A keyboard type power meter power supply real-time identity arbitration method, storage medium, device and equipment

Abstract

The application provides a keyboard type electric meter power supply real-time identity arbitration method, a storage medium, a device and equipment, including: in the power supply starting stage, collecting the typing rhythm characteristics of the user when inputting a fixed sentence on the electric meter keyboard, generating an identity anchor; in the power supply process, randomly or triggeredly requiring the user to input the fixed sentence again, collecting the typing rhythm characteristics of the current input in real time, and comparing the similarity with the identity anchor; based on the similarity comparison result and the input behavior quality, dynamically adjusting the credit score; according to the credit score, executing the corresponding power supply strategy. The application improves the identity arbitration security and real-time performance in the keyboard type electric meter power supply scene through the dynamic identity verification and power supply control, without relying on the traditional password, and realizes the improvement of the power supply resource allocation efficiency and user experience.

Description

Technical Field

[0001] This invention relates to the field of power metering and identity authentication technology, specifically to a method, storage medium, device, and equipment for real-time identity arbitration of power supply using a keypad-type electricity meter. Background Technology

[0002] In existing electricity management technologies, identity verification and power supply control typically rely on user pre-registration, pre-stored identity profiles, or biometric databases. For example, smart meter systems often use passwords, IC cards, or fingerprint recognition to verify user identity, but these methods have significant drawbacks: pre-stored templates require users to register in advance and maintain the database, which not only increases system deployment costs but also introduces privacy risks; once a user's template is stolen, attackers can impersonate legitimate users to perform electricity operations.

[0003] Therefore, there is an urgent need for a keypad meter power supply identity arbitration method that does not require pre-stored identity files, can continuously verify the operator's identity in real time, and has a hardware security mechanism that causes the user to forget their identity when power is off. Summary of the Invention

[0004] Based on this, in order to solve the above-mentioned technical problems in existing power supply scenarios, a real-time identity arbitration method, storage medium, device and equipment for keypad-type electricity meter power supply is proposed.

[0005] This invention protects a real-time identity arbitration method for power supply of a keypad-type electricity meter, comprising: during the power supply start-up phase, collecting the typing rhythm features of the user when inputting a fixed sentence on the meter keypad to generate an identity anchor; during the power supply process, randomly or triggeredly requesting the user to input the fixed sentence again, collecting the typing rhythm features of the current input in real time, and comparing them with the identity anchor for similarity; dynamically adjusting the credit score based on the similarity comparison result and the quality of the input behavior; and executing the corresponding power supply strategy according to the credit score.

[0006] Furthermore, during the power supply startup phase, the typing rhythm characteristics of the user when entering a fixed sentence on the meter keypad are collected to generate an identity anchor, including: displaying a fixed standard identity declaration sentence in response to the user pressing the start key; collecting raw data of the user's complete keyboard behavior when typing the sentence, including key intervals, press duration, force variation, backspace mode, and pause distribution; generating an anchor template through the edge computing unit within the meter, extracting the deep feature embedding vector of the raw data as the identity anchor for this electricity usage session; evaluating the intrinsic behavioral quality of the anchor template in real time, generating an initial credit score, and determining whether to grant initial power supply and overdraft limit based on the initial credit score; starting power supply, and temporarily storing the anchor template and the current credit score in the secure isolation area memory of the keypad MCU, binding it with the current session ID.

[0007] Furthermore, an anchor template is generated through the edge computing unit within the meter, and a deep feature embedding vector is extracted from the original data as the identity anchor for this electricity consumption session. This includes: generating a multi-dimensional deep feature embedding vector through real-time inference on the edge computing unit using a pre-trained lightweight temporal convolutional network; the input layer of the temporal convolutional network receives the original keyboard event stream, which includes the key press timestamp, release timestamp, force sensor analog-to-digital converter value, and backspace key trigger flag for each key; the first layer of the temporal convolutional network performs local pattern capture on the key interval sequence through multiple one-dimensional causal convolutional kernels and outputs a feature map; and a channel attention mechanism is used to process the feature channels at each time step. The algorithm employs row weighting to enhance rhythm patterns relevant to user habits; a global average pooling layer compresses variable-length sequences into fixed-length multidimensional vectors as anchor templates; the multidimensional floating-point array of the anchor templates is forcibly stored in the secure area of ​​the keyboard microcontroller unit, which is configured by a memory protection unit to allow access only by the neural network processor and a dedicated encryption engine; when a power interruption detects a drop in mains power, a session termination signal is sent to the keyboard microcontroller unit; the microcontroller unit's interrupt service routine executes at the highest priority, calling a hardware random number generator to generate multi-bit random data, which is then directly overwritten into the static random access memory area storing the anchor templates via the direct memory access controller.

[0008] Furthermore, during the power supply process, the user is randomly or triggered to re-enter a fixed sentence. The typing rhythm features of the current input are collected in real time and compared with the identity anchor for similarity. This includes: responding to the user's operation on the meter or random system checks, requiring the user to re-enter the fixed sentence, and collecting the typing rhythm feature vector of this input; calculating the cosine similarity between the feature vector and the anchor template in memory to obtain an identity continuity score; analyzing the intrinsic behavioral quality of this input, including stability and hesitation, to generate a real-time status score; and calculating the final decision weight based on the identity continuity score and the real-time status score.

[0009] Furthermore, the feature vector and the anchor template in memory are used to calculate the cosine similarity to obtain the identity continuity score, including: when the cosine similarity is greater than 0 (first threshold), the continuity factor is set as the first continuity factor; when the cosine similarity is between the first threshold and the second threshold, the continuity factor is set as the second continuity factor; when the cosine similarity is less than the second threshold, the continuity factor is set as the third continuity factor, and a locking operation is triggered.

[0010] Furthermore, the inherent behavioral quality of this input is analyzed, including stability and hesitation, to generate a real-time state score. This includes: when the standard deviation of the key interval of this input is less than a preset stability threshold and the number of backspaces is less than a preset error threshold, the state factor is set as the first state factor; when the standard deviation of the key interval of this input is greater than or equal to the preset stability threshold, or the number of backspaces is greater than or equal to the preset error threshold, or the fluctuation amplitude of the force sensor value is greater than the preset disorder threshold, the state factor is set as the second state factor; the state factor is multiplied by the preset base score to obtain the real-time state score.

[0011] Furthermore, based on the identity continuity score and the real-time status score, the final decision weight is calculated, including: weighting and summing the identity continuity score and the real-time status score to obtain the final decision weight; when the final decision weight is lower than the first weight threshold, power supply is interrupted and the device is locked; when the final decision weight is lower than the second weight threshold but higher than or equal to the first weight threshold, subsequent operations are restricted, overdraft requests are rejected, and power supply is reduced; when the final decision weight is higher than or equal to the second weight threshold, normal power supply is maintained and accumulated credit is allowed for overdraft.

[0012] This invention protects a real-time identity arbitration device for a keypad-type electricity meter, comprising: a data acquisition module, used to acquire the typing rhythm features of a user inputting a fixed sentence on the meter keypad during the power supply startup phase, and generate an identity anchor; a comparison module, used to randomly or triggerfully request the user to input the fixed sentence again during the power supply process, acquire the typing rhythm features of the current input in real time, and compare the similarity with the identity anchor; and an adjustment and execution module, used to dynamically adjust the credit score based on the similarity comparison results and the quality of the input behavior; and execute the corresponding power supply strategy according to the credit score.

[0013] This invention protects an electronic device, comprising: a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor and the memory communicate via the bus. The machine-readable instructions are executed by the processor to perform the steps of a keypad-type electricity meter power supply real-time identity arbitration method.

[0014] This invention protects a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of a real-time identity arbitration method for power supply to a keypad-type electricity meter.

[0015] This invention protects a real-time identity arbitration method for power supply in a keypad-type electricity meter. It generates an identity anchor by collecting the typing rhythm features of a user when inputting a fixed sentence on the meter's keypad. During power supply, the user is randomly or triggered to re-enter the sentence, and the similarity between the current typing rhythm features and the identity anchor is compared in real time. Simultaneously, a credit score is dynamically adjusted based on the quality of the input behavior, and a corresponding power supply strategy is executed according to the credit score. This solution, based on dynamic identity verification and power supply control using biometric behavioral features, eliminates the need for traditional passwords or hardware tokens, improving the security and real-time performance of identity arbitration in power supply scenarios. Furthermore, it enables rapid local erasure of stored identity information in the event of a power outage, effectively preventing identity misuse or illegal electricity use. The credit scoring mechanism also flexibly adjusts the power supply strategy, effectively improving the allocation efficiency of power resources and the user experience. Attached Figure Description

[0016] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This application provides a flowchart of an authentication and power supply control method based on typing rhythm features. Figure 2 This application provides a flowchart of keyboard behavior-based identity anchoring and initial power supply. Figure 3 This application provides a flowchart of a method for generating and destroying keyboard user behavior anchor templates. Figure 4 A flowchart of an authentication method during power supply provided in an embodiment of this application. Figure 5 A flowchart of a continuous factor setting method provided in an embodiment of this application. Figure 6 A flowchart of a real-time status scoring calculation method provided in this application embodiment. Figure 7 This application provides a flowchart of a power supply control method based on a weighted threshold. Figure 8 The diagram shows the keypad power supply control device provided in this application embodiment. Figure 9 The keyboard meter power supply control electronic device provided in the embodiments of this application is shown in the figure. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. Based on the embodiments of this application, every other embodiment obtained by those skilled in the art without inventive effort falls within the scope of protection of this application.

[0018] Research has revealed that current power supply scenarios rely primarily on pre-registered templates or static passwords for user authentication, which has significant shortcomings. Furthermore, static passwords are easily stolen or shared, failing to effectively distinguish whether the current operator is the same person as the initial power user. Additionally, attackers who obtain template files stored on the device can forge input behavior to continue using the power supply. Moreover, existing technologies lack real-time evaluation and credit adjustment mechanisms for dynamic changes in input behavior, making it difficult to respond to operator changes or abnormal behavior during power supply, resulting in low power supply security and a poor user experience.

[0019] Based on this, please refer to Figure 1 This application provides a real-time identity arbitration method for power supply of a keypad-type electricity meter, including: During the power-on phase, S101 collects the typing rhythm features of users when they input fixed sentences on the meter keypad to generate identity anchors. S102 During the power supply process, the user is randomly or triggered to input the fixed sentence again, and the typing rhythm features of the current input are collected in real time and compared with the identity anchor for similarity. S103 Based on the similarity comparison results and the quality of the input behavior, dynamically adjust the credit score; and execute the corresponding power supply strategy according to the credit score.

[0020] This embodiment provides a real-time identity arbitration method for power supply of keypad-type electricity meters. It generates an identity anchor during the anchoring stage and performs continuous or triggered verification during the power supply process. This allows the method to determine whether the current operator is the same person as the initial electricity user without relying on a pre-registration template, effectively preventing power supply risks caused by operator changes.

[0021] Simultaneously, the credit score is dynamically adjusted based on similarity comparison results and input behavior quality, and a differentiated power supply strategy is implemented accordingly, enabling direct demotion or power interruption for operations not performed by the account holder. The identity anchor exists only in session memory and is hardware-erased upon power interruption, preventing attackers from stealing templates and enhancing anti-spoofing capabilities.

[0022] In step S101, during the power supply startup phase, the typing rhythm characteristics of the user when entering a fixed sentence on the meter keypad are collected to generate an identity anchor.

[0023] When a user first activates the power supply, the system analyzes the user's typing patterns when entering fixed sentences on the meter keypad, extracting unique typing rhythm features. Based on these features, it generates an identity anchor unique to this electricity usage session, which serves as the benchmark for verifying the continuity of the operator's identity in subsequent transactions. Typing rhythm features include keystroke speed, pause length, key pressure, and backspace key usage patterns, all of which collectively constitute a unique identifier for the user's input behavior.

[0024] In one implementation, reference Figure 2 Step S101 includes: S1011. In response to the user pressing the start button, display a fixed standard identity declaration. In this step, after the user presses the start power button on the meter, the meter screen immediately displays a fixed standard identity declaration. In a real-world production scenario, the declaration could be "I confirm that I am responsible for this electricity usage," serving as a sampling carrier for user input. This declaration remains consistent across all electricity usage sessions to ensure that subsequent comparisons are based on the same information.

[0025] S1012. Collect the complete keyboard behavior raw data when the user types the sentence. The raw data includes key interval, press duration, force variation, backspace mode, and pause distribution.

[0026] When a user begins typing the displayed statement on the meter's keypad, the meter uses sensors on the keypad to collect raw data of the complete keystroke behavior in real time. Optionally, the system records the press and release timestamps of each key to calculate the key interval and press duration, acquires the force value of each keystroke through a force sensor to generate a force variation curve, monitors the frequency and timing of backspace key usage, and analyzes the distribution of pauses between individual keys and the overall input process, including the duration and location of hesitation intervals. This raw data constitutes a complete snapshot of rhythmic behavior, which is linked to the user.

[0027] S1013. An anchor template is generated by the edge computing unit in the meter, and the deep feature embedding vector of the original data is extracted as the identity anchor for this electricity consumption session.

[0028] In this step, the edge computing unit inside the meter immediately processes the collected raw data in real time and extracts deep feature embedding vectors through a pre-trained lightweight neural network. These vectors serve as the identity anchors for this electricity usage session and are used for subsequent identity continuity comparisons.

[0029] In one implementation, reference Figure 3 Step S1013 includes: S10131. A multi-dimensional deep feature embedding vector is generated by real-time inference on the edge computing unit through a pre-trained lightweight temporal convolutional network.

[0030] The input layer of the temporal convolutional network receives a raw keyboard event stream, which includes the key press timestamp, release timestamp, force sensor analog-to-digital converter value, and backspace key trigger flag for each key. In this step, the edge computing unit invokes a pre-trained lightweight temporal convolutional network to perform real-time inference on the raw keyboard event stream. The input layer of the network receives a continuous keyboard event stream, where each event includes a key press timestamp and release timestamp to calculate temporal relationships, a force sensor analog-to-digital converter value to quantify key pressure, and a backspace key trigger flag to identify corrective actions. This keyboard data is organized into a temporal sequence and used as the network input.

[0031] S10132. The first layer of the temporal convolutional network captures local patterns of the key interval sequence through multiple one-dimensional causal convolutional kernels and outputs a feature map; the feature channels of each time step are weighted through a channel attention mechanism, and a rhythmic pattern strongly coupled with user habits is generated based on the feature channel weighting; the global average pooling layer compresses the variable-length sequence into a fixed-length multi-dimensional vector as the anchor template.

[0032] The first layer of the temporal convolutional network uses multiple one-dimensional causal convolutional kernels, each with a size of 3, to capture local patterns in the key press interval sequence, extracting local features such as the speed variation of consecutive key presses, and outputting a feature map. Subsequently, a channel attention mechanism is used to weight the feature channels at each time step, reinforcing rhythmic patterns highly correlated with the user's personal habits. Optionally, it selects the pause preferences or intensity variation patterns of specific key positions. Finally, a global average pooling layer compresses the variable-length feature map into a fixed-length 128-dimensional vector, which corresponds to the anchor template.

[0033] S10133, The multidimensional floating-point array of the anchor template is forcibly stored in the secure area of ​​the keyboard microcontroller unit, which is configured by the memory protection unit to allow access only by the neural network processor and the dedicated encryption engine.

[0034] The 128-dimensional floating-point array of the generated anchor template is forcibly stored in the static random access memory secure area of ​​the keyboard microcontroller unit. This area is configured by the memory protection unit to allow only the neural network processor and the dedicated encryption engine to read and write access. The main control central processing unit cannot read it directly, thereby preventing the template from being illegally stolen.

[0035] S10134. When a power interruption is detected and a drop in the main power supply is detected, a session end signal is sent to the keyboard microcontroller unit. The interrupt service routine of the microcontroller unit is executed at the highest priority, and the hardware random number generator is called to generate multi-bit random data. The random data is directly overwritten into the static random access memory area of ​​the storage anchor template through the direct memory access controller.

[0036] In this step, when power is interrupted for any reason, the independent chip, upon detecting a mains power failure, immediately sends a session termination signal to the keyboard microcontroller unit via a dedicated hardware interrupt line. Simultaneously, the microcontroller unit's interrupt service routine executes with the highest priority. First, it calls the hardware random number generator to generate 128-bit random data. Then, it directly overwrites this random data into the static random access memory area storing the anchor template via the direct memory access controller. In an optional implementation, the memory overwrite should be performed multiple times, more than three times, to ensure thorough data erasure. Only after the overwrite is complete does the microcontroller unit perform a system reset, thus preventing any physical probe from capturing the user's biometric data at the moment of power failure.

[0037] S1014. Evaluate the intrinsic behavioral quality of the anchor template in real time, generate an initial credit score, and decide whether to grant an initial power supply and overdraft limit based on the initial credit score.

[0038] Specifically, while generating the anchor template, the system evaluates the template's inherent behavioral quality in real time, including key indicators such as the stability of the input rhythm, hesitation, and smoothness of intensity. Based on preset rules, it generates an initial credit score, ranging from 10 to 100 in one optional mode. Depending on the initial credit score, the system determines whether to provide initial power and the allowed overdraft limit. Optionally, based on factory settings, if the initial credit score is higher than 60, normal power supply is allowed with a certain overdraft limit; if it is lower than 30, only basic guaranteed power is provided.

[0039] S1015. Power on and temporarily store the anchor template and current credit score in the secure isolation area memory of the keyboard MCU, and bind them to the current session ID.

[0040] In this step, after completing the initial assessment and determining the power supply strategy, the system starts supplying power. Simultaneously, the generated anchor template and current credit score are temporarily stored only in the secure, isolated memory area of ​​the keyboard microcontroller unit and bound to a unique identifier ID for the current session. It is important to note that this memory area is hardware-protected and no other irrelevant data is written to it; specifically, it is not written to the main controller flash memory, ensuring that personal data is only valid during the current session.

[0041] In step S102, during the power supply process, the user is randomly or triggered to input the fixed sentence again, and the typing rhythm features of the current input are collected in real time and compared with the identity anchor for similarity.

[0042] Specifically, during the power supply period, the system randomly checks or responds to user operations on the meter, requiring the user to re-enter the same fixed sentence as in the startup phase, and collects the typing rhythm feature vector of this input in real time. Then, it compares the similarity with the identity anchor stored in memory to determine whether the current operator is the same person as the initial anchor.

[0043] The comparison results are used for subsequent credit score adjustments and power supply strategy implementation. In one implementation, see [link to implementation details]. Figure 4 Step S102 includes: S1021. In response to the user's operation on the meter or random system sampling, the user is required to re-enter the fixed sentence, and the typing rhythm feature vector of this input is collected.

[0044] In this step, whenever the user performs any operation on the meter—in one optional implementation, this could be purchasing more electricity, checking the balance, or responding to an alarm—the system automatically triggers a verification request, requiring the user to re-enter a fixed standard identity declaration sentence. Additionally, for long-term electricity consumption scenarios, the system will randomly pop up a heartbeat verification window, requiring the user to re-enter the sentence. During the user's typing process, the meter collects the complete raw keyboard behavior data in real time and extracts a deep feature embedding vector of the same dimension as the startup phase through an edge computing unit, serving as the typing rhythm feature vector for this input.

[0045] S1022. Calculate the cosine similarity between the feature vector and the anchor template in memory to obtain the identity continuity score; analyze the intrinsic behavioral quality of the current input, including stability and hesitation, and generate a real-time status score.

[0046] In this step, the system calculates the cosine similarity between the collected feature vector and the anchor template stored in memory, obtaining a similarity value between 0 and 1. This value serves as the basis for identity continuity scoring. Simultaneously, the system analyzes the inherent behavioral quality of the input, including key data such as the stability of key spacing, the frequency of backspace key usage, and the smoothness of the force curve. Alternatively, other behavioral data can be selected to generate a real-time status score reflecting the naturalness of the current input behavior.

[0047] S1023. Calculate the final decision weight based on the identity continuity score and the real-time status score. In this step, the system weights and combines the identity continuity score and the real-time status score to calculate a final decision weight, which is used to determine the subsequent power supply strategy. The identity continuity score reflects the consistency of the operator's identity, and the real-time status score reflects the intrinsic quality of the input behavior; together, they determine the power supply strategy available to the user in the current power consumption session.

[0048] In one implementation, step S1022 involves calculating the cosine similarity between the feature vector and the anchor template in memory to obtain an identity continuity score, which is then referenced. Figure 5 include: S10221. When the cosine similarity is greater than the first threshold, the continuous factor is set as the first continuous factor.

[0049] In this step, one optional approach is to determine that if the cosine similarity between the current feature vector and the anchor template is greater than 0.9, then it is determined to be the same person, and the continuity factor is set to 1.0, indicating a high degree of identity consistency.

[0050] S10222 When the cosine similarity is between the first threshold and the second threshold, the continuous factor is set to the second continuous factor.

[0051] In this step, if the cosine similarity is between 0.7 and 0.9, it is judged that the user's consistency is questionable, and the continuous factor is set to 0.6, indicating that there is a certain degree of uncertainty in the identity.

[0052] S10223. When the cosine similarity is less than the second threshold, the continuous factor is set to the third continuous factor, and a locking operation is triggered.

[0053] In this step, if the cosine similarity is less than 0.7, it is judged to be highly suspicious or not the person in question. The continuous factor is set to 0, and a locking operation is immediately triggered, power supply is interrupted, and the event is recorded.

[0054] In one implementation, step S1022 involves analyzing the intrinsic behavioral quality of the current input, including stability and hesitation, and generating a real-time state score, which is then referenced. Figure 6 include: S10224. When the standard deviation of the key interval of the current input is less than the preset stability threshold and the number of backspaces is less than the preset error threshold, the state factor is set to the first state factor.

[0055] In this step, if the standard deviation of the key interval of the current input is less than the preset stability threshold and the number of backspaces is less than the preset error threshold, it is judged as stable and smooth, and the state factor is set to 1.0.

[0056] S10225. When the standard deviation of the key interval of the current input is greater than or equal to the preset stability threshold, or the number of backspaces is greater than or equal to the preset error threshold, or the fluctuation amplitude of the force sensor value is greater than the preset disorder threshold, the state factor is set to the second state factor.

[0057] In this step, if the standard deviation of the key interval is greater than or equal to the preset stability threshold, or the number of backspaces is greater than or equal to the preset error threshold, or the fluctuation amplitude of the force sensor value is greater than the preset disorder threshold, it is judged as disorder, mania, or too many input errors. The state factor is set to a value between 0.2 and 0.5. The specific value is dynamically adjusted according to the degree of abnormality. The dynamic adjustment strategy is determined by the specific scenario.

[0058] S10226. Combine the state factor with the preset basic score to obtain the real-time state score.

[0059] Optionally, in this step, the determined state factors are multiplied by a preset base score to obtain a real-time state score, which reflects the intrinsic quality of the current input behavior.

[0060] In one implementation, step S1023, refer to Figure 7 ,include: S10231. The identity continuity score and the real-time status score are weighted and summed to obtain the final decision weight.

[0061] The identity continuity score and the real-time status score are weighted and summed, with the identity continuity score having a weight of 70% and the real-time status score having a weight of 30%. The calculation formula is W_new = W_old * identity continuity score + real-time status score adjustment, where W_old is the current credit score, the identity continuity score is determined by the continuity factor, and the real-time status score adjustment is calculated from the status factor and the preset base score.

[0062] S10232. When the final decision weight is lower than the first weight threshold, the power supply is interrupted and the device is locked.

[0063] In one optional approach, if the final decision weight W_new is less than 20, the power supply is immediately interrupted, the device is locked, and a message is displayed on the screen indicating that the operator may have changed and the power supply has been terminated.

[0064] S10233. When the final decision weight is lower than the second weight threshold but higher than or equal to the first weight threshold, restrict subsequent operations, reject overdraft requests, and reduce power supply.

[0065] In one alternative approach, if the final decision weight W_new is between 20 and 60, then subsequent user operations on the meter are restricted, all overdraft recharge requests are rejected, and the power supply is reduced to the basic guaranteed power, such as providing only lighting power.

[0066] S10234. When the final decision weight is higher than or equal to the second weight threshold, maintain normal power supply and allow accumulated credit to be overdrawn.

[0067] In one alternative approach, if the final decision weight W_new is greater than or equal to 60, normal power supply is maintained, allowing users to perform normal power consumption operations and allowing accumulated credit to be overdrawn, for example, up to 10 kWh.

[0068] In step S103, the credit score is dynamically adjusted based on the similarity comparison results and the quality of the input behavior; and the corresponding power supply strategy is executed according to the credit score.

[0069] Specifically, after each verification, the system dynamically updates the credit score of the current electricity consumption session based on the identity continuity score and real-time status score calculated in step S102. This credit score is a dynamic value between 0 and 100, and its adjustment rules comprehensively consider identity consistency and the intrinsic quality of input behavior.

[0070] For example, if the identity continuity score indicates the same person and the input behavior is stable and fluent, the credit score will be improved; if the identity continuity score is suspicious or the input behavior is disordered, the credit score will be deducted accordingly. Subsequently, the system automatically executes the corresponding power supply strategy based on the updated credit score. When the credit score is higher than or equal to 60, the system maintains normal power supply, allowing users to perform normal electricity usage operations and allowing accumulated credit for overdraft, such as overdrafting up to 10 kWh. When the credit score is between 30 and 60, the system prohibits overdraft, requires users to recharge before using electricity, and maintains normal power supply priority. When the credit score is lower than 30, the system delays the response to recharge requests, reduces the power supply to the basic guaranteed power, such as providing only lighting power, and records the event to the local black box.

[0071] In one optional implementation, when disordered or agitated input behavior is detected multiple times consecutively, a protection period, preferably 15 minutes, is initiated, during which no power supply requests are accepted to ensure power safety. Through this dynamic adjustment mechanism, the system can assess the credibility of the current operator in real time without pre-storing user profiles and flexibly adjust the power supply strategy, thereby effectively preventing identity theft and abnormal power consumption.

[0072] Please see Figure 8 , Figure 8This is a schematic diagram of a keypad-type electricity meter power supply real-time identity arbitration device provided in an embodiment of this application. Figure 8 As shown, the generating apparatus 800 includes: The data acquisition module 810 is used to collect the typing rhythm characteristics of users when they enter fixed sentences on the keypad of the electricity meter during the power supply start-up phase, and generate identity anchors. The comparison module 820 is used to randomly or trigger the user to input the fixed sentence again during the power supply process, collect the typing rhythm features of the current input in real time, and compare the similarity with the identity anchor. The adjustment execution module 830 is used to dynamically adjust the credit score based on the similarity comparison results and the quality of the input behavior; and to execute the corresponding power supply strategy according to the credit score.

[0073] Furthermore, when the acquisition module 810 is used to collect the typing rhythm features of the user when inputting a fixed sentence on the meter keypad during the power-on startup phase and to generate an identity anchor, the acquisition module 810 is used for: In response to the user pressing the start button, a fixed standard identity declaration statement is displayed; Collect raw data of complete keyboard behavior when the user types the sentence. The raw data includes key intervals, press duration, force variation, backspace mode, and pause distribution. An anchor template is generated by the edge computing unit inside the electricity meter, and the deep feature embedding vector of the original data is extracted as the identity anchor for this electricity consumption session. The inherent behavioral quality of the anchor template is evaluated in real time, an initial credit score is generated, and an initial power supply and overdraft limit are determined based on the initial credit score. Power is activated, and the anchor template and current credit score are temporarily stored in the secure isolation area memory of the keyboard MCU and bound to the current session ID.

[0074] Furthermore, when the acquisition module 810 generates an anchor template through the edge computing unit within the electricity meter and extracts the deep feature embedding vector of the original data as the identity anchor for this electricity consumption session, the acquisition module 810 is used to: A pre-trained lightweight temporal convolutional network performs real-time inference on the edge computing unit to generate multi-dimensional deep feature embedding vectors. The input layer of the temporal convolutional network receives the original keyboard event stream, which includes the key press timestamp, release timestamp, force sensor analog-to-digital converter value, and backspace key trigger flag for each key. The first layer of the temporal convolutional network captures local patterns of the key interval sequence through multiple one-dimensional causal convolutional kernels and outputs a feature map; the feature channels of each time step are weighted through a channel attention mechanism to enhance the rhythmic patterns related to user habits; the global average pooling layer compresses the variable-length sequence into a fixed-length multi-dimensional vector as the anchoring template. The multidimensional floating-point array of the anchor template is forcibly stored in the secure area of ​​the keyboard microcontroller unit, which is configured by the memory protection unit to allow access only to the neural network processor and the dedicated encryption engine. When a power interruption is detected and a drop in the main power supply is detected, a session termination signal is sent to the keyboard microcontroller unit. The interrupt service routine of the microcontroller unit is executed at the highest priority, calling the hardware random number generator to generate multi-bit random data, and directly overwriting the random data into the static random access memory area of ​​the storage anchor template through the direct memory access controller.

[0075] Furthermore, when the comparison module 820 is used to randomly or trigger the user to input the fixed sentence again during power supply, collect the typing rhythm features of the current input in real time, and compare the similarity with the identity anchor, the comparison module 820 is used to: In response to user operations on the meter or random system checks, the user is required to re-enter the fixed sentence, and the typing rhythm feature vector of this input is collected. The feature vector is compared with the anchor template in memory using cosine similarity calculation to obtain the identity continuity score; the intrinsic behavioral quality of the current input, including stability and hesitation, is analyzed to generate a real-time status score. The final decision weight is calculated based on the identity continuity score and the real-time status score.

[0076] Furthermore, when the comparison module 820 calculates the cosine similarity between the feature vector and the anchor template in memory to obtain an identity continuity score, the comparison module 820 is used to: When the cosine similarity is greater than the first threshold, the continuous factor is set to the first continuous factor; When the cosine similarity is between the first threshold and the second threshold, the continuous factor is set to the second continuous factor; When the cosine similarity is less than the second threshold, the continuous factor is set to the third continuous factor, and a locking operation is triggered.

[0077] Furthermore, when the comparison module 820 analyzes the intrinsic behavioral quality of the current input, including stability and hesitation, and generates a real-time state score, the comparison module 820 is used to: When the standard deviation of the key interval of the current input is less than the preset stability threshold and the number of backspaces is less than the preset error threshold, the state factor is set to the first state factor. When the standard deviation of the key interval of the current input is greater than or equal to the preset stability threshold, or the number of backspaces is greater than or equal to the preset error threshold, or the fluctuation amplitude of the force sensor value is greater than the preset disorder threshold, the state factor is set to the second state factor. The real-time status score is obtained by multiplying the status factor by the preset base score.

[0078] Furthermore, when calculating the final decision weight based on the identity continuity score and the real-time status score, the comparison module 820 is used to: The final decision weight is obtained by weighting and summing the identity continuity score and the real-time status score. When the final decision weight is lower than the first weight threshold, power is interrupted and the device is locked. When the final decision weight is lower than the second weight threshold but higher than or equal to the first weight threshold, subsequent operations are restricted, overdraft requests are rejected, and power supply is reduced. When the final decision weight is higher than or equal to the second weight threshold, normal power supply is maintained and accumulated credit is allowed to be overdrawn.

[0079] The real-time identity arbitration device for a keypad-type electricity meter provided in this application generates an identity anchor by collecting the typing rhythm features of a user when inputting a fixed sentence on the meter keypad during the power supply startup phase. During the power supply process, the user is randomly or triggered to input the fixed sentence again, and the typing rhythm features of the current input are collected in real time and compared with the identity anchor for similarity. Based on the similarity comparison results and the quality of the input behavior, the credit score is dynamically adjusted. According to the credit score, the corresponding power supply strategy is executed. In this way, real-time identity arbitration can be achieved based on the user's unique typing rhythm features without additional hardware, improving power supply security and the convenience of user identity verification. At the same time, the power supply strategy is dynamically adjusted through the credit score, realizing flexible and secure electricity management.

[0080] Please see Figure 9 , Figure 9 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 9 As shown, the electronic device 900 includes a processor 910, a memory 920, and a bus 930.

[0081] The memory 920 stores machine-readable instructions executable by the processor 910. When the electronic device 900 is running, the processor 910 and the memory 920 communicate via the bus 930. When the machine-readable instructions are executed by the processor 910, they can perform the operations described above. Figure 1 The steps of the keypad-type electricity meter power supply real-time identity arbitration method in the method embodiment shown are specifically implemented in the method embodiment and will not be repeated here.

[0082] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, can perform the above-described actions. Figure 1 The steps of the keypad-type electricity meter power supply real-time identity arbitration method in the method embodiment shown are specifically implemented in the method embodiment and will not be repeated here.

[0083] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0084] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the shown or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0085] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0086] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0087] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0088] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The scope of protection of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this application. Such modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for real-time identity arbitration of power supply in a keypad-type electricity meter, characterized in that, include: During the power supply startup phase, the typing rhythm characteristics of users when they enter fixed sentences on the meter keypad are collected to generate identity anchors; In response to the user pressing the start button, a fixed standard identity declaration statement is displayed; Collect raw data of complete keyboard behavior when the user types the sentence. The raw data includes key intervals, press duration, force variation, backspace mode, and pause distribution. An anchor template is generated by the edge computing unit inside the electricity meter, and the deep feature embedding vector of the original data is extracted as the identity anchor for this electricity consumption session. A pre-trained lightweight temporal convolutional network performs real-time inference on the edge computing unit to generate multi-dimensional deep feature embedding vectors. The input layer of the temporal convolutional network receives the original keyboard event stream, which includes the key press timestamp, release timestamp, force sensor analog-to-digital converter value, and backspace key trigger flag for each key. The first layer of the temporal convolutional network captures local patterns of the key interval sequence through multiple one-dimensional causal convolutional kernels and outputs a feature map; the feature channels of each time step are weighted through a channel attention mechanism, and a rhythmic pattern strongly coupled with user habits is generated based on the weighted feature channels; the global average pooling layer compresses the variable-length sequence into a fixed-length multi-dimensional vector, which serves as the anchoring template. The multidimensional floating-point array of the anchor template is forcibly stored in the secure area of ​​the keyboard microcontroller unit, which is configured by the memory protection unit to allow access only to the neural network processor and the dedicated encryption engine. When a power interruption is detected and a drop in the main power supply is detected, a session end signal is sent to the keyboard microcontroller unit. The interrupt service routine of the microcontroller unit is executed at the highest priority, and the hardware random number generator is called to generate multi-bit random data. The random data is directly overwritten into the static random access memory area of ​​the storage anchor template through the direct memory access controller. The inherent behavioral quality of the anchor template is evaluated in real time, an initial credit score is generated, and an initial power supply and overdraft limit are determined based on the initial credit score. Power is activated, and the anchor template and current credit score are temporarily stored in the secure isolation area memory of the keyboard MCU and bound to the current session ID; During the power supply process, the user is randomly or triggered to re-enter the fixed sentence, and the typing rhythm features of the current input are collected in real time and compared with the identity anchor for similarity. Based on the similarity comparison results and the quality of input behavior, the credit score is dynamically adjusted; Based on the credit score, the corresponding power supply strategy is executed.

2. The method according to claim 1, characterized in that, During the power supply process, the user is randomly or triggered to input the fixed sentence again. The typing rhythm features of the current input are collected in real time and compared with the identity anchor for similarity, including: In response to user operations on the meter or random system checks, the user is required to re-enter the fixed sentence, and the typing rhythm feature vector of this input is collected. The feature vector is compared with the anchor template in memory using cosine similarity calculation to obtain the identity continuity score; at the same time, the intrinsic behavioral quality of the current input is analyzed, including stability and hesitation, to generate a real-time status score. The final decision weight is calculated based on the identity continuity score and the real-time status score.

3. The method according to claim 2, characterized in that, The step of calculating the cosine similarity between the feature vector and the anchor template in memory to obtain the identity continuity score includes: When the cosine similarity is greater than the first threshold, the continuous factor is set to the first continuous factor; When the cosine similarity is between the first threshold and the second threshold, the continuous factor is set to the second continuous factor; When the cosine similarity is less than the second threshold, the continuous factor is set to the third continuous factor, and a locking operation is triggered.

4. The method according to claim 2, characterized in that, The analysis assesses the intrinsic behavioral quality of the input, including stability and hesitation, and generates a real-time state score, including: When the standard deviation of the key interval of the current input is less than the preset stability threshold and the number of backspaces is less than the preset error threshold, the state factor is set to the first state factor. When the standard deviation of the key interval of the current input is greater than or equal to the preset stability threshold, or the number of backspaces is greater than or equal to the preset error threshold, or the fluctuation amplitude of the force sensor value is greater than the preset disorder threshold, the state factor is set to the second state factor. The real-time status score is obtained by combining the state factors with the preset basic score.

5. The method according to claim 2, characterized in that, The calculation of the final decision weight based on the identity continuity score and the real-time status score includes: The final decision weight is obtained by weighting and summing the identity continuity score and the real-time status score. When the final decision weight is lower than the first weight threshold, power is interrupted and the device is locked. When the final decision weight is lower than the second weight threshold but higher than or equal to the first weight threshold, subsequent operations are restricted, overdraft requests are rejected, and power supply is reduced. When the final decision weight is higher than or equal to the second weight threshold, normal power supply is maintained and accumulated credit is allowed to be overdrawn.

6. A keypad-type real-time identity arbitration device powered by an electricity meter, characterized in that, include: The data acquisition module is used to collect the typing rhythm characteristics of users when they enter fixed sentences on the meter keypad during the power supply startup phase, and generate identity anchors. In response to the user pressing the start button, a fixed standard identity declaration statement is displayed; Collect raw data of complete keyboard behavior when the user types the sentence. The raw data includes key intervals, press duration, force variation, backspace mode, and pause distribution. An anchor template is generated by the edge computing unit inside the electricity meter, and the deep feature embedding vector of the original data is extracted as the identity anchor for this electricity consumption session. A pre-trained lightweight temporal convolutional network performs real-time inference on the edge computing unit to generate multi-dimensional deep feature embedding vectors. The input layer of the temporal convolutional network receives the original keyboard event stream, which includes the key press timestamp, release timestamp, force sensor analog-to-digital converter value, and backspace key trigger flag for each key. The first layer of the temporal convolutional network captures local patterns of the key interval sequence through multiple one-dimensional causal convolutional kernels and outputs a feature map; the feature channels of each time step are weighted through a channel attention mechanism, and a rhythmic pattern strongly coupled with user habits is generated based on the weighted feature channels; the global average pooling layer compresses the variable-length sequence into a fixed-length multi-dimensional vector, which serves as the anchoring template. The multidimensional floating-point array of the anchor template is forcibly stored in the secure area of ​​the keyboard microcontroller unit, which is configured by the memory protection unit to allow access only to the neural network processor and the dedicated encryption engine. When a power interruption is detected and a drop in the main power supply is detected, a session end signal is sent to the keyboard microcontroller unit. The interrupt service routine of the microcontroller unit is executed at the highest priority, and the hardware random number generator is called to generate multi-bit random data. The random data is directly overwritten into the static random access memory area of ​​the storage anchor template through the direct memory access controller. The inherent behavioral quality of the anchor template is evaluated in real time, an initial credit score is generated, and an initial power supply and overdraft limit are determined based on the initial credit score. Power is activated, and the anchor template and current credit score are temporarily stored in the secure isolation area memory of the keyboard MCU and bound to the current session ID; The comparison module is used to randomly or trigger the user to input the fixed sentence again during the power supply process, collect the typing rhythm features of the current input in real time, and compare the similarity with the identity anchor. The adjustment execution module is used to dynamically adjust the credit score based on the similarity comparison results and the quality of the input behavior; and to execute the corresponding power supply strategy according to the credit score.

7. An electronic device, characterized in that, include: The device includes a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the memory via the bus. The machine-readable instructions are executed by the processor to perform the steps of the real-time identity arbitration method for power supply of a keypad-type electricity meter as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of the real-time identity arbitration method for power supply of a keypad-type electricity meter as described in any one of claims 1 to 5.

Images