An intelligent power distribution system suitable for a variety of home appliances

Abstract

The application discloses a kind of intelligent power distribution system suitable for multiple household appliances, it is related to intelligent power distribution technical field, to solve the technical problem that traditional household power distribution system relies on fixed threshold value cannot adapt to the characteristics of multiple types of appliances, concurrent interference and extreme working condition of multiple equipment, including: multi-modal collaborative fingerprint acquisition module, for collecting the electrical parameter characteristics and non-electrical parameter characteristics of different operating scenarios of household appliances, based on two types of characteristics to build multi-dimensional associated multi-modal collaborative appliance fingerprint database;AI adaptive threshold control module is in communication connection with multi-modal collaborative fingerprint acquisition module, for obtaining multi-dimensional power consumption data and establishing a user power consumption behavior model that integrates multiple factors, based on model identification different power consumption scenarios and dynamically adjusting the load protection threshold of power distribution system, threshold response priority and new equipment access adaptation threshold.The application has the advantages of reducing the incidence of false tripping and protection not timely, guaranteeing the safe operation of household appliances.

Description

Technical Field

[0001] This invention relates to the field of intelligent power distribution technology, and more specifically, to an intelligent power distribution system applicable to a variety of household appliances. Background Technology

[0002] With the rapid development of the smart home industry, modern home appliances are characterized by "diversification of types, complexity of scenarios, and frequent access." They not only cover devices with different characteristics such as resistive loads (e.g., rice cookers, electric kettles), inductive loads (e.g., air conditioners, washing machines), and capacitive loads (e.g., LED lights, smart charging piles), but also often involve multiple devices operating concurrently (e.g., microwave ovens, range hoods, TVs, and humidifiers working simultaneously during dinner time), frequent access of new devices (e.g., adding a robot vacuum cleaner), and extreme operating conditions (e.g., air conditioners running at full load due to high summer temperatures, dampness affecting line insulation during the rainy season, and sudden rises and falls in grid voltage). These scenarios place stringent demands on the adaptability, safety, and intelligence level of home power distribution systems.

[0003] However, traditional household power distribution systems have significant technical limitations: they rely on fixed load protection thresholds and determine whether to trigger trip protection based solely on a single electrical parameter (such as the line's rated current). They cannot collect in-depth electrical parameters such as harmonic components, time-series power curves, and multi-device coupling interference during appliance operation, nor can they correlate with environmental parameters such as temperature and humidity, line impedance differences corresponding to the physical installation location of the equipment, and non-electrical parameters such as equipment type identification. This makes it difficult to construct accurate models of the equipment's power consumption characteristics. Consequently, the system has extremely poor adaptability to mutual interference when multiple types of appliances are operating concurrently, the inrush current of new equipment during initial startup, and load fluctuations under extreme conditions. For example, the instantaneous inrush current when an inductive air conditioner starts can easily trigger a false trip at the fixed threshold, while aging lines leading to increased impedance and overload conditions fail to provide timely protection because the threshold is not dynamically adjusted. Ultimately, this fails to meet the actual needs of safe and stable operation of diverse household appliances. Therefore, we propose an intelligent power distribution system suitable for various household appliances. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent power distribution system applicable to a variety of household appliances, so as to solve the technical problems of traditional household power distribution systems that rely on fixed thresholds and cannot adapt to the characteristics of multiple types of appliances, concurrent interference of multiple devices, and extreme working conditions.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: an intelligent power distribution system applicable to various household appliances, comprising:

[0006] The multimodal collaborative fingerprint acquisition module is used to collect electrical parameter features and non-electrical parameter features of household appliances under different operating scenarios, and to build a multi-dimensional associated multimodal collaborative appliance fingerprint database based on the two types of features;

[0007] The AI ​​adaptive threshold control module communicates with the multimodal collaborative fingerprint acquisition module to acquire multi-dimensional electricity consumption data and establish a user electricity consumption behavior model that integrates multiple factors. Based on the model, it identifies different electricity consumption scenarios and dynamically adjusts the load protection threshold, threshold response priority, and new device access adaptation threshold of the power distribution system.

[0008] The cross-scenario intelligent early warning and self-healing module is connected to the multimodal collaborative fingerprint acquisition module and the AI ​​adaptive threshold control module respectively. It is used to monitor the operation of electrical appliances and the status of the power grid. When an abnormality is detected, it triggers an early warning and executes linkage protection actions, while generating a self-healing strategy.

[0009] The power distribution execution and privacy protection module communicates with the AI ​​adaptive threshold control module and the cross-scenario intelligent early warning and self-healing module. It is used to control the power supply of home appliances according to the output instructions of the corresponding modules, and at the same time, it performs desensitization processing on the collected power consumption data.

[0010] The device type dynamic adaptation module communicates with the multimodal collaborative fingerprint acquisition module and the AI ​​adaptive threshold control module. It is used to match similar power consumption models for newly connected devices and generate initial protection thresholds, and continuously optimize the thresholds to achieve dynamic adaptation between new devices and the system.

[0011] Preferably, the multimodal collaborative fingerprint acquisition module includes:

[0012] The full-scenario electrical parameter acquisition unit is used to collect harmonic component data, time-series power curve change data, coupling interference data when multiple devices are running concurrently, inrush current data, inrush voltage data and start-up duration data when a new device is started for the first time, forming a complete set of electrical parameter characteristics.

[0013] The multi-dimensional non-electrical parameter acquisition unit is used to collect environmental parameters affecting electrical load, electrical usage frequency data, equipment physical installation location parameters, and equipment type identification parameters. The environmental parameters include temperature, humidity, and air pressure. The electrical usage frequency data is obtained based on historical switch status statistics. The equipment physical installation location parameters are used to analyze the impact of line impedance on electrical parameters. The equipment type identification parameters are obtained by reading the communication protocol when the equipment is connected or by manual input by the user. The equipment type identification parameters are used to distinguish the load characteristics of different types of electrical appliances.

[0014] The fingerprint association construction and update unit is used to associate and map the electrical parameter data output by the full-scene electrical parameter acquisition unit with the non-electrical parameter data output by the multi-dimensional non-electrical parameter acquisition unit to construct a multimodal collaborative electrical fingerprint database containing "device independent features - device collaborative features - environmental location features - device type features - new device features".

[0015] To quantify the correlation of multimodal features, the fingerprint association construction and update unit uses a feature fusion algorithm to calculate the feature association weights, as shown in the following formula:

[0016]

[0017] Among them, W i,j Let cov(F) be the association weight between the i-th electrical parameter feature and the j-th non-electrical parameter feature. i ,F j ) represents the electrical parameter characteristic F i With non-electrical parameter characteristics F j covariance, var(F i ) represents the electrical parameter characteristic F i The variance, var(F) j ) represents the non-electrical parameter characteristic F j The variance, α i Electrical parameter characteristics F i Importance coefficient, β j Non-electrical parameter characteristic F j Importance coefficient.

[0018] Preferably, the AI ​​adaptive threshold control module includes:

[0019] The multi-dimensional data acquisition unit is used to continuously collect household electricity consumption data, line impedance data, extreme operating condition data, load characteristic data associated with equipment type, and new equipment access data.

[0020] The household electricity consumption data includes the time period, duration, and number of concurrent users for each appliance.

[0021] The line impedance data is obtained based on the physical installation location of the equipment.

[0022] The line impedance data includes line resistance and capacitance parameters;

[0023] The extreme operating condition data includes power consumption parameters during periods of high summer temperatures, humid rainy seasons, sudden drops in grid voltage, and sudden increases in grid voltage.

[0024] The load characteristic data associated with the device type includes typical power consumption parameter ranges for resistive loads, inductive loads, and capacitive loads.

[0025] The new device access data includes the new device startup parameters and operating parameters.

[0026] Preferably, the AI ​​adaptive threshold control module further includes:

[0027] The integrated behavior analysis unit, based on the data collected by the multi-dimensional data acquisition unit, uses an improved deep learning algorithm to identify regular electricity consumption patterns, special electricity consumption scenarios, extreme operating conditions, and electricity consumption trends after new equipment is connected. It establishes a user electricity consumption behavior model that integrates line characteristics, extreme operating conditions, equipment types, and new equipment connection. The improved deep learning algorithm integrates CNN, LSTM, and attention mechanisms. The special electricity consumption scenarios include multiple devices operating concurrently during family gatherings and long-term electricity consumption during holidays. The model can quantify the influence coefficient of line impedance on protection thresholds, threshold correction coefficients under extreme operating conditions, load adaptation coefficients for different equipment types, and threshold adjustment coefficients for new equipment connection.

[0028] The hierarchical threshold adjustment unit adjusts the load protection threshold according to the scenario information output by the fusion behavior analysis unit, with the priority order being extreme working conditions > unstable new device access scenarios > special power consumption scenarios > normal power consumption periods. For extreme working conditions, an additional threshold buffer margin is added. For multi-device concurrent scenarios, the threshold anti-interference margin is adjusted according to the collaborative interference feature sub-library data. For new device access scenarios, an initial adaptation threshold is set based on the threshold adjustment coefficient of the new device access, and it is gradually optimized after the new device is running stably.

[0029] The graded threshold adjustment unit uses a dynamic threshold calculation formula to determine the load protection threshold, as follows:

[0030]

[0031] Among them, I th For the dynamically adjusted load protection threshold, I base W is the baseline protection threshold. k,env K represents the association weight between the k-th environmental feature and the electrical parameter feature. env,k W is the threshold influence coefficient of the k-th environmental feature. m,dev K represents the association weight between the m-th device type feature and the electrical parameter feature. dev,m Let f(m) be the threshold influence coefficient for the m-th device type feature;

[0032] The dynamic self-optimization unit automatically extracts multimodal collaborative fingerprint features for the corresponding scenario and updates them to the database when it detects unexpected tripping, misjudgment caused by coupling interference, or new equipment access adaptation deviation. At the same time, it optimizes the line impedance influence coefficient, extreme condition correction coefficient, equipment type load adaptation coefficient, and new equipment access threshold adjustment coefficient to improve the threshold adjustment accuracy in subsequent scenarios. It also periodically updates the collaborative interference feature sub-library and equipment type feature sub-library based on newly added equipment combination data and new equipment data.

[0033] Preferably, the cross-scenario intelligent early warning and self-healing module includes:

[0034] The multi-dimensional risk monitoring unit is used to continuously monitor the trend of electrical harmonic distortion, the state of power grid harmonics, the degree of concurrent coupling interference of multiple devices, the trend of line aging impedance, and the operational stability after the connection of new equipment.

[0035] The multi-dimensional risk monitoring unit uses a risk assessment algorithm to calculate the abnormal risk value, as shown in the following formula:

[0036]

[0037] Where R is the abnormal risk value, I real I is the actual load current. th The dynamic protection threshold is γ, which is the current fluctuation correction coefficient, and W is the W value. q,risk Let ΔX be the weight of the q-th risk factor. q The deviation of the q-th risk factor;

[0038] The graded early warning generation unit generates a first-level aging early warning when the periodic change in the harmonic distortion rate of electrical appliances reaches a preset mild threshold, and generates a second-level emergency early warning when it reaches a preset severe threshold.

[0039] A power grid anomaly warning is generated when the harmonic content of the power grid exceeds the harmonic standard.

[0040] When the degree of coupling interference exceeds the safety threshold of the cooperative interference feature sub-library, a device cooperative interference warning is generated.

[0041] A line aging warning is generated when the periodic change rate of line impedance exceeds the preset aging threshold.

[0042] A new equipment adaptation warning is generated when the deviation of the operating parameters of the new equipment exceeds the preset adaptation range or the frequency of parameter fluctuation is too high.

[0043] Preferably, the cross-scenario intelligent early warning and self-healing module further includes:

[0044] The intelligent self-healing and linkage protection unit pushes maintenance suggestions for Level 1 aging warnings and automatically cuts off the power supply to corresponding electrical appliances for Level 2 emergency warnings; it activates the built-in active filter component and adjusts the protection threshold for power grid anomaly warnings; it cuts off the power supply to low-priority devices according to their importance for device interference warnings, with the order of importance being: devices maintaining basic living needs > devices meeting daily functional needs > non-essential devices that enhance convenience; it generates line maintenance plans for line aging warnings and reduces the upper limit of the power supply load on the corresponding line; it adjusts the protection threshold for new equipment for new equipment adaptation warnings, and if the adjustment is still unstable, it cuts off the power supply to the new equipment and pushes adaptation suggestions; the unit also generates a self-healing strategy library adapted to the current scenario based on historical fault handling data, real-time operating conditions, and equipment type characteristics, realizing autonomous optimization of fault handling;

[0045] The fault tracing unit, when a power supply abnormality occurs, combines a multimodal collaborative fingerprint database, line impedance data, equipment type characteristic data, and new equipment access data to locate the root cause of the fault. The root cause of the fault includes electrical appliance faults, line aging, power grid problems, equipment collaborative interference, and new equipment adaptation problems, thus avoiding the shortcomings of traditional systems that can only judge the fault result but cannot trace the source.

[0046] Preferably, the power distribution execution and privacy protection module includes:

[0047] The priority power supply control unit is used to receive the hierarchical threshold instructions and new device access adaptation instructions output by the AI ​​adaptive threshold control module, and to perform power supply control according to the importance level of the devices. The importance level of the devices includes core devices, regular devices, auxiliary devices, and newly connected unstable devices. Core devices refer to devices that maintain the basic living needs of the family or ensure special health needs. Regular devices refer to devices that meet the daily functional needs of the family. Auxiliary devices refer to non-essential devices that improve the convenience of life. Under extreme working conditions, priority is given to ensuring the power supply of core devices. During the new device adaptation phase, the power supply priority of new devices is appropriately reduced.

[0048] The self-healing execution unit is used to receive the self-healing strategy output by the cross-scenario intelligent early warning and self-healing module, execute actions such as filter activation, load shedding, line load adjustment, and new device threshold optimization, and record self-healing process data for strategy optimization.

[0049] The data anonymization and privacy protection unit anonymizes the collected raw electricity consumption data, deleting user-identifiable information, which includes data on the correlation between specific electricity consumption time and family members' activities, and privacy information on the specific location of the device. Only appliance characteristic parameters, scene characteristic parameters, and device type characteristic parameters are extracted for system analysis. The appliance characteristic parameters include harmonics and power, and the scene characteristic parameters include environment and number of concurrent users.

[0050] The status feedback unit feeds back the power supply control status, self-healing execution status, data desensitization status and new equipment operation status to the AI ​​adaptive threshold control module, cross-scenario intelligent early warning and self-healing module and equipment type dynamic adaptation module, forming a complete control closed loop.

[0051] The priority power supply control unit uses a power supply priority allocation algorithm to determine the power supply sequence of the devices, as shown in the following formula:

[0052]

[0053] Among them, P dev Prioritize power supply to devices, S base Here, R represents the equipment baseline priority, R is the anomaly risk value, 0.3 is the risk correction factor, and W...t,use For the weight of the t-th feature, T t,use Let be the duration coefficient for the t-th feature used.

[0054] Preferably, the device type dynamic adaptation module includes:

[0055] The new device identification unit is used to determine the type of a newly connected device by detecting electrical signals when the device is connected, reading communication protocols, or using device type identifiers input by the user. The electrical signal detection includes starting impact feature detection, the communication protocol reading includes reading the model information of the smart device, and the device type includes kitchen functional equipment, audio-visual entertainment equipment, and health auxiliary equipment.

[0056] The feature matching unit, based on the type of the new device and the initially collected electrical and non-electrical parameter features, matches the historical power consumption models of the same or similar types of devices from the device type feature sub-library of the multimodal collaborative electrical fingerprint database, and extracts the protection threshold parameters and adaptation environment parameters of similar devices as the initial reference data for the new device.

[0057] Preferably, the device type dynamic adaptation module further includes:

[0058] The dynamic adaptation and optimization unit continuously collects power consumption data under different operating conditions during the operation of new equipment. These different operating conditions include no-load, full-load, and standby states. The data is compared and analyzed with the initial reference data to generate a protection threshold adjustment curve specific to the new equipment. The power supply control strategy of the new equipment is gradually optimized until the operating parameters of the new equipment are stable within the adaptation range. At the same time, the stable characteristic data of the new equipment is updated to the multimodal collaborative electrical fingerprint database to enrich the equipment type characteristic sub-library.

[0059] The dynamic adaptation optimization unit uses a threshold optimization algorithm to update the new device-specific protection threshold, and the specific formula is as follows:

[0060]

[0061] Among them, I new,th (k+1) represents the new device-specific protection threshold after the (k+1)th optimization, I new,th (k) represents the new device-specific protection threshold before the k-th optimization, R represents the abnormal risk value of the new device operation, and I new,real I represents the actual operating current of the new equipment. th 0.15 is the system dynamic protection threshold, and 0.15 is the optimization step size coefficient.

[0062] Preferably, the multimodal collaborative fingerprint acquisition module, AI adaptive threshold control module, cross-scenario intelligent early warning and self-healing module, power distribution execution and privacy protection module, and device type dynamic adaptation module achieve data interaction through a dual-link communication method. The dual-link communication method includes wired communication and encrypted wireless communication. The encrypted wireless communication is used to cope with wired line fault scenarios. At the same time, an anti-interference algorithm is added to the wireless communication. The anti-interference algorithm includes frequency hopping technology and signal filtering to avoid interference from commonly used household wireless signals and electromagnetic devices on system communication, and to ensure the transmission stability of new device access data and fault early warning data. Commonly used household wireless signals include short-range wireless communication signals and home network signals. Electromagnetic devices refer to household devices that generate electromagnetic radiation during operation.

[0063] Compared with the prior art, the beneficial effects of the present invention are:

[0064] 1. This invention utilizes a combined design of a "multimodal collaborative fingerprint acquisition module + AI adaptive threshold control module" to simultaneously collect electrical parameters (harmonic components, inrush current, and coupled interference data) and multidimensional non-electrical parameters (ambient temperature and humidity, equipment location, and type identification) of electrical appliances across all scenarios. This constructs a fingerprint database containing features of "device independence, collaboration, environment, type, and new equipment." Furthermore, by combining an improved deep learning algorithm integrating CNN, LSTM, and attention mechanisms, a user electricity behavior model is established that correlates line characteristics, extreme operating conditions, and equipment types. This dynamically adjusts the load protection threshold and response priority (e.g., priority is higher in extreme operating conditions than during normal periods). This design fundamentally solves the problem that traditional fixed thresholds cannot adapt to the characteristics of multiple types of electrical appliances, concurrent interference from multiple devices, and extreme operating conditions, significantly reducing the incidence of false tripping and untimely protection, and ensuring the safe operation of household appliances.

[0065] 2. This invention addresses the limitation of dynamic thresholds, which can only handle explicit load anomalies but cannot identify latent faults, by designing a cross-scenario intelligent early warning and self-healing module. Through a multi-dimensional risk monitoring unit, it tracks in real-time trends in electrical harmonic distortion, line aging impedance changes, and grid harmonic status. A risk assessment algorithm quantifies abnormal risk values ​​(corresponding to actual current, dynamic thresholds, and deviations from risk factors) to generate tiered early warnings (such as pushing maintenance suggestions for mildly aging appliances and automatically cutting off power for severely aging appliances). Simultaneously, it triggers self-healing actions (such as activating active filtering when the grid is abnormal and cutting off power to low-priority devices when there is interference). This design overcomes the shortcomings of traditional systems that can only trip after a fault occurs, unable to predict or self-heal in advance, thus preventing the escalation of latent faults such as line aging and grid harmonic pollution.

[0066] 3. This invention addresses the issues of slow initial adaptation response of dynamic thresholds to new devices and chaotic power supply priorities under extreme conditions. Through a device type dynamic adaptation module, it quickly matches similar power consumption models for new devices and generates initial protection thresholds based on a multimodal fingerprint database. Then, a dynamic optimization unit continuously collects idle, full-load, and standby data of the new devices, iteratively adjusting the specific thresholds. Simultaneously, the power distribution execution and privacy protection module prioritizes power supply according to the order of "core devices > conventional devices > auxiliary devices," prioritizing basic living and health needs under extreme conditions, and anonymizing power consumption data. This design not only solves the problem of delayed initial adaptation to new devices but also ensures the stability of power supply to core devices and user privacy, further improving system compatibility and practicality. Attached Figure Description

[0067] Figure 1 This is a schematic diagram of the system framework of the present invention. Detailed Implementation

[0068] To facilitate understanding of the technical solution of the present invention by those skilled in the art, the technical solution of the present invention will now be further described in conjunction with the accompanying drawings.

[0069] Example 1, such as Figure 1 As shown, the present invention provides an intelligent power distribution system applicable to various household appliances, comprising:

[0070] The multimodal collaborative fingerprint acquisition module is used to collect electrical parameter features and non-electrical parameter features of household appliances under different operating scenarios, and to build a multi-dimensional associated multimodal collaborative appliance fingerprint database based on the two types of features;

[0071] The AI ​​adaptive threshold control module communicates with the multimodal collaborative fingerprint acquisition module to acquire multi-dimensional electricity consumption data and establish a user electricity consumption behavior model that integrates multiple factors. Based on the model, it identifies different electricity consumption scenarios and dynamically adjusts the load protection threshold, threshold response priority, and new device access adaptation threshold of the power distribution system.

[0072] The cross-scenario intelligent early warning and self-healing module is connected to the multimodal collaborative fingerprint acquisition module and the AI ​​adaptive threshold control module respectively. It is used to monitor the operation of electrical appliances and the status of the power grid. When an abnormality is detected, it triggers an early warning and executes linkage protection actions, while generating a self-healing strategy.

[0073] The power distribution execution and privacy protection module communicates with the AI ​​adaptive threshold control module and the cross-scenario intelligent early warning and self-healing module. It is used to control the power supply of home appliances according to the output instructions of the corresponding modules, and at the same time, it performs desensitization processing on the collected power consumption data.

[0074] The device type dynamic adaptation module communicates with the multimodal collaborative fingerprint acquisition module and the AI ​​adaptive threshold control module. It is used to match similar power consumption models for newly connected devices and generate initial protection thresholds, and continuously optimize the thresholds to achieve dynamic adaptation between new devices and the system.

[0075] In an embodiment of the present invention, the multimodal collaborative fingerprint acquisition module includes:

[0076] The full-scenario electrical parameter acquisition unit is used to collect harmonic component data, time-series power curve change data, coupling interference data (including at least voltage fluctuation amplitude and current mutual interference frequency) when electrical appliances are running independently, inrush current data, inrush voltage data and start-up duration data when new equipment is started for the first time, forming a complete set of electrical parameter features.

[0077] The multi-dimensional non-electrical parameter acquisition unit is used to collect environmental parameters affecting electrical loads, electrical usage frequency data, equipment physical installation location parameters, and equipment type identification parameters. Environmental parameters include temperature, humidity, and air pressure. Electrical usage frequency data is obtained based on historical switch status statistics. Equipment physical installation location parameters are used to correlate and analyze the impact of line impedance on electrical parameters. Equipment type identification parameters are obtained by reading the communication protocol when the equipment is connected or by manual input by the user. Equipment type identification parameters are used to distinguish the load characteristics of different types of electrical appliances.

[0078] The fingerprint association construction and update unit is used to associate and map the electrical parameter data output by the full-scenario electrical parameter acquisition unit with the non-electrical parameter data output by the multi-dimensional non-electrical parameter acquisition unit to build a multimodal collaborative electrical fingerprint database containing "device independent features - device collaborative features - environmental location features - device type features - new device features". At the same time, when a new device is connected, the device parameters change, or the environment changes over a long period of time, the corresponding feature data in the database is automatically updated to maintain the timeliness of the database.

[0079] To quantify the correlation of multimodal features, the fingerprint association construction and update unit uses a feature fusion algorithm to calculate the feature association weights, as shown in the following formula:

[0080]

[0081] Among them, W i,j The correlation weight between the i-th electrical parameter feature and the j-th non-electrical parameter feature is the core indicator for measuring the degree of correlation between the two types of features. The higher the weight, the stronger the ability of the two types of features to jointly reflect the electrical characteristics of the equipment.

[0082] cov(F i ,F j ) represents the electrical parameter characteristic F i With non-electrical parameter characteristics Fj The covariance is used to quantify the direction and strength of the linear association between two types of features. Positive values ​​indicate positive correlation, negative values ​​indicate negative correlation, and the larger the absolute value, the stronger the association.

[0083] var(F i ) represents the electrical parameter characteristic F i The variance reflects the degree of dispersion of the feature data and is used to normalize the covariance results to eliminate the influence of dimensions.

[0084] var(F j ) represents the non-electrical parameter characteristic F j The variance, with respect to var(F) i Consistency is maintained to ensure the objectivity of the correlation calculation results;

[0085] α i Electrical parameter characteristics F i The importance coefficient is set according to the role of the feature in device identification and load judgment. Core features (such as harmonic features) have higher coefficients, while non-core features have lower coefficients.

[0086] β j Non-electrical parameter characteristic F j Importance coefficient, setting logic and α i Consistency is used to distinguish the priority of different non-electrical parameter characteristics (such as ambient temperature and frequency of use);

[0087] This feature fusion algorithm constructs its operational logic based on "feature correlation as the core and feature importance as a supplement." First, it quantifies the linear correlation between electrical parameter features and non-electrical parameter features by calculating the covariance. Then, it introduces feature variance for normalization to avoid bias in correlation calculation due to differences in feature dimensions. Finally, it applies a weighted correction to the correlation result by multiplying the importance coefficients of electrical parameter features and non-electrical parameter features, highlighting the dominant role of core features in the correlation weight. Ultimately, it obtains the comprehensive correlation weight of the two types of features, providing a quantitative basis for the construction of a multimodal fingerprint database.

[0088] This algorithm addresses the problems of "incomplete information and low recognition accuracy" in traditional single-dimensional feature acquisition. By quantifying the association weights of multimodal features, it achieves deep fusion of electrical and non-electrical parameters. This avoids redundant features interfering with database construction efficiency while highlighting the characterization of core features in terms of device power consumption. This allows the multimodal collaborative fingerprint database to more accurately reflect the actual power consumption status of devices, providing high-quality data support for subsequent threshold adjustments, risk warnings, and other functions, and enhancing the system's adaptability to diverse household appliances.

[0089] In an embodiment of the present invention, the AI ​​adaptive threshold control module includes:

[0090] The multi-dimensional data acquisition unit is used to continuously collect household electricity consumption data, line impedance data, extreme operating condition data, load characteristic data associated with equipment type, and new equipment access data.

[0091] Household electricity consumption data includes the time period, duration, and number of concurrent users for each appliance;

[0092] The line impedance data is obtained based on calculations of the physical installation location of the equipment.

[0093] Line impedance data includes line resistance and capacitance parameters;

[0094] Extreme operating condition data includes power consumption parameters during periods of high summer temperatures, humid rainy seasons, sudden drops in grid voltage, and sudden increases in grid voltage.

[0095] The load characteristic data associated with the equipment type includes typical power parameter ranges for resistive, inductive, and capacitive loads.

[0096] The data accessed by the new device includes the device's startup parameters and operating parameters.

[0097] The integrated behavior analysis unit, based on data collected by the multi-dimensional data acquisition unit, uses an improved deep learning algorithm to identify regular electricity consumption patterns, special electricity consumption scenarios, extreme operating conditions, and electricity consumption trends after new equipment is connected. It establishes a user electricity consumption behavior model that integrates line characteristics, extreme operating conditions, equipment types, and new equipment connection. The improved deep learning algorithm integrates CNN, LSTM, and attention mechanisms. Special electricity consumption scenarios include multiple devices operating concurrently during family gatherings and long-term electricity consumption during holidays. The model can quantify the influence coefficient of line impedance on protection threshold, threshold correction coefficient under extreme operating conditions, load adaptation coefficient of different equipment types, and threshold adjustment coefficient for new equipment connection.

[0098] The tiered threshold adjustment unit adjusts the load protection threshold according to the scenario information output by the fusion behavior analysis unit, with the priority order being extreme operating conditions > unstable new device access scenarios > special power consumption scenarios > normal power consumption periods. For extreme operating conditions, additional threshold buffer margin is added. For multi-device concurrent scenarios, the threshold anti-interference margin is adjusted based on the collaborative interference feature sub-library data. For new device access scenarios, an initial adaptation threshold is set based on the threshold adjustment coefficient of the new device access, and it is gradually optimized after the new device is running stably.

[0099] The dynamic self-optimization unit automatically extracts multimodal collaborative fingerprint features in the corresponding scenario and updates them to the database when it detects unexpected tripping, misjudgment caused by coupling interference, or new equipment access adaptation deviation. At the same time, it optimizes the line impedance influence coefficient, extreme condition correction coefficient, equipment type load adaptation coefficient, and new equipment access threshold adjustment coefficient to improve the threshold adjustment accuracy in subsequent scenarios. It also periodically updates the collaborative interference feature sub-library and equipment type feature sub-library based on newly added equipment combination data and new equipment data.

[0100] The graded threshold adjustment unit uses a dynamic threshold calculation formula to determine the load protection threshold, as follows:

[0101]

[0102] Among them, I th The dynamically adjusted load protection threshold is the core basis for the power distribution system to determine whether to trigger protection action, and directly determines the sensitivity and safety of the protection.

[0103] I base The baseline protection threshold is set based on the rated load-bearing capacity of the household power distribution line and serves as the basic reference value for dynamic threshold adjustment.

[0104] W k,env The correlation weight between the k-th environmental feature and the electrical parameter feature is given by W. i,j The derivation shows that the greater the impact of the environmental characteristics on the equipment's power consumption parameters, the higher the weight.

[0105] K env,k The threshold influence coefficient for the k-th environmental feature is set according to the actual impact of the environmental feature on the load. For example, the coefficient is higher in high-temperature environments to expand the threshold buffer space.

[0106] W m,dev The association weight between the m-th device type feature and the electrical parameter feature is given by W. i,j The derivation shows that this reflects the close correlation between equipment type and power consumption parameters;

[0107] K dev,m The threshold influence coefficient for the m-th device type feature is set according to the load fluctuation characteristics of the device type. The coefficient is higher for devices with large load fluctuations to avoid false triggering of protection.

[0108] The dynamic threshold calculation formula is based on a benchmark threshold and adjusted based on multiple scenario factors. First, a benchmark protection threshold is used as the initial value to ensure the basic safety of threshold adjustment. Then, the impact of environmental factors (such as temperature and humidity) on load protection is quantified by weighted summation of environmental feature association weights and environmental threshold influence coefficients. Simultaneously, weighted summation of equipment type feature association weights and equipment threshold influence coefficients is introduced to reflect the differences in load characteristics of different equipment types (such as resistive and inductive loads). Finally, the environmental impact item and the equipment type impact item are added together and multiplied by the benchmark threshold to obtain the dynamically adjusted load protection threshold, achieving precise matching between the threshold and the scenario.

[0109] This algorithm breaks through the limitations of traditional fixed thresholds, which "cannot adapt to changes in scenarios," and achieves dynamic and intelligent adjustment of load protection thresholds. By combining environmental and equipment type factors, it can avoid false protection caused by abnormal load fluctuations in special environments (such as high temperature and humidity), and can set protection thresholds differently according to equipment characteristics, ensuring the safe and stable operation of various household appliances. At the same time, the correlation weights are derived from the multimodal feature fusion results mentioned above, ensuring the scientific nature and consistency of threshold adjustment and improving the system's adaptability to complex household electricity usage scenarios.

[0110] In an embodiment of the present invention, the cross-scenario intelligent early warning and self-healing module includes:

[0111] The multi-dimensional risk monitoring unit is used to continuously monitor the electrical harmonic distortion trend, power grid harmonic status, degree of concurrent coupling interference of multiple devices, line aging impedance trend, and operational stability after the connection of new equipment. The electrical harmonic distortion trend is obtained based on the periodic distortion rate change. The power grid harmonic status is obtained based on the harmonic content of the input voltage of the public power grid. The degree of concurrent coupling interference of multiple devices is obtained based on the difference between the coupling interference data and the threshold of the cooperative interference feature sub-library. The line aging impedance trend is obtained based on the periodic change rate of the line impedance. The line aging impedance trend is obtained by comparing the initial impedance associated with the physical installation location of the equipment with the real-time impedance. The operational stability after the connection of new equipment is obtained based on the deviation of the operating parameters of the new equipment from the adaptation threshold and the parameter fluctuation frequency. The operational stability of the new equipment is used to determine whether the new equipment is suitable for the current power distribution environment.

[0112] The graded early warning generation unit generates a first-level aging early warning when the periodic change in the harmonic distortion rate of electrical appliances reaches a preset mild threshold, and generates a second-level emergency early warning when it reaches a preset severe threshold.

[0113] A power grid anomaly warning is generated when the harmonic content of the power grid exceeds the harmonic standard.

[0114] When the degree of coupling interference exceeds the safety threshold of the cooperative interference feature sub-library, a device cooperative interference warning is generated.

[0115] A line aging warning is generated when the periodic change rate of line impedance exceeds the preset aging threshold.

[0116] A new equipment adaptation warning is generated when the deviation of the operating parameters of the new equipment exceeds the preset adaptation range or the frequency of parameter fluctuation is too high.

[0117] The intelligent self-healing and linkage protection unit pushes maintenance suggestions for Level 1 aging warnings and automatically cuts off the power supply to corresponding appliances for Level 2 emergency warnings; it activates the built-in active filter component and adjusts the protection threshold for power grid anomaly warnings; it cuts off the power supply to low-priority devices according to their importance for device interference warnings, with the order of importance being: devices that maintain basic living needs > devices that meet daily functional needs > non-essential devices that enhance convenience; it generates line maintenance plans for line aging warnings and reduces the upper limit of the power supply load on the corresponding line; it adjusts the protection threshold for new equipment for new equipment adaptation warnings, and if the adjustment is still unstable, it cuts off the power supply to the new equipment and pushes adaptation suggestions; the unit also generates a self-healing strategy library adapted to the current scenario based on historical fault handling data, real-time operating conditions, and equipment type characteristics, realizing autonomous optimization of fault handling;

[0118] The fault tracing unit, when a power supply abnormality occurs, combines a multimodal collaborative fingerprint database, line impedance data, equipment type characteristic data, and new equipment access data to locate the root cause of the fault. The root cause of the fault includes the fault of the electrical appliance itself, line aging, power grid problems, equipment collaborative interference, and new equipment adaptation problems, thus avoiding the shortcomings of traditional systems that can only judge the fault result but cannot trace the source.

[0119] The multi-dimensional risk monitoring unit uses a risk assessment algorithm to calculate the abnormal risk value. The specific formula is as follows:

[0120]

[0121] Among them, R is the abnormal risk value, which is the core indicator for determining whether to trigger an early warning. The larger the value, the higher the current fault or abnormal risk. An early warning will be triggered when the set threshold is reached.

[0122] I real The actual load current is the total load current of household appliances or the load current of a single device collected in real time, reflecting the current actual electricity consumption situation.

[0123] I th The dynamic protection threshold serves as a reference standard for determining whether the load is abnormal.

[0124] γ is the current fluctuation correction coefficient, which is used to correct the interference of instantaneous current spikes or troughs on risk assessment, so that the assessment results are more in line with the actual power consumption status.

[0125] W q,riskThe weight of the q-th risk factor reflects the degree of impact of the risk factor on system safety. For example, the weight of line aging is higher than that of slight harmonic distortion.

[0126] ΔX q The deviation of the q-th risk factor quantifies the degree of deviation between the actual value of the risk factor and the safety standard value. The larger the deviation, the higher the deviation.

[0127] This risk assessment algorithm constructs its operational logic based on "threshold matching degree as the foundation and multiple risk factors as a supplement." First, it calculates the current matching degree by comparing the actual load current with the dynamic protection threshold, initially determining whether the current load is close to the protection threshold. Then, it introduces a current fluctuation correction coefficient to correct for the interference of instantaneous current fluctuations on the matching degree. Next, it quantifies the comprehensive impact of various risks (such as harmonic distortion and line aging) by weighted summation of the correlation weights and deviations of multiple risk factors. Finally, it multiplies the current matching degree correction result with the comprehensive impact of the risk factors to obtain the abnormal risk value, achieving a precise quantitative assessment of fault risk.

[0128] This algorithm solves the problems of traditional early warning systems that "rely on a single indicator and have a high false alarm / missed alarm rate." Through multi-dimensional risk quantification assessment, it achieves comprehensive monitoring of household electrical safety. It can accurately identify overload risks approaching protection thresholds and comprehensively consider hidden risks such as harmonic distortion and aging lines, thus detecting potential faults in advance. Furthermore, the risk value calculation is deeply integrated with the aforementioned dynamic thresholds and feature correlation weights, ensuring the consistency and scientific nature of the early warning logic, effectively reducing the interference of false alarms on users, and improving the reliability and practicality of the system's safety protection.

[0129] In an embodiment of the present invention, the power distribution execution and privacy protection module includes:

[0130] The priority power supply control unit is used to receive the hierarchical threshold instructions and new device access adaptation instructions output by the AI ​​adaptive threshold control module, and to perform power supply control according to the importance level of the device. The importance level of the device includes core devices, regular devices, auxiliary devices, and newly connected unstable devices. Core devices refer to devices that maintain the basic living needs of the family or ensure special health needs. Regular devices refer to devices that meet the daily functional needs of the family. Auxiliary devices refer to non-essential devices that improve the convenience of life. Under extreme working conditions, priority is given to ensuring the power supply of core devices. During the new device adaptation phase, the power supply priority of new devices is appropriately reduced.

[0131] The self-healing execution unit is used to receive cross-scenario intelligent early warning and self-healing strategy output by the self-healing module, and execute actions such as filter activation, load shedding, line load adjustment, and new device threshold optimization. At the same time, it records self-healing process data for strategy optimization.

[0132] The data anonymization and privacy protection unit anonymizes the collected raw electricity consumption data, removing user-identifiable information, including data related to specific electricity consumption time and family members' activities, and privacy information about the specific location of devices. Only appliance characteristic parameters, scene characteristic parameters, and device type characteristic parameters are extracted for system analysis. Appliance characteristic parameters include harmonics and power, and scene characteristic parameters include environment and number of concurrent users. The anonymized data is stored locally only and is not uploaded to the external cloud to prevent leakage of user electricity consumption privacy.

[0133] The status feedback unit feeds back the power supply control status, self-healing execution status, data desensitization status and new equipment operation status to the AI ​​adaptive threshold control module, cross-scenario intelligent early warning and self-healing module and equipment type dynamic adaptation module, forming a complete control closed loop.

[0134] The priority power supply control unit uses a power supply priority allocation algorithm to determine the power supply sequence of the devices. The specific formula is as follows:

[0135]

[0136] Among them, P dev The power supply priority for equipment determines the order in which equipment receives power when power resources are scarce; the higher the priority, the more likely it is to receive power.

[0137] S base The baseline priority of the equipment is set according to the importance of the equipment in family life, with core equipment (such as equipment that ensures basic living) having the highest baseline priority and auxiliary equipment having the lowest.

[0138] R is the abnormal risk value, which is used to adjust the priority. The higher the risk, the more significant the downgrade of the priority of non-core equipment.

[0139] 0.3 is the risk correction coefficient, which controls the impact of abnormal risks on priority and avoids excessive power outages of non-core equipment due to excessively high risks;

[0140] W t,use The weight of the t-th usage feature reflects the degree of influence of this usage feature (such as usage duration or usage frequency) on the priority.

[0141] T t,use Let t be the duration coefficient of the t-th usage feature, which quantifies the urgency of the current usage demand for the device. For example, the duration coefficient of the device in use is higher than that of the idle device.

[0142] This power supply priority allocation algorithm constructs its operational logic based on "baseline priority and adjustment based on risk and usage demand." First, it uses the equipment's baseline priority as the initial value, clarifying the inherent priority differences between core equipment, regular equipment, and auxiliary equipment. Then, it adjusts the baseline priority using abnormal risk values; higher risks appropriately reduce the priority of non-core equipment to ensure the power supply safety of core equipment. Finally, it introduces a weighted sum of usage characteristic weights and duration coefficients to reflect the urgency of the equipment's current usage demand; higher usage demand leads to a more significant priority increase. Ultimately, it obtains the actual power supply priority of the equipment, guiding power supply control actions.

[0143] This algorithm enables intelligent priority management of power supply for household appliances, solving the problem of core equipment power outages caused by the traditional "one-size-fits-all" power supply approach. By combining equipment importance, current risks, and usage needs, it prioritizes power supply to core equipment under overload or abnormal risk conditions to meet basic household living and safety needs, while also accommodating the reasonable usage needs of regular equipment, avoiding excessive restrictions that could impact user convenience. Furthermore, the priority calculation is closely integrated with the aforementioned risk assessment and feature association weights, ensuring the consistency and rationality of power supply control logic and improving the system's power dispatching capabilities in complex power usage scenarios.

[0144] In an embodiment of the present invention, the device type dynamic adaptation module includes:

[0145] The new device identification unit is used to determine the type of newly connected device by detecting electrical signals when the device is connected, reading communication protocols, or using device type identifiers input by the user. Electrical signal detection includes starting impact feature detection, communication protocol reading includes reading the model information of smart devices, and device type includes kitchen functional equipment, audio-visual entertainment equipment, and health assistance equipment.

[0146] The feature matching unit, based on the type of the new device and the initially collected electrical and non-electrical parameter features, matches the historical power consumption models of the same or similar types of devices from the device type feature sub-database of the multimodal collaborative electrical fingerprint database, and extracts the protection threshold parameters and adaptation environment parameters of similar devices as the initial reference data for the new device.

[0147] The dynamic adaptation and optimization unit continuously collects power consumption data under different operating conditions during the operation of new equipment, including no-load, full-load, and standby states. It compares and analyzes the data with the initial reference data to generate a protection threshold adjustment curve specific to the new equipment. It gradually optimizes the power supply control strategy of the new equipment until the operating parameters of the new equipment are stable within the adaptation range. At the same time, it updates the stable characteristic data of the new equipment to the multimodal collaborative electrical fingerprint database to enrich the equipment type characteristic sub-library.

[0148] The dynamic adaptation optimization unit uses a threshold optimization algorithm to update the protection threshold specific to the new device. The specific formula is as follows:

[0149]

[0150] Among them, I new,th (k+1) is the protection threshold specific to the new device after the (k+1)th optimization. It is the final threshold adapted to the power consumption characteristics of the new device and becomes increasingly accurate with the number of optimizations.

[0151] I new,th (k) is the new device-specific protection threshold before the kth optimization, which serves as the basic reference value for this optimization. The initial value is derived from the similar device model.

[0152] R represents the abnormal risk value of the new equipment operation. The lower the risk, the closer the current threshold is to the optimal value, and the smaller the adjustment range.

[0153] I new,real The actual operating current of the new equipment is collected in real time, reflecting the true power consumption characteristics of the equipment.

[0154] I th The system's dynamic protection threshold serves as a reference standard for judging whether the actual current of new equipment is reasonable.

[0155] 0.15 is the optimization step size coefficient, which controls the magnitude of each threshold optimization and avoids excessive adjustment that may cause threshold fluctuations or adaptation lag.

[0156] This threshold optimization algorithm constructs its operational logic based on "the current threshold as the foundation, and risk and actual current as the adjustment criteria." First, the new device-specific threshold after the k-th optimization is used as the initial value to ensure the continuity of the optimization process. Then, the current operating status of the new device is determined by the abnormal risk value; the lower the risk, the more reasonable the current threshold, and the smaller the adjustment range. Simultaneously, the ratio of the new device's actual current to the system's dynamic threshold is introduced to quantify the matching degree between the actual power consumption status and the system's adaptive threshold; the higher the matching degree, the smaller the adjustment range. Finally, the risk correction term and the current matching term are multiplied and added to the current threshold to obtain the threshold after the (k+1)-th optimization, achieving gradual and precise adaptation of the new device's threshold.

[0157] This algorithm solves the problem of "initial threshold mismatch and easy triggering of false protection" when new devices are connected. Through dynamic iterative optimization, it achieves accurate adaptation between new devices and the power distribution system. It can quickly generate initial thresholds based on similar device models, shortening the adaptation cycle after new devices are connected. It can also gradually optimize the thresholds by continuously collecting actual power consumption data, making the thresholds increasingly closer to the actual power consumption characteristics of the new devices. Simultaneously, the optimization process combines anomaly risks with the system's dynamic thresholds to ensure the safety and rationality of threshold adjustments, effectively reducing the probability of false protection in the initial stage of new device connection and improving the system's compatibility and adaptation efficiency for new household appliances.

[0158] In embodiments of the present invention, the multimodal collaborative fingerprint acquisition module, the AI ​​adaptive threshold control module, the cross-scenario intelligent early warning and self-healing module, the power distribution execution and privacy protection module, and the device type dynamic adaptation module achieve data interaction through a dual-link communication method. The dual-link communication method includes wired communication and encrypted wireless communication. The wired communication includes RS485 bus and power line carrier, while the encrypted wireless communication includes a communication method combining Mesh network and national cryptographic SM4 encryption. The encrypted wireless communication is used to deal with wired line fault scenarios. At the same time, an anti-interference algorithm is added to the wireless communication. The anti-interference algorithm includes frequency hopping technology and signal filtering to avoid interference from commonly used household wireless signals and electromagnetic devices to the system communication, and to ensure the transmission stability of new device access data and fault early warning data. Commonly used household wireless signals include short-range wireless communication signals and home network signals. Electromagnetic devices refer to household devices that generate electromagnetic radiation during operation.

[0159] The anti-interference algorithm uses a communication quality assessment formula to select the optimal communication link. The specific formula is as follows:

[0160]

[0161] Among them, Q link Scoring the quality of communication links is a core indicator for judging the quality of links, and the scoring results are directly used for link selection decisions.

[0162] S represents the received signal strength, which reflects the strength of the effective signal in the link and is a key parameter for evaluating the basic transmission capability of the link.

[0163] N represents the background noise intensity, which refers to the inherent noise level when there is no target signal in the link, and will affect the clear transmission of the signal.

[0164] I represents the interference signal strength, which refers to the magnitude of interference signals generated by other wireless signals and electromagnetic devices in the home environment. It is an important factor affecting link stability.

[0165] L represents the total number of data features, i.e., the number of feature dimensions contained in the data to be transmitted, reflecting the complexity of the data;

[0166] W l,data The correlation weight between the type l transmitted data and electrical parameter characteristics reflects the degree of correlation between this type of data and the core electrical parameters of the system; the closer the correlation, the higher the weight.

[0167] D l The urgency coefficient for transmitting the first type of data is set according to the purpose of the data. Emergency warning data needs to be transmitted first, so the coefficient is the highest. Historical statistical data has low timeliness requirements, so the coefficient is the lowest.

[0168] This communication quality assessment formula is based on the principle of "signal quality as the foundation, supplemented by data characteristics." First, the basic signal quality of the communication link is determined by calculating the ratio of received signal strength to total interference (background noise + interfering signal). Then, the impact of data characteristics on link selection is reflected by weighted summation of the correlation weights and transmission urgency coefficients for different types of transmitted data. Finally, the weighted results of the basic signal quality and data characteristics are multiplied to obtain a comprehensive link quality score. A higher score indicates that the link is more suitable for the current data transmission requirements, thus achieving dynamic selection of the optimal communication link.

[0169] This algorithm enables intelligent and precise communication link selection, breaking the limitations of traditional fixed link selection methods. By evaluating links from both signal quality and data characteristics, it ensures stable transmission of critical information such as emergency warning data and new device access data, while also adapting to the complex wireless interference environment of homes, reducing communication interruptions or data loss. This improves the reliability of data interaction throughout the power distribution system and provides stable data transmission support for the collaborative operation of various system modules.

[0170] The embodiments disclosed in this invention are preferred embodiments, but are not limited thereto. Those skilled in the art can easily understand the spirit of this invention based on the above embodiments and make different extensions and variations, but as long as they do not depart from the spirit of this invention, they are all within the protection scope of this invention.

Claims

1. An intelligent power distribution system applicable to various household appliances, characterized in that, include: The multimodal collaborative fingerprint acquisition module is used to collect electrical parameter features and non-electrical parameter features of household appliances under different operating scenarios, and to build a multi-dimensional associated multimodal collaborative appliance fingerprint database based on the two types of features; The AI ​​adaptive threshold control module communicates with the multimodal collaborative fingerprint acquisition module to acquire multi-dimensional electricity consumption data and establish a user electricity consumption behavior model that integrates multiple factors. Based on the model, it identifies different electricity consumption scenarios and dynamically adjusts the load protection threshold, threshold response priority, and new device access adaptation threshold of the power distribution system. The cross-scenario intelligent early warning and self-healing module is connected to the multimodal collaborative fingerprint acquisition module and the AI ​​adaptive threshold control module respectively. It is used to monitor the operation of electrical appliances and the status of the power grid. When an abnormality is detected, it triggers an early warning and executes linkage protection actions, while generating a self-healing strategy. The power distribution execution and privacy protection module communicates with the AI ​​adaptive threshold control module and the cross-scenario intelligent early warning and self-healing module. It is used to control the power supply of home appliances according to the output instructions of the corresponding modules, and at the same time, it performs desensitization processing on the collected power consumption data. The device type dynamic adaptation module communicates with the multimodal collaborative fingerprint acquisition module and the AI ​​adaptive threshold control module. It is used to match similar power consumption models for newly connected devices and generate initial protection thresholds, and continuously optimize the thresholds to achieve dynamic adaptation between new devices and the system.

2. The intelligent power distribution system applicable to various household appliances according to claim 1, characterized in that, The multimodal collaborative fingerprint acquisition module includes: The full-scenario electrical parameter acquisition unit is used to collect harmonic component data, time-series power curve change data, coupling interference data when multiple devices are running concurrently, inrush current data, inrush voltage data and start-up duration data when a new device is started for the first time, forming a complete set of electrical parameter characteristics. The multi-dimensional non-electrical parameter acquisition unit is used to collect environmental parameters affecting electrical load, electrical usage frequency data, equipment physical installation location parameters, and equipment type identification parameters. The environmental parameters include temperature, humidity, and air pressure. The electrical usage frequency data is obtained based on historical switch status statistics. The equipment physical installation location parameters are used to analyze the impact of line impedance on electrical parameters. The equipment type identification parameters are obtained by reading the communication protocol when the equipment is connected or by manual input by the user. The equipment type identification parameters are used to distinguish the load characteristics of different types of electrical appliances. The fingerprint association construction and update unit is used to associate and map the electrical parameter data output by the full-scene electrical parameter acquisition unit with the non-electrical parameter data output by the multi-dimensional non-electrical parameter acquisition unit to construct a multimodal collaborative electrical fingerprint database containing "device independent features - device collaborative features - environmental location features - device type features - new device features". To quantify the correlation of multimodal features, the fingerprint association construction and update unit uses a feature fusion algorithm to calculate the feature association weights, as shown in the following formula: Among them, W i,j Let cov(F) be the association weight between the i-th electrical parameter feature and the j-th non-electrical parameter feature. i ,F j ) represents the electrical parameter characteristic F i With non-electrical parameter characteristics F j covariance, var(F i ) represents the electrical parameter characteristic F i The variance, var(F) j ) represents the non-electrical parameter characteristic F j The variance, α i Electrical parameter characteristics F i Importance coefficient, β j Non-electrical parameter characteristic F j Importance coefficient.

3. The intelligent power distribution system applicable to various household appliances according to claim 1, characterized in that, The AI ​​adaptive threshold control module includes: The multi-dimensional data acquisition unit is used to continuously collect household electricity consumption data, line impedance data, extreme operating condition data, load characteristic data associated with equipment type, and new equipment access data. The household electricity consumption data includes the time period, duration, and number of concurrent users for each appliance. The line impedance data is obtained based on the physical installation location of the equipment. The line impedance data includes line resistance and capacitance parameters; The extreme operating condition data includes power consumption parameters during periods of high summer temperatures, humid rainy seasons, sudden drops in grid voltage, and sudden increases in grid voltage. The load characteristic data associated with the device type includes typical power consumption parameter ranges for resistive loads, inductive loads, and capacitive loads. The new device access data includes the new device startup parameters and operating parameters.

4. The intelligent power distribution system applicable to various household appliances according to claim 3, characterized in that, The AI ​​adaptive threshold control module also includes: The integrated behavior analysis unit, based on the data collected by the multi-dimensional data acquisition unit, uses an improved deep learning algorithm to identify regular electricity consumption patterns, special electricity consumption scenarios, extreme operating conditions, and electricity consumption trends after new equipment is connected. It establishes a user electricity consumption behavior model that integrates line characteristics, extreme operating conditions, equipment types, and new equipment connection. The improved deep learning algorithm integrates CNN, LSTM, and attention mechanisms. The special electricity consumption scenarios include multiple devices operating concurrently during family gatherings and long-term electricity consumption during holidays. The model can quantify the influence coefficient of line impedance on protection thresholds, threshold correction coefficients under extreme operating conditions, load adaptation coefficients for different equipment types, and threshold adjustment coefficients for new equipment connection. The hierarchical threshold adjustment unit adjusts the load protection threshold according to the scenario information output by the fusion behavior analysis unit, with the priority order being extreme working conditions > unstable new device access scenarios > special power consumption scenarios > normal power consumption periods. For extreme working conditions, an additional threshold buffer margin is added. For multi-device concurrent scenarios, the threshold anti-interference margin is adjusted according to the collaborative interference feature sub-library data. For new device access scenarios, an initial adaptation threshold is set based on the threshold adjustment coefficient of the new device access, and it is gradually optimized after the new device is running stably. The graded threshold adjustment unit uses a dynamic threshold calculation formula to determine the load protection threshold, as follows: Among them, I th For the dynamically adjusted load protection threshold, I base W is the baseline protection threshold. k,env K represents the association weight between the k-th environmental feature and the electrical parameter feature. env,k W is the threshold influence coefficient of the k-th environmental feature. m,dev K represents the association weight between the m-th device type feature and the electrical parameter feature. dev,m Let f(m) be the threshold influence coefficient for the m-th device type feature; The dynamic self-optimization unit automatically extracts multimodal collaborative fingerprint features for the corresponding scenario and updates them to the database when it detects unexpected tripping, misjudgment caused by coupling interference, or new equipment access adaptation deviation. At the same time, it optimizes the line impedance influence coefficient, extreme condition correction coefficient, equipment type load adaptation coefficient, and new equipment access threshold adjustment coefficient to improve the threshold adjustment accuracy in subsequent scenarios. It also periodically updates the collaborative interference feature sub-library and equipment type feature sub-library based on newly added equipment combination data and new equipment data.

5. The intelligent power distribution system applicable to various household appliances according to claim 4, characterized in that, The cross-scenario intelligent early warning and self-healing module includes: The multi-dimensional risk monitoring unit is used to continuously monitor the trend of electrical harmonic distortion, the state of power grid harmonics, the degree of concurrent coupling interference of multiple devices, the trend of line aging impedance, and the operational stability after the connection of new equipment. The multi-dimensional risk monitoring unit uses a risk assessment algorithm to calculate the abnormal risk value, as shown in the following formula: Where R is the abnormal risk value, I real I is the actual load current. th The dynamic protection threshold is γ, which is the current fluctuation correction coefficient, and W is the W value. q,risk Let ΔX be the weight of the q-th risk factor. q The deviation of the q-th risk factor; The graded early warning generation unit generates a first-level aging early warning when the periodic change in the harmonic distortion rate of electrical appliances reaches a preset mild threshold, and generates a second-level emergency early warning when it reaches a preset severe threshold. A power grid anomaly warning is generated when the harmonic content of the power grid exceeds the harmonic standard. When the degree of coupling interference exceeds the safety threshold of the cooperative interference feature sub-library, a device cooperative interference warning is generated. A line aging warning is generated when the periodic change rate of line impedance exceeds the preset aging threshold. A new equipment adaptation warning is generated when the deviation of the operating parameters of the new equipment exceeds the preset adaptation range or the frequency of parameter fluctuation is too high.

6. The intelligent power distribution system applicable to various household appliances according to claim 5, characterized in that, The cross-scenario intelligent early warning and self-healing module also includes: The intelligent self-healing and linkage protection unit pushes maintenance suggestions for Level 1 aging warnings and automatically cuts off the power supply to corresponding electrical appliances for Level 2 emergency warnings; it activates the built-in active filter component and adjusts the protection threshold for power grid anomaly warnings; it cuts off the power supply to low-priority devices according to their importance for device interference warnings, with the order of importance being: devices maintaining basic living needs > devices meeting daily functional needs > non-essential devices that enhance convenience; it generates line maintenance plans for line aging warnings and reduces the upper limit of the power supply load on the corresponding line; it adjusts the protection threshold for new equipment for new equipment adaptation warnings, and if the adjustment is still unstable, it cuts off the power supply to the new equipment and pushes adaptation suggestions; the unit also generates a self-healing strategy library adapted to the current scenario based on historical fault handling data, real-time operating conditions, and equipment type characteristics, realizing autonomous optimization of fault handling; The fault tracing unit, when a power supply abnormality occurs, combines a multimodal collaborative fingerprint database, line impedance data, equipment type characteristic data, and new equipment access data to locate the root cause of the fault. The root cause of the fault includes electrical appliance faults, line aging, power grid problems, equipment collaborative interference, and new equipment adaptation problems, thus avoiding the shortcomings of traditional systems that can only judge the fault result but cannot trace the source.

7. The intelligent power distribution system applicable to various household appliances according to claim 6, characterized in that, The power distribution execution and privacy protection module includes: The priority power supply control unit is used to receive the hierarchical threshold instructions and new device access adaptation instructions output by the AI ​​adaptive threshold control module, and to perform power supply control according to the importance level of the devices. The importance level of the devices includes core devices, regular devices, auxiliary devices, and newly connected unstable devices. Core devices refer to devices that maintain the basic living needs of the family or ensure special health needs. Regular devices refer to devices that meet the daily functional needs of the family. Auxiliary devices refer to non-essential devices that improve the convenience of life. Under extreme working conditions, priority is given to ensuring the power supply of core devices. During the new device adaptation phase, the power supply priority of new devices is appropriately reduced. The self-healing execution unit is used to receive the self-healing strategy output by the cross-scenario intelligent early warning and self-healing module, execute actions such as filter activation, load shedding, line load adjustment, and new device threshold optimization, and record self-healing process data for strategy optimization. The data anonymization and privacy protection unit anonymizes the collected raw electricity consumption data, deleting user-identifiable information, which includes data on the correlation between specific electricity consumption time and family members' activities, and privacy information on the specific location of the device. Only appliance characteristic parameters, scene characteristic parameters, and device type characteristic parameters are extracted for system analysis. The appliance characteristic parameters include harmonics and power, and the scene characteristic parameters include environment and number of concurrent users. The status feedback unit feeds back the power supply control status, self-healing execution status, data desensitization status and new equipment operation status to the AI ​​adaptive threshold control module, cross-scenario intelligent early warning and self-healing module and equipment type dynamic adaptation module, forming a complete control closed loop. The priority power supply control unit uses a power supply priority allocation algorithm to determine the power supply sequence of the devices, as shown in the following formula: Among them, P dev Prioritize power supply to devices, S base Here, R represents the equipment baseline priority, R is the anomaly risk value, 0.3 is the risk correction factor, and W... t,use For the weight of the t-th feature, T t,use Let be the duration coefficient for the t-th feature used.

8. The intelligent power distribution system applicable to various household appliances according to claim 7, characterized in that, The device type dynamic adaptation module includes: The new device identification unit is used to determine the type of a newly connected device by detecting electrical signals when the device is connected, reading communication protocols, or using device type identifiers input by the user. The electrical signal detection includes starting impact feature detection, the communication protocol reading includes reading the model information of the smart device, and the device type includes kitchen functional equipment, audio-visual entertainment equipment, and health auxiliary equipment. The feature matching unit, based on the type of the new device and the initially collected electrical and non-electrical parameter features, matches the historical power consumption models of the same or similar types of devices from the device type feature sub-library of the multimodal collaborative electrical fingerprint database, and extracts the protection threshold parameters and adaptation environment parameters of similar devices as the initial reference data for the new device.

9. A smart power distribution system applicable to various household appliances according to claim 8, characterized in that, The device type dynamic adaptation module also includes: The dynamic adaptation and optimization unit continuously collects power consumption data under different operating conditions during the operation of new equipment. These different operating conditions include no-load, full-load, and standby states. The data is compared and analyzed with the initial reference data to generate a protection threshold adjustment curve specific to the new equipment. The power supply control strategy of the new equipment is gradually optimized until the operating parameters of the new equipment are stable within the adaptation range. At the same time, the stable characteristic data of the new equipment is updated to the multimodal collaborative electrical fingerprint database to enrich the equipment type characteristic sub-library. The dynamic adaptation optimization unit uses a threshold optimization algorithm to update the new device-specific protection threshold, and the specific formula is as follows: Among them, I new,th (k+1) represents the new device-specific protection threshold after the (k+1)th optimization, I new,th (k) represents the new device-specific protection threshold before the k-th optimization, R represents the abnormal risk value of the new device operation, and I new,real I represents the actual operating current of the new equipment. th 0.15 is the system dynamic protection threshold, and 0.15 is the optimization step size coefficient.

10. The intelligent power distribution system applicable to various household appliances according to claim 1, characterized in that, The multimodal collaborative fingerprint acquisition module, AI adaptive threshold control module, cross-scenario intelligent early warning and self-healing module, power distribution execution and privacy protection module, and device type dynamic adaptation module interact with each other through a dual-link communication method. The dual-link communication method includes wired communication and encrypted wireless communication. The encrypted wireless communication is used to deal with wired line fault scenarios. At the same time, an anti-interference algorithm is added to the wireless communication. The anti-interference algorithm includes frequency hopping technology and signal filtering to avoid interference from commonly used household wireless signals and electromagnetic devices on system communication, and to ensure the transmission stability of data from new device access and fault early warning data. Commonly used household wireless signals include short-range wireless communication signals and home network signals. Electromagnetic devices refer to household devices that generate electromagnetic radiation during operation.

Images