A remote control system and method for a smart waterproof socket that adapts to multiple scenarios

Abstract

This invention relates to the field of waterproof sockets and discloses a remote control system and method for intelligent waterproof sockets that adapts to multiple scenarios. The system includes acquiring multi-source sensing data of the socket deployment environment, constructing a multi-scenario adaptive input dataset by combining parameter fusion strategies and signal interference compensation mechanisms; identifying environmental features of the multi-scenario adaptive input dataset and extracting typical application modes under different scenarios using a socket state label extraction algorithm; dynamically analyzing the socket's operating status based on a state mapping relationship model and employing a remote control strategy optimization algorithm to identify the degree of matching between the current operating mode and the environmental scenario; introducing an intelligent decision mechanism to verify the multi-scenario control suggestion set before execution, and identifying potential control conflicts by combining historical operating data and real-time control commands; and managing the remote control behavior of the socket based on the optimized execution strategy. This invention has the advantage of improving the versatility of intelligent waterproof sockets.

Description

Technical Field

[0001] This invention relates to the field of waterproof sockets, specifically to a remote control system and method for intelligent waterproof sockets that adapts to multiple scenarios. Background Technology

[0002] Smart waterproof sockets are primarily used in outdoor or high-humidity environments, such as gardens, kitchens, and bathrooms. They typically feature a waterproof structure and basic remote control functionality. Currently, most smart waterproof sockets are designed for single-scene adaptation, such as being suitable only for specific humid environments like home balconies or bathrooms, lacking automatic recognition and function adjustment mechanisms for different scene conditions. For example, when sockets are deployed in open-air construction sites or agricultural irrigation areas, the power requirements, power fluctuation frequencies, and usage periods of the connected devices vary significantly. However, traditional sockets cannot automatically switch protection levels or remote control strategies based on external environmental parameters, easily leading to false alarms, delayed control, or malfunctions. Therefore, it is essential to design a multi-scene adaptable remote control system and method for smart waterproof sockets to improve their versatility. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a remote control system and method for intelligent waterproof sockets that adapts to multiple scenarios, which has the advantage of improving the versatility of intelligent waterproof sockets and solves the problems mentioned in the background technology.

[0004] To achieve the aforementioned goal of improving the versatility of smart waterproof sockets, this invention provides the following technical solution: a remote control method for a multi-scenario adaptable smart waterproof socket, comprising the following steps:

[0005] Acquire multi-source sensing data of the socket deployment environment, and combine parameter fusion strategy and signal interference compensation mechanism to construct multi-scenario adaptive input dataset;

[0006] Environmental features are identified in the multi-scenario adaptive input dataset, and typical application patterns in different scenarios are extracted using the socket status label extraction algorithm. A status mapping relationship model reflecting the relationship between scenario and device status is constructed.

[0007] Based on the state mapping relationship model, a remote control strategy optimization algorithm is used to dynamically analyze the socket operation status, identify the degree of matching between the current operation mode and the environmental scenario, and generate a multi-scenario control suggestion set.

[0008] An intelligent decision-making mechanism is introduced to verify the multi-scenario control suggestion set before execution, identify potential control conflicts by combining historical operation data and real-time control commands, and optimize the final execution strategy through a control path filtering mechanism.

[0009] Based on the optimized execution strategy, the remote control behavior of the socket is managed by issuing commands and outputting remote control execution reports.

[0010] Preferably, the process of constructing a multi-scenario adapted input dataset is as follows:

[0011] Multi-source sensing data from the deployment site of the sockets are collected, and the collected data is preprocessed. A feature filtering algorithm based on information gain is used to remove low-relevance and redundant dimensions.

[0012] By integrating historical operating environment information and typical abnormal scenario data, time series data are normalized and restructured through parameter fusion strategies.

[0013] A signal interference compensation mechanism is introduced, and wavelet denoising and Kalman filtering are used to jointly compensate for the noise signal generated in multi-source sensing acquisition.

[0014] The final result is an input dataset that is adaptable to multiple scenarios and can be used in real time.

[0015] Preferably, the process of environmental feature recognition for multi-scene adapted input datasets is as follows:

[0016] A multi-scale convolutional neural network is used to extract features from the normalized input data, and an attention mechanism is used to focus on environmental change factors.

[0017] By combining geographic location information with time series, a scene category mapping table is established;

[0018] For regions with ambiguous boundaries or multiple environments coexisting, fuzzy clustering methods are used to model the probability distribution of scene labels, and the recognition results are marked as scene feature labels.

[0019] Preferably, the process of extracting typical application patterns in different scenarios using the socket status label extraction algorithm is as follows:

[0020] Construct a joint index structure of socket usage logs and environmental scenario tags, and mine typical working modes through state transition graphs;

[0021] A socket status recognition network is used to encode the current, voltage, and power fluctuation patterns under different scenarios, and the stable usage interval is extracted as a representative state segment.

[0022] By combining the frequent pattern mining algorithm with the scene label clustering results, high-frequency state combinations are extracted and marked as typical application patterns.

[0023] Preferably, the process of constructing a state mapping relationship model that reflects the relationship between the scene and the device state is as follows:

[0024] Using typical application patterns as nodes, a socket state space graph is constructed, and the edge weights in the graph are weighted by environmental scene labels.

[0025] Graph neural networks are used to propagate features from the state-space graph to capture state transition patterns and higher-order relationships in different scenarios.

[0026] A state confidence estimation module is introduced to dynamically score the matching degree of each state node in combination with real-time environmental perception data.

[0027] The output reflects the state mapping relationship model between the scene and the device state.

[0028] Preferably, the process of dynamically analyzing the socket's operating status using a remote control strategy optimization algorithm is as follows:

[0029] Receive the state mapping relationship model and real-time environmental input, and perform multi-scenario compatibility analysis on the current socket operating status;

[0030] By modeling the socket state change process using dynamic Bayesian networks, the regulatory potential and intervention feasibility are evaluated based on the state transition probability.

[0031] By combining user preference strategies and energy-saving priorities, multiple control schemes are generated using a multi-objective reinforcement learning algorithm.

[0032] The output includes a set of multi-scenario control suggestions, including the timing, method, expected returns, and risk warnings.

[0033] Preferably, the process of identifying potential control conflicts by combining historical operating data with real-time control commands is as follows:

[0034] The current control recommendations are matched with the socket operation logs and historical user behavior sequences to perform feature comparison on potential conflict patterns;

[0035] Construct a control command conflict diagram and classify and analyze typical conflict types;

[0036] The conflict risk value of control commands is calculated based on the conflict probability estimation model, and high-risk commands are marked for intervention and early warning.

[0037] The analysis results are used as input conditions for the control path filtering mechanism.

[0038] Preferably, the process of optimizing the final execution strategy through the control path filtering mechanism is as follows:

[0039] Construct a multi-path control scheme set graph, mapping each path to a control chain consisting of instruction sequence, environmental scenario and desired state;

[0040] Use control chart reachability analysis to identify execution deadlocks or backtracking chains in the path;

[0041] By introducing a path cost function and comprehensively considering energy consumption, latency, conflict risk, and user satisfaction, the effective path set with the lowest cost is selected.

[0042] The final remote control execution strategy is generated after conflict avoidance, risk reduction, and goal optimization.

[0043] Preferably, the process of outputting the remote control execution report is as follows:

[0044] The control instructions in the execution strategy are parsed one by one, and the instruction issuance time, response delay and state transition effect are recorded.

[0045] The execution results are compared with the original control targets to generate a control deviation statistics table and a task completion analysis chart.

[0046] By combining real-time feedback with historical control records, a remote control execution report is generated.

[0047] A remote control system for a smart waterproof socket that adapts to multiple scenarios includes:

[0048] Environmental perception construction module: Collects and integrates multi-source environmental perception data to construct an input dataset with multi-scenario adaptability;

[0049] Scene recognition module: Identifies environmental features and extracts typical application patterns of sockets, and establishes a mapping relationship model between scenes and states;

[0050] Regulation strategy generation module: Analyzes the current state based on a mapping relationship model and generates remote regulation suggestions for multiple scenarios;

[0051] Conflict identification module: Verifies control recommendations before implementation, identifies potential control conflicts, and selects the optimal control path;

[0052] Command execution module: issues remote control commands and generates a control execution report that includes execution results and analysis.

[0053] Compared with the prior art, the present invention provides a remote control system and method for a smart waterproof socket that adapts to multiple scenarios, and has the following beneficial effects:

[0054] 1. By collecting multi-source data such as temperature and humidity, light intensity, noise, electromagnetic interference intensity and user interaction behavior, and using parameter fusion and signal interference compensation mechanisms, redundancy and noise interference in the sensing data are effectively eliminated, ensuring that the subsequent model has a stable input foundation and adaptability in various complex scenarios.

[0055] 2. By using multi-scale convolutional neural networks and attention mechanisms to accurately identify environmental change features, and combining the socket operation behavior patterns to construct state labels, a high-dimensional mapping relationship between the scene and the device state is further constructed, enabling the system to accurately understand the device operation rules under different environments, laying the foundation for personalized decision-making on control strategies.

[0056] 3. By using dynamic Bayesian networks and multi-objective reinforcement learning algorithms, combined with user preferences and energy-saving needs, the current state is analyzed in multiple dimensions and the optimal control suggestions are output, making the control scheme targeted, feasible and forward-looking, and improving the system's intelligent response capability and energy efficiency management level in complex environments.

[0057] 4. By combining conflict detection and path cost function screening, it can effectively identify and avoid problems such as time overlap, state contradiction and control conflict in control instructions, ensure the optimal execution path of the final strategy under multi-objective constraints, and reduce the risk of system miscontrol.

[0058] 5. By recording and evaluating the execution status of each instruction in real time and outputting a graphical control report, not only is the control process made transparent and traceable, but data support is also provided for subsequent system optimization and user behavior modeling, thereby improving the overall intelligence level of the system and user satisfaction. Attached Figure Description

[0059] Figure 1 This is a schematic diagram of the method of the present invention;

[0060] Figure 2 This is a schematic diagram of the structure of the present invention. Detailed Implementation

[0061] 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, and 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.

[0062] Example 1: Please refer to Figure 1 As shown in the figure, a remote control method for a multi-scenario adaptable smart waterproof socket according to an embodiment of the present invention includes the following steps:

[0063] S1: Acquire multi-source sensing data of the socket deployment environment, and construct a multi-scenario adaptive input dataset by combining parameter fusion strategy and signal interference compensation mechanism.

[0064] The process of constructing the multi-scenario adaptive input dataset in S1 is as follows:

[0065] Collect multi-source sensing data from the socket deployment site, including temperature and humidity, light intensity, environmental noise, electromagnetic interference intensity, and user interaction behavior records; deploy various sensors in the actual socket deployment environment, including temperature and humidity sensors, light sensors, environmental noise detection devices, and electromagnetic interference intensity sensors; collect user operation behavior data through the user interaction recording module; and collect sensor data in real time through edge devices and transmit it to the local data processing unit to ensure the timeliness and integrity of the data.

[0066] The collected data is preprocessed, and a feature filtering algorithm based on information gain is used to remove low-relevance redundant dimensions. The original multi-source data is converted in format and synchronized in time, missing values ​​are filled, outliers are removed, and the correlation between each feature and the target variable is calculated using the information gain algorithm. The data are sorted according to the information gain and redundant features with information gain below the threshold are removed to reduce feature dimensions and noise interference.

[0067] By integrating historical operating environment information and typical abnormal scenario data, time-series data are normalized and restructured using a parameter fusion strategy. Environmental parameter data and typical abnormal scenario data collected during past operations are extracted from historical databases. A fusion model is designed to match current data with historical data according to time and scenario labels to achieve cross-time-series fusion. A parameter fusion strategy is used to normalize multi-source time-series data to eliminate dimensional differences. The normalized data is then restructured, and the time-series data is divided into fixed windows to construct a multi-dimensional time series matrix that includes environmental features and user behavior, ensuring that the fused data can accurately reflect the dynamic changes under different scenarios.

[0068] A signal interference compensation mechanism is introduced, employing wavelet denoising and Kalman filtering to jointly compensate for noise signals generated during multi-source sensor acquisition. The acquired raw signals are analyzed in both the time and frequency domains to identify noise types and characteristics. The signal is decomposed using wavelet transform, and high-frequency noise components are effectively filtered out using multi-scale analysis methods, while preserving the main trends and details of the signal. A Kalman filter is applied to the denoised signal, and combined with the system state-space model, the signal is dynamically predicted and corrected, further improving the accuracy and stability of the signal. Through this joint processing, measurement errors caused by environmental interference, electromagnetic waves, etc., during multi-source sensor acquisition are significantly reduced.

[0069] The final result is an input dataset that is adaptable to multiple scenarios and can be used in real time.

[0070] S2: Environmental feature recognition is performed on the multi-scenario adaptation input dataset, and typical application patterns under different scenarios are extracted using the socket status label extraction algorithm to construct a state mapping relationship model that reflects the relationship between scenario and device status.

[0071] The process of environmental feature recognition for the multi-scene adaptive input dataset in S2 is as follows:

[0072] A multi-scale convolutional neural network is used to extract features from the normalized input data, and an attention mechanism is used to focus on significant environmental change factors. The normalized multi-dimensional time-series data is input into the multi-scale convolutional neural network, and multiple layers of convolutional kernels of different scales are designed to capture multi-level and multi-scale information of environmental features. The low-level convolution extracts local detail features, and the high-level convolution extracts global environmental patterns. Combined with the attention mechanism module, the weights of each feature channel and time period are dynamically adjusted to highlight significant changes in the environment, suppress irrelevant or noise interference signals, and output a weighted environmental feature vector that reflects the most discriminative environmental information in the current input data.

[0073] By combining geographic location information and time series, a scene category mapping table is established to achieve preliminary judgment of typical environmental scenes. The geographic location information of the socket is obtained using GPS or network positioning module, and the time information is synchronized to the feature data. Based on historical data and expert experience, typical environmental scene categories are predefined, and a scene category mapping table is established, which includes geographic coordinate range and time feature template. For the currently input environmental feature vector, combined with real-time geographic location and time label, matching and retrieval are performed in the mapping table. According to the matching degree and time period pattern, the preliminary classification judgment of the current scene is completed, and the corresponding scene category label is assigned.

[0074] For regions with ambiguous boundaries or multiple coexisting environments, fuzzy clustering is used to model the probability distribution of scene labels, ensuring the flexibility and continuity of environmental judgment. For regions with ambiguous geographical boundaries and complex overlapping environments, fuzzy clustering algorithm is used to perform cluster analysis on environmental features. Fuzzy clustering allows a single data point to belong to multiple categories with different membership degrees, outputting the probability distribution of each scene category, reflecting the mixed characteristics of the environment. The probability distribution model effectively solves the transitional zones and mixed environments that hard classification cannot cover, ensuring the continuity and robustness of the judgment results. The membership degree threshold is dynamically adjusted to support soft assignment of scene labels and multi-label recognition.

[0075] The recognition results are labeled as scene feature labels, which serve as the associated input for subsequent state label extraction and state mapping modeling.

[0076] The process of extracting typical application patterns under different scenarios using the socket status label extraction algorithm in S2 is as follows:

[0077] A joint index structure of socket usage logs and environmental scene tags is constructed, and typical working modes are mined through state transition graphs. Socket usage log data, including time series such as current, voltage, and power, as well as corresponding environmental scene feature tags, are collected. A joint index structure is designed to associate and store the usage logs with the environmental scene tags, enabling fast query and retrieval based on timestamps and scene categories. Based on the joint index, a socket state transition graph is constructed, where nodes represent different socket states, and edges represent the transition relationships between states and their probabilities. By traversing the state transition graph, path mining technology is used to identify frequently occurring state transition sequences, revealing typical working modes and usage behaviors.

[0078] A socket state recognition network is used to encode the current, voltage, and power fluctuation patterns under different scenarios and extract stable usage intervals as representative state segments. The socket state recognition network is designed with input time-series data such as current, voltage, and power. The network encodes the features of electrical parameter fluctuations under different scenarios, learns to capture typical state change patterns and dynamic features, and uses the sliding window technique to divide the time-series data, identify and extract continuous and stable usage intervals as representative state segments to avoid interference from short-term anomalies or noise. The state recognition network is trained or fine-tuned for different scenarios to improve the accuracy of scene-specific pattern recognition.

[0079] By combining frequent pattern mining algorithms with scene label clustering results, high-frequency state combinations are extracted and marked as typical application patterns. The frequent pattern mining algorithm is used to analyze the state fragment sequence to identify frequently occurring state combinations and their temporal association rules. The frequent state combinations are combined with the scene label clustering results to associate typical scenes with their corresponding high-frequency usage patterns. The confidence and support of the mining results are evaluated to select representative and stable high-frequency patterns. The selected state combinations are marked as typical application patterns to guide the formulation of smart socket management, anomaly detection, and optimization strategies.

[0080] The process of constructing the state mapping relationship model reflecting the association between the scene and the device state in S2 is as follows:

[0081] Using typical application patterns as nodes, a socket state space graph is constructed, and the edge weights in the graph are weighted by environmental scenario labels. The extracted typical application patterns are used as nodes in the state space graph, and each node represents a device operating state or a combination of states. Based on the socket state transition log, transition edges between state nodes are established. Each edge represents a possible transition from one state to another. Combined with environmental scenario labels, each state transition edge is assigned a weight. The weight is adjusted based on the frequency or importance of the transition in the corresponding scenario. The edge weights dynamically reflect the impact of environmental conditions on state transitions, enhancing the expressive ability of the state space graph to adapt to multiple scenarios.

[0082] Graph neural networks (GNNs) are used to propagate features from the state space graph, capturing state transition patterns and higher-order relationships under different scenarios. A GNN model is designed, taking node features and weighted edge information as inputs. Feature propagation is achieved through multi-layer graph convolution operations, fusing the states of neighboring nodes and their environmental weighted influences, and learning higher-order dependencies and state transition patterns between nodes. The GNN model captures complex state interaction patterns and scene-driven change trends, improving the expressive power and prediction accuracy of state mapping. During training, historical data is used for supervision to optimize model parameters to accurately simulate device state dynamics.

[0083] A state confidence estimation module is introduced to dynamically score the matching degree of each state node in conjunction with real-time environmental perception data. The state confidence estimation module is designed to take real-time environmental perception data and current state node features as input, and use machine learning or Bayesian inference methods to calculate the matching confidence score of each state node with the current environmental conditions. The activation weights of the nodes in the state space graph are dynamically adjusted so that the model can reflect the probability and reliability of the state in the current environment in real time. The confidence score helps the state mapping model to make more accurate state predictions and anomaly detection.

[0084] The output reflects the state mapping relationship model between the scene and the device state.

[0085] Example 2: Figure 1 As shown, a remote control method for a smart waterproof socket adaptable to multiple scenarios also includes the following steps:

[0086] S3: Based on the state mapping relationship model, a remote control strategy optimization algorithm is used to dynamically analyze the socket operation status, identify the degree of matching between the current operation mode and the environmental scenario, and generate a multi-scenario control suggestion set.

[0087] The process of dynamically analyzing the socket's operating status using the remote control strategy optimization algorithm in S3 is as follows:

[0088] The system receives the state mapping relationship model and real-time environmental input, and performs multi-scenario compatibility analysis on the current socket operating status. It receives the current socket operating status and its scenario-related characteristics output from the state mapping relationship model in real time, obtains real-time environmental input data from multiple sources of sensors, and combines the state mapping model and real-time environmental data to analyze the compatibility and adaptability of the current operating status in different scenarios, and assesses whether the device status matches the environmental conditions or whether there are potential risks.

[0089] By modeling the state change process of a socket using a dynamic Bayesian network, the potential for regulation and the feasibility of intervention are evaluated based on the state transition probabilities. A dynamic Bayesian network model is constructed, using state nodes and transition edges in the state mapping relationship model as variable nodes and dependencies. Using historical state transition data and real-time collected data, the transition probability distribution between states is calculated. Based on the current state and transition probabilities, future state trends and possible abnormal or unstable states are predicted. Combined with equipment operation constraints and safety thresholds, the regulatory potential and the feasibility and necessity of intervention on each state transition path are evaluated.

[0090] Combining user preference strategies and energy-saving priorities, a multi-objective reinforcement learning algorithm is used to generate multiple control schemes. User preferences and energy-saving priorities are used as reward function design elements in multi-objective reinforcement learning. A reinforcement learning environment model is designed, with the state space including the current state of the device and the state of the environment, and the action space containing different control strategies. The multi-objective reinforcement learning algorithm is trained to optimize energy consumption and response performance while ensuring device safety and user experience. Multiple candidate control schemes are generated, covering different control timings, methods, and expected effects, and supporting flexible selection based on real-time needs.

[0091] The output includes a set of multi-scenario control suggestions, including the timing, method, expected returns, and risk warnings.

[0092] S4: Introducing an intelligent decision-making mechanism to verify the multi-scenario control suggestion set before execution, combining historical operation data and real-time control instructions to identify potential control conflicts, and optimizing the final execution strategy through a control path filtering mechanism.

[0093] The process of identifying potential control conflicts by combining historical operating data and real-time control commands in S4 is as follows:

[0094] The current control suggestions are matched with the socket operation logs and historical user behavior sequences to compare features of potential conflict patterns; the currently generated remote control suggestions are collected, including control time, action type and target status; the socket operation logs and historical user behavior sequences are obtained, including previous control command execution records and corresponding device response status; the current control suggestions and historical control behaviors are time-series aligned to match possible overlaps or contradictions in time or action; a conflict feature vector is designed, covering dimensions such as time conflict, action conflict, and overlapping control rights, and feature comparison analysis is performed.

[0095] Construct a control command conflict graph and classify and analyze typical conflict types such as time overlap, state inconsistency, and control struggle. Use control commands as nodes in the graph to establish a conflict graph structure, with edges representing potential conflict relationships between two commands. According to the conflict type, the edges are classified into categories: time overlap conflict, state inconsistency conflict, and control struggle conflict.

[0096] Calculate the conflict risk value of control commands and mark high-risk commands for intervention and early warning. The formula is as follows:

[0097] R i =α*P time (i)+β*P state (i)+γ*P ctrl (i)

[0098] In the formula, R i Let P be the conflict risk value of the i-th control instruction. rime (i) represents the probability of time overlap and conflict, P state (i) represents the probability of state inconsistency conflict, P ctrl (i) represents the probability of control contest, and α, β, and γ are weighting coefficients;

[0099] Input control command characteristics and their historical conflict frequency are combined with current environmental variables to conduct dynamic risk assessment, calculate the conflict risk value of each control recommendation command, reflect the possibility and severity of potential conflicts, mark commands with conflict risk values ​​exceeding the threshold as high risk, trigger the intervention early warning mechanism, notify the system administrator or automatically start the conflict resolution process;

[0100] The analysis results are used as input conditions for the control path screening mechanism. Conflict risk markers and conflict graph structure information are input into the control path screening module. The control path screening mechanism uses this information to eliminate or adjust high-risk control schemes, prioritizes control paths with low conflict risks, and dynamically optimizes the final control strategy execution sequence by combining control effects and user preferences. This ensures the safety and stability of the remote control process and reduces the risk of malfunctions and equipment failures.

[0101] The process of optimizing the final execution strategy through the control path filtering mechanism in S4 is as follows:

[0102] Construct a multi-path control scheme set graph, mapping each path to a control chain consisting of a sequence of instructions, environmental scenarios, and desired states; collect multiple candidate control schemes, each containing a series of control instructions covering different time points and environmental scenarios, abstract these control schemes into path nodes, and construct a control scheme set graph, where each path represents a possible control execution chain, and each node in the path consists of control instructions, corresponding environmental scenario characteristics, and the desired device state, forming a complete control chain description;

[0103] The control chart reachability analysis method is used to identify potential execution deadlocks or backtracking chains in the path; the reachability analysis algorithm is used to detect the execution dependencies and transition relationships between nodes in each path, identify path nodes with deadlocks, i.e., circular dependencies in the instruction execution order that cause execution to stall, check whether there are backtracking chains in the path, i.e., repeated state switching or multiple undoing operations that affect execution efficiency and system stability, and mark the problematic path nodes.

[0104] A path cost function is introduced, comprehensively considering energy consumption, latency, conflict risk, and user satisfaction, to select the set of effective paths with the lowest cost; the path cost function is designed to include the following weight indicators:

[0105] Energy cost: The estimated total energy consumption for executing all control commands under this path;

[0106] Delay cost: The time required from the current moment to complete the last instruction in the path;

[0107] Conflict risk cost: The cumulative risk score of high-risk conflict instructions included in the path;

[0108] User satisfaction cost: a score of the effectiveness of regulation based on user preferences and historical feedback.

[0109] Cost calculation and sorting are performed on all paths, and the set of paths with the lowest cost and no deadlock or backoff issues is selected first. Multi-objective optimization method is used to ensure that the control scheme is balanced among multiple objectives.

[0110] The final remote control execution strategy is generated after conflict avoidance, risk reduction, and goal optimization.

[0111] S5: Based on the optimized execution strategy, it manages the remote control behavior of the socket by issuing commands and outputs a remote control execution report.

[0112] The process of outputting the remote control execution report in S5 is as follows:

[0113] The system parses each control command in the execution strategy, records the command issuance time, response delay, and state transition effect; it automatically parses each control command contained in the final executed remote control strategy, records the actual issuance timestamp of each command, collects the response time of the device to the command, calculates the response delay, i.e. the time difference between the command issuance and the device execution, monitors changes in device status, records whether the state transition after command execution achieves the expected effect, and collects key performance indicators.

[0114] The execution results are compared with the original control targets to generate a control deviation statistics table and a task completion analysis chart. The actual state data after the execution of each instruction is summarized and compared with the preset expected state in the remote control strategy. Control deviations are calculated, including response delay deviations, state achievement deviations, and abnormal fluctuations. The completion rate of control tasks is statistically analyzed, and a task completion analysis chart is generated to intuitively display the achievement of each control target, identify problems or deviations in the control execution, and facilitate subsequent optimization and fault diagnosis.

[0115] By combining real-time feedback with historical control records, a remote control execution report is generated.

[0116] Example 3: Please refer to Figure 2 As shown, a remote control system for a smart waterproof socket that adapts to multiple scenarios includes:

[0117] Environmental perception construction module: Collects and integrates multi-source environmental perception data to construct an input dataset with multi-scenario adaptability;

[0118] Scene recognition module: Identifies environmental features and extracts typical application patterns of sockets, and establishes a mapping relationship model between scenes and states;

[0119] Regulation strategy generation module: Analyzes the current state based on a mapping relationship model and generates remote regulation suggestions for multiple scenarios;

[0120] Conflict identification module: Verifies control recommendations before implementation, identifies potential control conflicts, and selects the optimal control path;

[0121] Command execution module: issues remote control commands and generates a control execution report that includes execution results and analysis.

[0122] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0123] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A remote control method for a smart waterproof socket adaptable to multiple scenarios, characterized in that, Includes the following steps: Acquire multi-source sensing data of the socket deployment environment, and combine parameter fusion strategy and signal interference compensation mechanism to construct multi-scenario adaptive input dataset; Environmental features are identified in the multi-scenario adaptive input dataset, and typical application patterns in different scenarios are extracted using the socket status label extraction algorithm. A status mapping relationship model reflecting the relationship between scenario and device status is constructed. The process of constructing a state mapping model that reflects the relationship between scene and device state is as follows: Using typical application patterns as nodes, a socket state space graph is constructed, and the edge weights in the graph are weighted by environmental scene labels. Graph neural networks are used to propagate features from the state-space graph to capture state transition patterns and higher-order relationships in different scenarios. A state confidence estimation module is introduced to dynamically score the matching degree of each state node in combination with real-time environmental perception data. The output reflects a state mapping model that connects the scene and the device state. Based on the state mapping relationship model, a remote control strategy optimization algorithm is used to dynamically analyze the socket operation status, identify the degree of matching between the current operation mode and the environmental scenario, and generate a multi-scenario control suggestion set. An intelligent decision-making mechanism is introduced to verify the multi-scenario control suggestion set before execution, identify potential control conflicts by combining historical operation data and real-time control commands, and optimize the final execution strategy through a control path filtering mechanism. Based on the optimized execution strategy, the remote control behavior of the socket is managed by issuing commands and outputting remote control execution reports.

2. The remote control method for a multi-scenario adaptable smart waterproof socket according to claim 1, characterized in that, The process of constructing a multi-scenario adapted input dataset is as follows: Multi-source sensing data from the deployment site of the sockets are collected, and the collected data is preprocessed. A feature filtering algorithm based on information gain is used to remove low-relevance and redundant dimensions. By integrating historical operating environment information and typical abnormal scenario data, time series data are normalized and restructured through parameter fusion strategies. A signal interference compensation mechanism is introduced, and wavelet denoising and Kalman filtering are used to jointly compensate for the noise signal generated in multi-source sensing acquisition. The final result is an input dataset that is adaptable to multiple scenarios and can be used in real time.

3. The remote control method for a multi-scenario adaptable smart waterproof socket according to claim 2, characterized in that, The process of environmental feature recognition for multi-scene adaptive input datasets is as follows: A multi-scale convolutional neural network is used to extract features from the normalized input data, and an attention mechanism is used to focus on environmental change factors. By combining geographic location information with time series, a scene category mapping table is established; For regions with ambiguous boundaries or multiple environments coexisting, fuzzy clustering methods are used to model the probability distribution of scene labels, and the recognition results are marked as scene feature labels.

4. The remote control method for a multi-scenario adaptable smart waterproof socket according to claim 3, characterized in that, The process of extracting typical application patterns in different scenarios using the socket status label extraction algorithm is as follows: Construct a joint index structure of socket usage logs and environmental scenario tags, and mine typical working modes through state transition graphs; A socket status recognition network is used to encode the current, voltage, and power fluctuation patterns under different scenarios, and the stable usage interval is extracted as a representative state segment. By combining the frequent pattern mining algorithm with the scene label clustering results, high-frequency state combinations are extracted and marked as typical application patterns.

5. The remote control method for a multi-scenario adaptable smart waterproof socket according to claim 4, characterized in that, The process of dynamically analyzing the socket's operating status using a remote control strategy optimization algorithm is as follows: Receive the state mapping relationship model and real-time environmental input, and perform multi-scenario compatibility analysis on the current socket operating status; By modeling the socket state change process using dynamic Bayesian networks, the regulatory potential and intervention feasibility are evaluated based on the state transition probability. By combining user preference strategies and energy-saving priorities, multiple control schemes are generated using a multi-objective reinforcement learning algorithm. The output includes a set of multi-scenario control suggestions, including the timing, method, expected returns, and risk warnings.

6. The remote control method for a multi-scenario adaptable smart waterproof socket according to claim 5, characterized in that, The process of identifying potential control conflicts by combining historical operating data with real-time control commands is as follows: The current control recommendations are matched with the socket operation logs and historical user behavior sequences to perform feature comparison on potential conflict patterns; Construct a control command conflict diagram and classify and analyze typical conflict types; Calculate the conflict risk value of control commands and mark high-risk commands for intervention and early warning; The analysis results are used as input conditions for the control path filtering mechanism.

7. The remote control method for a multi-scenario adaptable smart waterproof socket according to claim 6, characterized in that, The process of optimizing the final execution strategy through the control path filtering mechanism is as follows: Construct a multi-path control scheme set graph, mapping each path to a control chain consisting of instruction sequence, environmental scenario and desired state; Use control chart reachability analysis to identify execution deadlocks or backtracking chains in the path; By introducing a path cost function and comprehensively considering energy consumption, latency, conflict risk, and user satisfaction, the effective path set with the lowest cost is selected. The final remote control execution strategy is generated after conflict avoidance, risk reduction, and goal optimization.

8. A remote control method for a multi-scenario adaptable smart waterproof socket according to claim 7, characterized in that, The process of generating a remote control execution report is as follows: The control instructions in the execution strategy are parsed one by one, and the instruction issuance time, response delay and state transition effect are recorded. The execution results are compared with the original control targets to generate a control deviation statistics table and a task completion analysis chart. By combining real-time feedback with historical control records, a remote control execution report is generated.

9. A remote control system for a smart waterproof socket that adapts to multiple scenarios, applied to the method described in any one of claims 1-8, characterized in that, include: Environmental perception construction module: Collects and integrates multi-source environmental perception data to construct an input dataset with multi-scenario adaptability; Scene recognition module: Identifies environmental features and extracts typical application patterns of sockets, and establishes a mapping relationship model between scenes and states; Regulation strategy generation module: Analyzes the current state based on a mapping relationship model and generates remote regulation suggestions for multiple scenarios; Conflict identification module: Verifies control recommendations before implementation, identifies potential control conflicts, and selects the optimal control path; Command execution module: issues remote control commands and generates a control execution report that includes execution results and analysis.

Images