Electric quantity prediction method and equipment of intelligent equipment, storage medium and intelligent equipment

Abstract

The invention relates to the technical field of electric quantity measurement, in particular to an electric quantity prediction method and equipment for intelligent equipment, a storage medium and the intelligent equipment, and the method comprises the steps: obtaining historical behavior information collected by at least one function module of the intelligent equipment; inputting the historical behavior information into a first prediction model, and predicting to obtain a future behavior to be generated of at least one functional module and a power consumption demand thereof; obtaining the residual electric quantity condition of a battery on the intelligent equipment; and inputting the residual electric quantity condition and the power consumption demand of the battery into a second prediction model, and determining the predicted runtime of the residual electric quantity of the battery. According to the scheme, the user can know the electric quantity state of the intelligent equipment in advance, and the problem that the door lock cannot be normally used due to sudden power failure is effectively avoided. Moreover, the method can accurately predict the variation trend of the available day number demand of the battery, reduces the charging times of the battery, and improves the user experience.

Description

Technical Field

[0001] This application relates to the field of power measurement technology, specifically to a power prediction method, device, storage medium, and smart device for intelligent devices. Background Technology

[0002] Batteries have unique characteristics in the use of smart locks. Due to current technological limitations, battery storage technology cannot yet store and release power on a large scale. Existing smart locks all have limited battery capacity. Therefore, reducing power consumption and using primary and backup batteries remains the only feasible way to improve the battery life of smart locks. At the same time, to ensure the safety of the battery system and prevent the lock from becoming unusable after the battery is depleted, accurate prediction of the smart lock's battery power requirements is essential.

[0003] There is a supply-demand imbalance in battery capacity and usage. On the one hand, capacity cannot be increased indefinitely, leading to insufficient supply within a limited capacity range, failing to meet user needs and even affecting normal door operation. On the other hand, frequent charging shortens battery life, and battery replacement increases inconvenience and energy waste. Therefore, there is a need for an accurate method to predict the time it takes for a battery to run out of power, in order to accurately forecast the changing trend of battery life and improve user experience. Summary of the Invention

[0004] In view of this, embodiments of this application provide a method for predicting the power consumption of a smart device to predict the power consumption of the battery over time, thereby accurately predicting the changing trend of the battery's available days and improving the user experience.

[0005] In a first aspect, embodiments of this application provide a method for predicting the battery level of a smart device, applied to a smart door lock or smart doorbell, the method comprising: The system acquires historical behavior information collected by at least one functional module of a smart device; inputs the historical behavior information into a first prediction model to predict the future behavior of at least one functional module and its power consumption requirements; acquires the remaining battery power of the smart device; and inputs the remaining battery power and the power consumption requirements into a second prediction model to determine the estimated operating time of the remaining battery power.

[0006] In conjunction with the first aspect, in one possible implementation, acquiring historical behavior information collected by at least one functional module of the smart device includes: The device receives at least one behavioral data reported by at least one functional module of the smart device within a preset period, wherein each behavioral data includes: the type, number, power consumption, and time consumption of a single behavior; the device processes the at least one behavioral data within the preset period to generate the historical behavioral information.

[0007] In conjunction with the first aspect, in another possible implementation, before inputting the remaining battery power and power consumption demand into the second prediction model, the method further includes: configuring a target weight coefficient for each functional module based on historical behavior information; and fusing the target weight coefficient of each functional module with data on future behaviors of each functional module to obtain fused data on future behaviors.

[0008] In conjunction with the first aspect, in another possible implementation, configuring the target weight coefficient for each of the functional modules includes: calculating the expected energy consumption ratio of each of the functional modules in the future preset period based on the historical behavior information of the at least one functional module, and using the expected energy consumption ratio as the target weight coefficient.

[0009] In conjunction with the first aspect, in yet another possible implementation, before using the first prediction model to predict future behavior and the corresponding power consumption demand, the method further includes: Configure at least one influencing factor, which is used to characterize one or more of the following: environmental conditions, hardware and software status of the smart device, usage behavior, and battery status. Based on the influencing factors of each behavior, the data on the power consumption and time consumption of each behavior in future pending behaviors are fused together to obtain the fused data on the power consumption demand of future pending behaviors. The fused data of the power consumption demand of future events is input into the first prediction model to obtain the corrected power consumption and time consumption of a single future event.

[0010] In conjunction with the first aspect, in another possible implementation, the at least one functional module includes one or more of the following modules: a locking module, an unlocking module, a communication module, a display module, at least one sensor module, and a standby module.

[0011] In conjunction with the first aspect, in yet another possible implementation, the method further includes: Record at least one actual power consumption data of the at least one functional module within the future target period; Compare the at least one actual power consumption data with at least one of the predicted power consumption demands, and calculate the deviation between the two power consumption data. Based on the deviation, the target weight coefficient and / or at least one influencing factor corresponding to each functional module are adaptively adjusted to optimize the first prediction model so that the optimized model improves the prediction accuracy of behavior and power consumption demand.

[0012] Secondly, embodiments of the present invention also provide an electronic device, including a memory and a processor, wherein the memory and the processor are connected; the memory stores computer instructions, and the processor executes the computer instructions to implement the power prediction method of the intelligent device described in the first aspect or any embodiment of the first aspect.

[0013] Thirdly, embodiments of the present invention also provide a computer-readable storage medium storing computer instructions for causing a computer to execute the power prediction method for a smart device described in the first aspect or any corresponding embodiment.

[0014] Fourthly, embodiments of the present invention also provide an intelligent device, including: an electronic device and at least one functional module; wherein, the at least one functional module is used to execute the function corresponding to the functional model and provide the corresponding data to the electronic device in a proactive reporting or passive receiving manner; The electronic device is connected to the at least one functional module, and the electronic device is used to actively or passively acquire the data of the at least one functional module, and execute the power prediction method of the smart device described in the first aspect or any corresponding embodiment.

[0015] Furthermore, embodiments of the present invention also provide a computer program product, including computer instructions, which are used to cause a computer to execute the power prediction method of a smart device according to the first aspect described above or any corresponding embodiment.

[0016] This invention provides a method, device, storage medium, and smart device for predicting the power consumption of a smart device. The method collects historical behavior information and inputs this information into a first prediction model to predict future possible behaviors and power consumption needs. Then, this power consumption demand and the remaining battery power of the smart device are input into a second prediction model to predict the remaining battery life. This method allows users to clearly know the battery status of their smart device in advance, effectively avoiding problems such as door lock malfunctions due to sudden power outages. Furthermore, this method can accurately predict the changing trend of battery life, reducing the number of battery charging cycles and improving the user experience.

[0017] Furthermore, this method uses a deep learning network to predict the remaining battery life of smart locks. Based on the power consumption of the previous period and key influencing factors, it predicts the power demand for the next period. By introducing a prediction model based on multiple key influencing factors, the accuracy and practicality of power demand prediction are improved. This also avoids user anxiety about battery life, such as frequent charging leading to accelerated battery wear, and prevents users from being locked out due to unforeseen battery depletion.

[0018] For smart lock manufacturers, the predicted power consumption can effectively remind users to schedule battery replacements and adjust high-power-consuming usage habits, thereby achieving stable battery usage. Attached Figure Description

[0019] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0020] Figure 1 This is a schematic diagram of the structure of a smart device provided in some embodiments of this application; Figure 2 This is a flowchart illustrating a power prediction method provided in some embodiments of this application; Figure 3 This is a schematic diagram of a power prediction method provided in some embodiments of this application; Figure 4 This is a flowchart illustrating another power prediction method provided in some embodiments of this application; Figure 5 This is a schematic diagram of a process for configuring target weight coefficients provided in some embodiments of this application; Figure 6 This is a schematic diagram of a process for configuring an influence factor provided in some embodiments of this application; Figure 7 This is a flowchart illustrating yet another power prediction method provided in some embodiments of this application; Figure 8 This is a schematic diagram of the structure of an electronic device provided in some embodiments of this application; Figure 9 This is a schematic diagram of the structure of another electronic device provided in some embodiments of this application. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] It is understood that before using the technical solutions disclosed in the various embodiments of the present invention, one should be informed and authorized in accordance with relevant laws and regulations through appropriate means regarding the type, scope of use, and usage scenarios of the personal information involved in the present invention.

[0023] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "at least one" means one or more, and "a plurality of" means two or more, unless otherwise explicitly specified.

[0024] As an optional application scenario of this invention, such as Figure 1 The diagram shown is a structural schematic of a system provided in an embodiment of this application. See also... Figure 1 The system 100 includes: a main control chip 10, a locking module 20, an unlocking module 30, a communication module 40, a display module 50, a sensor module 60, a standby module 70, etc.

[0025] The main control chip 10 is connected to various modules, including the locking module 20, the unlocking module 30, the communication module 40, the display module 50, the sensor module 60, and the standby module 70.

[0026] Specifically, the locking module 20 includes mechanical structures such as a motor, clutch, main locking tongue (slanted tongue), square locking tongue (top and bottom hooks), and lock body.

[0027] The unlocking module 30 includes: a biometric unit, a password input unit, a physical keypad, a card / RFID identification unit, and a wireless near-field unlocking module. For example, it may include a Bluetooth antenna / NFC coil for unlocking via Bluetooth or NFC from a mobile phone.

[0028] Optionally, the biometric unit may include at least a fingerprint recognition module, a face recognition module, and a vein / fingerprint recognition module.

[0029] The communication module 40 includes: a wireless communication chip and antenna, a Wi-Fi module, a Bluetooth module (BLE), a Zigbee / Z-Wave module; a wired communication interface (mainly used for engineering installation and debugging), a Micro-USB / USB-C interface for temporary power supply or data debugging, a serial port (such as TTL), etc.

[0030] The display module 50 includes at least: status indicator lights, display screen, touch color screen, voice module, etc.

[0031] Sensor module 60, or sensor module group, includes at least the following: (1) Basic status sensors, such as the latch position sensor (micro switch / Hall sensor), are used to detect whether the latch is in the pop-out (locked) or retracted (unlocked) state.

[0032] (2) Door status sensor (such as reed switch / Hall sensor) is used to detect whether the door is closed (magnet is close) or open (magnet is far away).

[0033] (3) Security sensors, including: anti-pry detection switch, impact / vibration sensor and / or environmental monitoring sensor. The anti-pry detection switch is triggered when the current panel is forcibly pried open. The environmental monitoring sensor includes, but is not limited to, temperature and humidity sensor, light sensor, etc.

[0034] In addition, the system 100 may also include other sensors, such as human proximity sensors and passive infrared sensors (PIR), but this embodiment does not limit this.

[0035] The standby module 70 includes a power system. This power system further includes a main battery compartment, a backup power interface, and an external power interface. The main battery compartment typically uses 4 or 8 AA (size 5) dry cell batteries to power most of the functional modules. The backup power interface is usually a 9V battery contact used to temporarily power the door when the battery is completely depleted. The external power interface, such as a Micro-USB port, is used for emergency power supply.

[0036] The main control chip 10 can be a microprocessor, such as an MCU. As the "brain" of the entire door lock system, it controls the operation of various modules and issues control commands to each functional module. Generally, in standby mode, the main control chip 10 is in a very low-power sleep mode.

[0037] Optionally, the system 100 may also include a real-time clock, a power management module, etc. Figure 1Not shown in the diagram. The real-time clock is used to maintain accurate time for purposes such as password expiration and log recording. The power management module is used to detect the main battery compartment's power level and report it to the main control chip 10.

[0038] Optionally, the system 100 described above can be a smart device, such as a smart lock or a smart doorbell.

[0039] The technical solutions of the embodiments of this application, the technical problems to be solved, and the beneficial effects that can be achieved will be described in detail below.

[0040] For smart lock users, battery life forecasting is generally divided into two types: "medium-to-long-term" and "short-term" forecasts. Medium-to-long-term forecasts primarily aim to guide energy-efficient use of the lock. Short-term forecasts primarily aim to guide lock replacement planning.

[0041] Due to differences in user habits regarding various smart devices (or smart locks), there is a lack of a universally applicable model or prediction method for forecasting the number of days the lock's battery will last. Furthermore, a single user's battery usage is also influenced by various factors, such as seasons, holidays, and weekend effects. This results in complex variations in power consumption, making accurate prediction challenging.

[0042] Currently, the battery prediction methods used are generally linear predictions based on one or more of the above combined factors. These methods have short prediction times and low accuracy. Some methods require complex modeling, resulting in high resource consumption and making it difficult for smart lock manufacturers to implement and promote them in engineering.

[0043] To overcome the aforementioned technical problems, this invention provides a method for predicting the battery life of smart devices. This method predicts the number of days the battery in a smart device's door lock can last. The prediction uses historical time-series data as the data source and employs data mining, deep learning, and other technologies to establish a battery consumption demand prediction model. This model is then used to predict the battery demand or consumption for a future period, generating and feeding back the prediction results. This allows users to manage their batteries scientifically and effectively, reducing battery waste and the number of charging cycles.

[0044] According to an embodiment of the present invention, an embodiment of a power prediction method is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0045] This embodiment provides a method for predicting the power consumption of a smart device, which may be a smart lock or a smart doorbell. Furthermore, this method can be executed by a main control chip 10 located in the smart lock or smart doorbell.

[0046] Figure 2 This is a flowchart illustrating a power prediction method provided in an embodiment of this application. The method includes: Step S101: Obtain historical behavior information collected by at least one functional module of the smart device.

[0047] The historical behavior information records the user's historical usage and power consumption of smart devices. Furthermore, this usage includes events such as opening, closing, locking, and activating the lock. Power consumption can be understood as the amount of battery power required after opening, closing, locking, and / or activating the lock. One example of historical behavior information is shown in Table 1 below.

[0048] Table 1

[0049] At least one functional module includes, but is not limited to, the aforementioned locking module 20, unlocking module 30, communication module 40, display module 50, sensor module 60, standby module 70, etc. These functional modules are used to collect and record historical behavior data, and the main control chip 10 analyzes these historical behavior data to obtain the power consumption corresponding to the historical behavior recorded by each functional module.

[0050] In this step, the implementation method for obtaining the aforementioned historical behavior information is as follows: the main control chip 10 actively and periodically sends request instructions to each functional module, requesting each functional module to report the historical behavior data collected within a specified period to the main control chip 10. Alternatively, a passive acquisition method can be used, such as periodically receiving historical behavior data actively reported by each functional module. This historical behavior data can serve as historical behavior information to provide a basis for subsequent processing.

[0051] Step S102: Input historical behavior information into the first prediction model to predict the future behavior of at least one of the functional modules and its power consumption requirements.

[0052] The first prediction model is a network model, specifically a Long Short-Term Memory (LSTM) network model. LSTM is an improved core model of Recurrent Neural Networks (RNNs); LSTM models can effectively capture long-distance dependencies in sequence data. In this embodiment, the first prediction model can be a trained LSTM model.

[0053] Please see Figure 2 In this step, after obtaining the first prediction model, historical behavior information sampled online is input into the first prediction model. This model is then used to predict future behaviors and the potential power consumption of each behavior, thus obtaining the power demand corresponding to the upcoming behaviors. In summary, all the content output by the first prediction model can be considered the prediction result. This prediction result includes a series of future behaviors, the number of times each behavior occurs, the power consumption of each behavior, and the duration of each occurrence.

[0054] Step S103: Obtain the remaining battery power on the smart device.

[0055] The remaining battery power status can be detected and reported to the main control chip via the power management module. The battery management module is used to detect and collect battery power data, and to report the remaining battery power status so that the main control chip can obtain the battery's remaining power status in real time. For example, the power management module might report a battery level of 80% and the current battery power value to the main control chip.

[0056] In this embodiment, the battery on the smart device can be understood as the main battery compartment in the power system. The remaining power of the battery can be understood as the remaining power of the main battery compartment.

[0057] It should be noted that step S103 and the aforementioned steps S101 and S102 can be executed in parallel or sequentially. Although in this embodiment, obtaining the remaining battery power on the smart device is step S103, the actual execution order can be before or simultaneously with S101 or S102. This embodiment does not restrict this.

[0058] Step S104: Input the remaining battery power and power consumption requirements into the second prediction model to determine the estimated operating time of the remaining battery power.

[0059] Specifically, the "remaining power status" in step S103 and the "power consumption demand" in step S102 are input into the second prediction model for prediction.

[0060] The second prediction model is a battery power prediction model. Optionally, this second prediction model is a battery power prediction model based on LASSO regression. It is trained using total battery power demand, number of days of use, and ambient temperature as input features, and real-time remaining battery power as the output result. During training, the training set and test set are divided in a 4:1 ratio. After training, a prediction model for the battery is obtained. This model is then used to predict future total battery power demand and power changes under temperature, and finally calculates the estimated runtime of the remaining battery power.

[0061] The power prediction method provided in this embodiment collects historical behavior information and inputs it into a first prediction model to predict future possible behaviors and power consumption needs. Then, this power consumption need and the remaining battery power of the smart device are input into a second prediction model to predict the remaining battery life. This method allows users to clearly know the battery status of their smart devices in advance, effectively avoiding problems such as door lock malfunctions due to sudden power outages. Furthermore, this method can accurately predict the changing trend of battery life, reducing the number of battery charging cycles and improving the user experience.

[0062] Optionally, in one specific implementation of this embodiment, such as Figure 4 As shown, step S101 above: obtaining historical behavior information collected by at least one functional module of the smart device, specifically includes: Step S101-1: Receive at least one behavioral data reported by at least one functional module of the smart device within a preset period.

[0063] Each behavioral data point includes: type, frequency, power consumption, and duration of each behavior. Specifically, the types of behaviors include: normal standby; entering / exiting; ringing the doorbell; active wake-up / false wake-up; fingerprint unlocking and automatic locking; palm vein unlocking and internal locking; password unlocking and automatic locking; card unlocking and automatic locking; face unlocking and automatic locking; remote door opening and automatic locking; internal door knob unlocking and automatic locking; internal door knob unlocking and external touch lock; automatic continuous snapshot when the peephole is unattended; image transmission when unattended; short-term stay snapshot; 60-second stay snapshot; and peephole message push.

[0064] Step S101-2: Process at least one behavioral data within a preset period to generate historical behavioral information.

[0065] In this embodiment, historical behavior information can be obtained through the following two methods: The first method: the direct method.

[0066] The controller can obtain behavioral records through the door lock log, such as when and what user action was recorded (i.e., the action type), the duration of the action, and power consumption. Power consumption can be directly measured using instruments such as ammeters and fuel meters. The testing method can be to measure and record each detected action, generating a log record; or, after multiple actions occur within a period of time, the test results can be statistically analyzed to generate multiple log records.

[0067] The second method: the indirect method.

[0068] The controller can track historical actions, such as the number of unlocking / unlocking events detected in a day and the total duration. Optionally, if the total number of unlocking / unlocking events in a day is 10, with a total duration of 10 seconds, and the doorbell rings 5 ​​times with a total duration of 5 seconds, the total power consumption for the day is 30 mAh. This can be calculated based on the theoretical power consumption ratio of each module. For example, the calculated power consumption might be: 10 / 15 × 30 = 20 mAh; 5 / 15 × 30 = 10 mAh.

[0069] In addition, it also includes: calculating the electricity required for the day based on the above data, the calculation formula is: Q=H×i / 3600, or Q=n×h×i / 3600, where i represents the average current, h represents the unit time, H represents the total time, and n represents the number of actions.

[0070] Furthermore, the above method includes a process of configuring weighting coefficients in the second prediction model before inputting the battery's remaining power and power consumption demand into the second prediction model. Specifically, as follows: Figure 5 As shown, the method includes: Step S201: Configure the target weight coefficient for each functional module based on historical behavior information.

[0071] The weighting coefficients represent the impact of each functional module on the overall battery power consumption. The target weighting coefficient can be understood as the final (used) weighting coefficient after adjustments or corrections to the initial weighting coefficients. Essentially, the sum of all target weighting coefficients for each functional module equals 1.

[0072] Step S202: Based on the target weight coefficient of each functional module, merge it with the data of the future behaviors of each functional module to obtain the fused data of the future behaviors.

[0073] Specifically, in step S201, target weight coefficients are configured for each functional module based on historical behavior information, which can quantify different functional modules (such as...). Figure 1 The differences in power consumption percentage and usage frequency of each functional module (as shown) during actual operation are analyzed. Step S202 embeds weighting coefficients into the second prediction model. This allows the second prediction model to no longer indiscriminately weight the power consumption data of each functional module when calculating the total power consumption demand. Instead, it highlights the dominant influence of high-priority, high-power-consuming modules on power consumption based on the weighting coefficients, thereby correcting the prediction logic of the second prediction model. This avoids prediction deviations caused by the average calculation of the power consumption percentage of each functional module, effectively reduces the prediction error of the estimated remaining battery power operating time, and improves the consistency between the prediction results and the actual usage time.

[0074] Further, step S201: One specific implementation of configuring the target weight coefficient for each functional module is: based on the historical behavior information of at least one functional module, calculate the expected energy consumption ratio of each functional module in the future preset period, and use the expected energy consumption ratio as the target weight coefficient.

[0075] For example, suppose the system includes the following functional modules: unlocking module, communication module, display module, and standby module. The system sets target weight coefficients for these functional modules. During the model training phase, the daily power consumption of each functional module is calculated based on historical behavior information. Suppose that on a certain historical day, the power consumption of the unlocking module, communication module, display module, and standby module were recorded as 6mAh, 40mAh, 4mAh, and 0.48mAh, respectively. Then, the total power consumption Q for that day can be calculated. 总 Q 总 = 6 + 40 + 4 + 0.48 = 50.48 (mAh / day).

[0076] One method for determining the target weight coefficient is to calculate the proportion of daily power consumption of each functional module to the total power consumption. In this example, based on Q... 总 The weight coefficients for these four modules are calculated as follows: The target weight coefficient (W1) corresponding to the unlocking module is approximately 0.119, where W1 = 6 / 50.48 ≈ 0.119. The target weight coefficient (W2) corresponding to the communication module is approximately 0.792, calculated as W2 = 40 / 50.48. The target weight coefficient (W3) corresponding to the display module is approximately 0.079, where W3 = 4 / 50.48 ≈ 0.079. The target weight coefficient (W4) for the standby module is W4 = 0.48 / 50.48 ≈ 0.01.

[0077] Optionally, the method also includes: verifying the target weight coefficients. The specific process is as follows: input the above target weight coefficients into the second prediction model, compare the model's predicted run time with the actual run time, and if the error exceeds 5%, iteratively update the weight coefficients based on the latest user behavior data.

[0078] In addition, in other embodiments, target weight coefficients can be configured for each functional module in other ways, such as by a preset algorithm or by setting based on other historical behavioral data. This embodiment does not limit this.

[0079] In this embodiment, a specific implementation of step S202 includes: firstly, collecting core data of each functional module corresponding to the future behavior to be performed, including the estimated number of unlocking attempts and basic power consumption per unlocking attempt for the unlocking module, the estimated data transmission frequency and basic power consumption per transmission attempt for the communication module, the estimated screen-on time and basic power consumption per unit time for the display module, and the estimated standby time and basic power consumption per unit time for the standby module. Then, using a weighted fusion algorithm, the core data of each functional module is multiplied by its corresponding target weight coefficient to obtain weighted data of the future behavior to be performed for each module. Finally, the weighted data of all modules are aggregated to generate fused data representing the overall power consumption characteristics of the future behavior to be performed.

[0080] For example, if the target weight coefficient of the unlocking module is 0.0021, and it is expected to unlock 10 times in the future with a basic power consumption of 20W per unlock, then... s, the target weight coefficient of the communication module is 0.0266, the estimated 50 transmissions are made and the basic power consumption per transmission is 50W. s, then first calculate the weighted data of the two modules as 0.0021×(10×20W) s), 0.0266×(50×50W Then, the weighted data from all modules is aggregated to obtain fused data, which provides an integrated input basis for subsequent power consumption prediction.

[0081] In some possible implementations, such as Figure 6 As shown, before step S102 of the above method: predicting the future behavior and its power consumption demand using the first prediction model, the method further includes: Step S301: Configure at least one influencing factor.

[0082] The at least one influencing factor is used to characterize one or more of the following: environmental conditions, hardware and software status, usage behavior, and battery status of the smart device. Specifically, the content of the at least one influencing factor includes: environmental conditions of the door lock (weather, humidity, electromagnetic interference (other electronic devices in the environment, WiFi strength, base station affects background power consumption)), door lock device status (door sagging, hardware aging or failure, software consumption (such as background processes, system updates, etc.)), door lock usage behavior (battery aging caused by frequency of actions, charging habits, etc.), and door lock battery status (battery degradation, health, calibration deviation, self-discharge), etc.

[0083] The effects of various influencing factors on the power consumption and time of a single action include the following four points: (1) Environmental conditions as influencing factors Weather (Temperature): In low-temperature environments, battery activity decreases and internal resistance increases. To drive the unlocking motor and other actuators to complete a single action, a higher voltage needs to be output to compensate for internal resistance losses, directly increasing power consumption per action. Simultaneously, low temperatures reduce the lubrication of motor mechanical components, increasing the resistance of the latch extension and retraction, prolonging the time required for a single unlocking action, further amplifying power consumption. High-temperature environments accelerate battery self-discharge, increasing static power consumption during standby and potentially triggering the device's thermal protection mechanism, reducing the communication module's transmission power and extending the time required for data transmission.

[0084] Humidity: High humidity can cause slight moisture in the internal circuitry of the door lock, increasing leakage current loss and thus increasing the basic power consumption per operation. If the humidity exceeds the standard and causes corrosion of the metal parts of the lock body, it will increase the frictional resistance of the lock tongue movement, prolong the time for each unlocking and locking operation, and thus increase the power consumption of the motor for continuous operation.

[0085] Electromagnetic interference (electronic devices, WiFi, base stations): Strong electromagnetic interference in the surrounding environment can interfere with the signal transmission stability of communication modules. A single data upload / command reception may require multiple retransmissions to complete, significantly extending the time of a single communication operation. At the same time, the communication module is constantly in a high-power wake-up state, increasing power consumption accordingly. Severe interference may also falsely trigger the sensing module, causing a single standby operation to be frequently woken up, resulting in additional unnecessary power consumption.

[0086] (2) Door lock device status influencing factors Door sagging: Door sagging will cause the bolt and latch to misalign. When unlocking, the motor needs to output more torque to overcome the jamming, which greatly increases power consumption. In addition, the bolt positioning detection time is extended, and the time taken for each action is increased. In extreme cases, "over-rotation" will occur, triggering overload protection and further increasing power consumption.

[0087] Hardware aging or failure: Hardware aging issues such as worn motor bearings and increased gear meshing clearance will reduce power transmission efficiency, requiring the motor to work continuously for a longer time for each unlocking, increasing both time and power consumption; sensor aging will lead to a decrease in detection accuracy, requiring multiple sampling confirmations for a single induction wake-up action, increasing time and power consumption; circuit faults (such as capacitor leakage) will directly increase the base power consumption of a single action.

[0088] Software consumption (background, system update): If a single unlocking action is accompanied by software tasks such as background log uploading and status synchronization, the processor and communication module will be activated additionally, increasing the total power consumption of a single action; a single device wake-up during a system update requires loading the update package and verifying the data, which takes much longer than a regular wake-up, and the processor runs at full load, resulting in a significant increase in power consumption.

[0089] (3) Door lock usage behavior influencing factors Frequency of operation: High-frequency unlocking / locking will accelerate battery polarization, increase the instantaneous discharge current of a single operation, reduce the battery energy conversion efficiency, and gradually increase the power consumption per operation; at the same time, high-frequency operation will increase the motor temperature, increase heat loss, and further increase the power consumption per operation.

[0090] Charging habits: Overcharging will cause battery gas evolution and capacity decay, reducing the effective output power of a single action and requiring more battery capacity to complete the same action; undercharging will keep the battery in a low voltage state for a long time, resulting in insufficient starting voltage for a single action, longer motor start-up time, and increased power consumption.

[0091] (4) Factors affecting door lock battery status Battery degradation: As the number of battery cycles increases, capacity decreases, internal resistance increases, energy loss per operation increases, and power consumption increases. At the same time, the voltage drop rate of a degraded battery is faster, which may lead to insufficient voltage in the middle of a single operation, triggering the retry mechanism, prolonging the time and adding to the power consumption.

[0092] Battery health: Batteries with low health have problems such as low charging and discharging efficiency and high self-discharge rate, and the static power consumption of a single standby operation is significantly increased; in addition, insufficient health will lead to unstable motor starting torque, a decrease in the success rate of a single unlocking action, an increase in the number of retries, and a simultaneous increase in time and power consumption.

[0093] Calibration deviation: Battery power calibration deviation can cause the device to misjudge the remaining power. To ensure the stability of a single action, the system will reserve more redundant power output, increasing the actual power consumption of a single action. If the current calibration is inaccurate, it will also lead to inaccurate power consumption monitoring, making it impossible to adjust the module power in time, and exacerbating the power waste of a single action.

[0094] Self-discharge: When the battery self-discharge rate is high, even if the device is in standby mode, the static power consumption of a single standby action will increase; and self-discharge will cause the battery voltage to drop slowly, requiring voltage compensation before each action is started, which will increase power consumption.

[0095] Step S302: Based on the influencing factors of each behavior, merge the power consumption and time consumption data of each behavior in the future to be performed to obtain the fused data of the power consumption demand of the future to be performed behavior.

[0096] Step S303: Input the fused data of the power consumption demand of future behaviors into the first prediction model to obtain the corrected power consumption and time consumption of a single future behavior.

[0097] One possible implementation is to assess the power consumption and time consumption of future actions. Assume individual actions are represented as x1, x2, x3, ... x nThe power consumption of a single action is q. i The time for a single action is t. i The actual power consumption is Q. 总 If the number of times is n, then the actual power consumption is predicted. .

[0098] Incorporate an impact factor, assuming it corresponds to x1, x2, x3, ..., x. n The influencing factors are k1, k2, k3, ..., k n Based on the influencing factors of each behavior, the estimated cost per action is [calculated as follows]. The estimated time for a single action is to be corrected. ,but , And thus obtain ; After weighted calculation, the proportion of a single action in the total consumption. When performing a weighted summation, without considering influencing factors, the total power consumption of all future actions is calculated as follows: Taking into account influencing factors, the total power consumption corresponding to all future actions is calculated as follows: .

[0099] Based on the above embodiments, this example uses the two core behaviors of a smart door lock, namely "unlocking behavior" and "communication behavior," to illustrate how the formula is used to combine influencing factor configuration with power consumption prediction. Specifically, step S301: Configure influencing factors. For unlocking and communication behaviors, select two types of influencing factors and assign values: for example, configure the unlocking behavior influencing factor, and due to door sagging (equipment status), set the correction coefficient k1=1.2 (consumption correction). =1.3 (Time Consumption Correction). Configure communication behavior impact factors: Due to surrounding electromagnetic interference (environmental conditions), set correction coefficient k2=1.4 (Consumption Correction). =1.5 (time consumption correction).

[0100] Step S302: Integrate influencing factors and behavioral data. Basic data for two types of behavior are known: Unlocking behavior: estimated daily frequency n1=10, basic consumption per transaction q1=20W·s, basic time per transaction t1=2s. Communication behavior: estimated daily frequency n2=50, basic consumption per transaction q2=5W·s, basic time per transaction t2=10s.

[0101] Based on the corrected data from the impact factor calculation, the following was obtained: Power consumption during unlocking: Duration ; Power consumption of communication activities: Duration .

[0102] Then combine the behavior weights (unlock weights) Communication weight Substituting the data, we obtain the fused data, which is the total power consumption Q'. 总 :

[0103] Step S303: Input model and output correction results. Input the above fused data into the first prediction model. Based on the fitting relationship of historical training, the first prediction model outputs the corrected single behavior data: the power consumption of a single unlocking behavior is 24W·s, and the time is 2.6s; the power consumption of a single communication behavior is 7W·s, and the time is 15s.

[0104] In this embodiment, by adding at least one influencing factor to the prediction model, these factors cover key dimensions such as the door lock's environmental conditions, device status, usage behavior, and battery status. After incorporating these factors into the first prediction model, the model no longer relies solely on the basic power consumption data of functional modules for calculation. Instead, it incorporates interference factors in the actual scenario (such as increased hardware power consumption due to high humidity, decreased actual output power caused by battery degradation, and additional power consumption of the communication module due to electromagnetic interference) to correct the prediction logic. By fitting a regression multinomial to quantify the influence weight of each factor on power consumption, prediction bias caused by environmental fluctuations, device aging, and differences in usage habits can be effectively reduced, making the predicted power consumption for a single action and the total power consumption more closely match the actual situation. This avoids user anxiety about battery life, such as frequent charging leading to accelerated battery wear, and also prevents users from being locked out due to unforeseen circumstances where the battery runs out.

[0105] Furthermore, different usage scenarios and their influencing factors (such as high humidity in rainy weather, latch jamming caused by door sagging, and frequent door opening behavior) will have varying impacts on door lock power consumption. Incorporating these factors into the model allows the first prediction model to adapt to diverse real-world application scenarios, avoiding insufficient model generalization due to training in a single scenario.

[0106] Meanwhile, through the synergistic correction of multi-dimensional factors, even if a certain factor experiences abnormal fluctuations (such as sudden electromagnetic interference), the model can maintain prediction stability through compensation calculations of other factors, thereby improving the model's anti-interference capability.

[0107] Furthermore, in another embodiment, the method of the above embodiment, after determining the expected runnable time of the remaining battery power in step S104, further includes a method flow for optimizing the first prediction model.

[0108] Specifically, such as Figure 7 As shown, the method includes: Step S105: Record at least one actual power consumption data of at least one functional module within the future target period.

[0109] Step S106: Compare at least one actual power consumption data with at least one of the predicted power consumption requirements, and calculate the deviation between the two power consumption data.

[0110] Step S107: Based on the deviation, adaptively adjust the target weight coefficient and / or at least one influencing factor corresponding to each functional module to optimize the first prediction model so that the optimized model improves the prediction accuracy of behavior and power consumption demand.

[0111] Specifically, this embodiment takes the unlocking module, communication module, display module, and standby module of a smart door lock as the objects and performs a model adaptive optimization process, as follows: According to step S105: Record the actual power consumption data within the future target period.

[0112] For example, if the target period is set to the next 7 days, the power management module of the smart door lock collects and records the actual power consumption data of the four functional modules in real time, with a sampling frequency of once per hour.

[0113] Unlocking module: 80 door openings within 7 days, actual total power consumption Q 开实 =180W·s; Communication module: Completed 350 data uploads / command receptions within 7 days, with actual total power consumption Q. 通实 =2300 W·s; Display module: Cumulative screen-on time of 120 minutes over 7 days, actual total power consumption Q 显实 =45000 W·s; Standby module: Actual total power consumption Q during continuous operation for 7 days 待实 =41000W·s.

[0114] At the same time, record the environmental and equipment status data within the target period (such as daily average humidity of 65% and door jamming twice).

[0115] According to step S106: compare the actual power consumption with the predicted power consumption, and calculate the deviation. Specifically, retrieve the predicted power consumption from the four modules output by the first prediction model before the start of the target period: Predicted power consumption Q for unlocking module 开预 =200W·s, deviation ΔQ 开 =|180-200| / 200=10%; Predicted power consumption Q of communication module 通预=2500W·s, deviation ΔQ 通 =|2300-2500| / 2500=8%; Predicted power consumption Q of the display module 显预 =48000 W·s, deviation ΔQ 显 =|45000-48000| / 48000=6.25%; Standby module predicted power consumption Q 待预 =43200W·s, deviation ΔQ 待 =|41000-43200| / 43200≈5.1%.

[0116] A deviation threshold of 5% is set. If the power consumption deviation of the unlocking module and communication module exceeds the threshold, model optimization needs to be triggered. According to step S107: the weight coefficients and influencing factors are adaptively adjusted based on the deviation to optimize the model. Specifically, the target weight coefficients are adjusted, and the weights of each module are corrected in reverse based on the deviation. The larger the deviation, the greater the weight adjustment. The adjustment formula is as follows: W i新 =W i旧 ×(1-(ΔQ i -ΔQ 阈值 ) / maxQE') Substitute the original weights (W) 开旧 =0.0021、W 通旧 =0.0266) and the deviation data, we can calculate: Unlock module new weight W 开新 =0.0021×(1-(10%-5%) / 10%)=0.00105; New weight W for communication module 通新 =0.0266×(1-{8%-5%} / 10%)=0.01862; If the deviation of the display module and the standby module does not exceed the threshold, the weight remains unchanged.

[0117] Then, the influencing factors were adjusted. The reasons for the deviations of the unlocking module and communication module were analyzed: the humidity was high during the target period, which led to increased resistance of the unlocking motor, signal attenuation of the communication module, and reduced additional power consumption (opposite to the predicted trend).

[0118] Specifically, regarding the environmental humidity influencing factor k 湿 The influence weight of this factor in the model was reduced, and the correction coefficient was adjusted from 0.15 to 0.08; the influence factor k of the door jamming was also adjusted. 卡 Since occasional stuttering did not significantly increase power consumption, its correction factor was adjusted from 0.2 to 0.12.

[0119] Finally, the adjusted weighting coefficients and influencing factors were input into the first prediction model, and the data from the next 7 days were used for verification. The results showed that the prediction deviations of power consumption of the unlocking module and the communication module were reduced to 4.2% and 3.8% respectively, both below the threshold of 5%, and the model prediction accuracy was significantly improved.

[0120] In this embodiment, step S105 records the actual power consumption data of the functional modules within the future target period, providing a realistic scenario-based data benchmark for model optimization; step S106 accurately locates the source of error in the model prediction by comparing the deviation between the actual value and the predicted value; step S107 adaptively adjusts the target weight coefficients and influencing factors of the functional modules based on the deviation, enabling the model to iteratively optimize according to dynamic factors such as the actual usage status of the device, environmental changes, and battery degradation. Compared with a prediction model with fixed parameters, this process can continuously reduce the deviation between predicted power consumption and actual power consumption, ensuring that the model maintains high prediction accuracy over a long period of time, and improving the predictive model's adaptability to complex scenarios.

[0121] In another embodiment, the power prediction method provided in this embodiment specifically includes: First, the construction and training process of the first prediction model: the first prediction model is built based on the LSTM algorithm and its prediction ability is learned.

[0122] In this embodiment, the LSTM algorithm is used to adapt to time series data (the pattern of user behavior and power consumption over time), and the core objective is to predict the number of future behaviors.

[0123] Specifically, in this embodiment, the model is trained to learn the patterns of behavior by using "historical behavior dataset + date information" as input. For example, based on the number of doorbell rings from January 1st to January 4th, the model predicts the number of doorbell rings on January 5th. This is subsequently extended to iteratively predict the daily occurrence frequency of each type of behavior within a certain future period (e.g., 30 days). As shown in Table 2 below, Table 2

[0124] Optionally, the first prediction model may also include at least one configurable parameter. Different lock models have different configurations, and the configuration parameters include at least one of the following: model, battery type, and battery capacity, as shown in Table 3 below.

[0125] Table 3

[0126] The second step is to predict future behaviors and power consumption based on historical behavior information and the trained first prediction model.

[0127] Specifically, the method for calculating future power consumption demand includes: combining the "predicted number of times the behavior occurs" and the "power consumption per behavior" (such as the power consumption per fingerprint unlock and the power consumption of WiFi connection), and calculating the total daily and periodic power consumption demand in the future according to the following formula: Q=n×h×I, where I=αT.

[0128] Where Q is the power demand (or power consumption), n is the number of actions, h is the duration of a single action, I is the current, and T is the temperature. α is the correlation coefficient between temperature and current.

[0129] It should be understood that α can also be a parameter used to quantify the relationship between ambient temperature (T) and the average current (I) of a certain type of electrical behavior of the smart door lock. α can be understood as the proportionality coefficient or correction coefficient of temperature affecting current.

[0130] The third step is to input the remaining battery power and the aforementioned power consumption requirements into the second prediction model to determine the estimated operating time of the remaining battery power.

[0131] Specifically, this embodiment introduces the battery degradation factor and calculates it according to the formula: Battery degradation Q = βT + Δbehavior + ∑cycle number. The remaining battery capacity can then be calculated as: Remaining battery capacity = Total battery capacity - Total future power consumption demand - Battery degradation. Finally, the number of days the remaining battery capacity is available is calculated or predicted: Available days = Remaining capacity ÷ Average daily power consumption demand, outputting the core prediction result.

[0132] In addition, the above methods also include: optimizing the accuracy of the prediction model so that the prediction results continue to closely approximate the true values.

[0133] One implementation involves collecting the "actual usage days" (e.g., 90 days for the first charge, 85 days for the second charge) each time the device is actually charged. This true value is then used to update the model parameters and correct the prediction logic. Through multiple cycles of "prediction - actual verification - parameter update," the predicted value gradually approaches the true value, for example, from a theoretical 100 days to a final actual 80 days, thereby improving prediction accuracy.

[0134] The method provided in this embodiment can predict battery power in advance and avoid the risk of power outages. Specifically, by accurately predicting the number of days the lock can be used, users can know in advance the usage time corresponding to the remaining battery power of the door lock, eliminating the need for frequent manual checks of the battery power. This also effectively avoids the security risks of the door lock being unable to unlock / lock due to sudden power outages, thus improving user safety.

[0135] Furthermore, this solution is adaptable to diverse scenarios and meets different needs. It can provide personalized predictions for different door lock models, different operating modes (fully automatic / semi-automatic), different environmental factors (high / low temperature), and different user behaviors (such as high-frequency snapshots and low-frequency unlocking). Whether for home users or commercial scenarios (such as office buildings and apartments), users can obtain power management references tailored to their specific needs, improving the solution's applicability and user experience.

[0136] The method described in this embodiment enables the prediction of power consumption demand for the next period based on power consumption data and key influencing factors for a given period.

[0137] Specifically, historical data on power consumption and key influencing factors are collected. Power consumption-related data is collected from the smart lock's historical data, including power consumption data (such as actual daily / hourly power consumption) and key influencing factor data (combined with the actual operating conditions of the lock, including user behavior data: number of times the door is opened, frequency of unlocking methods, number of snapshots, etc.; environmental parameter data: temperature, humidity, and other meteorological information; lock hardware parameter data: standby / wake-up power consumption, WiFi / Bluetooth connection power consumption, abnormal bolt rotation power consumption, etc.; and battery status data: number of charging cycles, degree of cycle degradation, etc.).

[0138] Arrange the two types of data in chronological order to construct a time series sample dataset; normalize the data (e.g., map data of different magnitudes such as power consumption and temperature to the [0,1] interval) to eliminate the interference of the difference in scale on model training; to ensure the model training and validation effect, split the preprocessed dataset into a training set (for model parameter fitting) and a test set (for model generalization ability validation) according to a preset ratio (e.g., 7:3 or 8:2).

[0139] A first prediction model (LSTM network model) for predicting electricity demand is established. At least one influencing factor is added to this model to decompose electricity consumption into basic electricity consumption and meteorologically sensitive electricity consumption, and the network's internal parameters are adjusted. Then, the normalized training set data is input into the LSTM network model for training. During training, momentum is introduced to dynamically modify the learning rate of each parameter, thus fitting the LSTM network model. The predicted and test data are combined, and the size of the test dataset is adjusted using the expected electricity consumption. This dataset is then input into the fitted LSTM network model. The training and test loss data are compared and displayed to determine if the model is overfitting.

[0140] After the LSTM network model passes evaluation and verification (meets the error threshold and has no overfitting), the online operating data of the smart lock is collected in real time (such as the current temperature, the number of times the door has been opened, the remaining battery capacity, etc.). The data is standardized and then input into the verified LSTM network to output the predicted power consumption value for a specified period in the future (such as the next day or the next week). Combined with the "available days derivation logic" in the patent, the accurate prediction of the available days of the smart lock is finally realized, providing users with a reference for power management.

[0141] Furthermore, this embodiment also provides a power prediction device for implementing the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0142] The power prediction device may include multiple modules for implementing the power prediction method described in the foregoing embodiments. Furthermore, the power prediction device is presented in the form of functional units, where a unit refers to an ASIC circuit, a processor and memory executing one or more software or fixed programs, and / or other devices that can provide the aforementioned functions.

[0143] Specifically, the further functional descriptions of each module and unit in the power prediction device are the same as those in the corresponding embodiments described above, and will not be repeated here.

[0144] Figure 8 A schematic diagram of an electronic device provided in an embodiment of the present invention. The electronic device may be as described above. Figure 1 The system described in the illustrated embodiment.

[0145] The following is a detailed reference. Figure 8 This diagram illustrates a structural schematic suitable for implementing an electronic device according to embodiments of the present invention. The electronic device may include a processor 810 (e.g., a central processing unit), which can perform various appropriate actions and processes based on a program stored in memory 820 or a program loaded from memory 820 into random access memory (RAM). The RAM also stores various programs and data required for the operation of the electronic device. The processor 810 and memory 820 are interconnected via a bus. Input / output (I / O) interfaces are also connected to the bus.

[0146] Optionally, in this embodiment, the processor 810 and the memory 820 can be used to implement the above-described... Figure 1 The main control chip 10 shown performs the functions described above. Figures 2 to 7 The power prediction method described above.

[0147] In addition, the electronic device also includes a power management module 830, which includes a battery or battery pack. The power management module 830 is connected to the processor 810 and the memory 820 via a bus, and is used to detect the current power level of the smart lock in real time, calculate the remaining power level, and report the data to the processor 810.

[0148] In addition, the electronic device may include more or fewer other modules, such as connecting the aforementioned functional modules via at least one interface, for example... Figure 1 As shown, it is connected to the locking module 20, unlocking module 30, communication module 40, display module 50, sensor module 60, etc.

[0149] It should be noted that in this embodiment, the electronic device may not include the power management module 830, such as... Figure 9 As shown, the power management module 830 is located outside the electronic device. In this case, the electronic device can be, as in the aforementioned embodiments, Figure 1 The main control chip 10 is shown.

[0150] This invention also provides a computer-readable storage medium in which the methods described in this invention can be implemented in hardware or firmware, or implemented as recordable on a storage medium, or implemented as computer code originally stored on a remote storage medium or a non-transitory machine-readable storage medium and to be stored on a local storage medium after being downloaded via a network, so that the methods described herein can be stored on such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware.

[0151] The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; furthermore, the storage medium can also include combinations of the above types of memory. It is understood that a computer, processor, microprocessor controller, or programmable hardware includes storage components capable of storing or receiving software or computer code, which, when accessed and executed by the computer, processor, or hardware, implements the methods shown in the above embodiments.

[0152] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and all such modifications and variations fall within the scope defined by the appended claims.

Claims

1. A method for predicting the power consumption of a smart device, characterized in that, Applied to smart locks or smart doorbells, the method includes: Obtain historical behavior information of at least one functional module of the smart device; The historical behavior information is input into the first prediction model to predict the future behavior of at least one of the functional modules and its power consumption requirements. Obtain the remaining battery power on the smart device; The remaining battery power and the power consumption demand are input into the second prediction model to determine the estimated operating time of the remaining battery power.

2. The method according to claim 1, characterized in that, The acquisition of historical behavior information collected by at least one functional module of the smart device includes: The device receives at least one behavioral data reported by at least one functional module of the smart device within a preset period, and each behavioral data includes: the type, number, power consumption and time consumption of a single behavior; The at least one behavioral data within the preset period is processed to generate the historical behavioral information.

3. The method according to claim 1, characterized in that, Before inputting the remaining battery power and power consumption demand into the second prediction model, the following steps are also included: Configure the target weight coefficient for each functional module based on historical behavior information; Based on the target weight coefficient of each functional module, the data of the future behavior to be performed of each functional module are fused together to obtain the fused data of the future behavior to be performed.

4. The method according to claim 3, characterized in that, The step of configuring the target weight coefficient for each functional module based on historical behavior information includes: Based on the historical behavior information of the at least one functional module, the expected energy consumption ratio of each functional module in the future preset period is calculated, and the expected energy consumption ratio is used as the target weight coefficient.

5. The method according to claim 1, characterized in that, Before using the first prediction model to predict future events and their power consumption, the method further includes: Configure at least one influencing factor, which is used to characterize one or more of the following: environmental conditions, hardware and software status of the smart device, usage behavior, and battery status. Based on the influencing factors of each behavior, the data on the power consumption and time consumption of each behavior in future pending behaviors are fused together to obtain the fused data on the power consumption demand of future pending behaviors. The fused data of the power consumption demand of future events is input into the first prediction model to obtain the corrected power consumption and time consumption of a single future event.

6. The method according to any one of claims 1 to 5, characterized in that, The at least one functional module includes one or more of the following modules: The system includes a locking module, an unlocking module, a communication module, a display module, at least one sensor module, and a standby module.

7. The method according to any one of claims 1 to 5, characterized in that, The method further includes: Record at least one actual power consumption data of the at least one functional module within the future target period; Compare the at least one actual power consumption data with at least one of the predicted power consumption demands, and calculate the deviation between the two power consumption data. Based on the deviation, the target weight coefficient and / or at least one influencing factor corresponding to each functional module are adaptively adjusted to optimize the first prediction model so that the optimized model improves the prediction accuracy of behavior and power consumption demand.

8. An electronic device, characterized in that, include: A memory and a processor, wherein the memory and the processor are connected; The memory stores computer instructions, and the processor executes the computer instructions to perform the power prediction method for the smart device according to any one of claims 1 to 7.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing a computer to execute the power prediction method of the smart device according to any one of claims 1 to 7.

10. A smart device, characterized in that, include: Electronic equipment and at least one functional module; The at least one functional module is used to execute the function corresponding to the functional model and provide the corresponding data to the electronic device in the form of active reporting or passive reception. The electronic device is connected to the at least one functional module, and the electronic device is used to acquire the data of the at least one functional module in an active or passive manner, and to execute the power prediction method of the smart device as described in any one of claims 1 to 7.

Images