequency up, ensure carbon data processing delay standard; Medium load scenario balance performance and power consumption.

Images