Remote control terminal and method for flexibly regulating and controlling load

Abstract

The invention relates to a flexible regulation and control load remote control terminal and method, and belongs to the technical field of power system automation and demand side management.The control terminal comprises an intelligent communication adaptation module, a coupling modeling module, a multi-scene self-adaption regulation and control module and a user interaction feedback module which are interconnected through an internal data bus; real-time interaction and cooperative work of data are realized; according to the invention, the breakthrough of the regulation and control mode from static presetting to dynamic optimization is realized, full-protocol compatibility and high-reliability transmission are realized, the real-time performance and interaction depth of user side response are improved, and the refinement and precision of control are improved.

Description

Technical Field

[0001] This invention belongs to the field of power system automation and demand-side management technology, specifically involving a flexible and controllable load remote control terminal and control method that integrates the Internet of Things, edge computing and artificial intelligence. It is suitable for distributed load aggregation and control in industrial, commercial and residential scenarios, and supports new power system applications such as virtual power plants and demand response. Background Technology

[0002] Patent application number CN2024106824259, invention title: Air conditioning control and debugging method and system based on flexible load control terminal, describes:

[0003] 1) Attribute Information Data Set Technology: The control terminal presets an attribute information data set, storing different attribute type information (such as air conditioner brand type, control address of control variables, communication format, air conditioner unit number, etc.) and corresponding data identifiers for each attribute type required for air conditioner control. By using the data identifiers in the instructions, the corresponding attribute information is quickly searched and matched to determine the downstream air conditioner unit attributes.

[0004] 2) Communication Protocol Conversion Technology: The State Grid 698 protocol frames are converted into the Modbus protocol format of the air conditioning unit to enable communication between the control terminal and the air conditioning unit. The downlink communication interface of the control terminal supports multiple interface methods such as 485, CAN, HPLC&RF, and LoRa, and can be directly connected to the control interface of the air conditioning unit's communication board, or connected via a communication card.

[0005] 3) Multi-mode control technology: The air conditioning unit can be controlled in various ways, with controllable parameter variables that can be set and relevant data from the air conditioning unit can be acquired and transmitted. When executing the control strategy, there are two modes: Mode 1 (continuously adjusting the air conditioning set temperature according to the set target power value, and stopping after the target is reached) and Mode 2 (adjusting according to the set temperature range during the set high power load time, and exiting after the time ends). It also supports issuing control commands individually for flexible control.

[0006] Patent application number CN2023113494348, invention title: A smart energy unit load flexible adjustment system, describes:

[0007] 1) Network transmission technology: The first network transmission module (fiber optic, power wireless private network, 5G channel and wireless virtual private network) is used to realize bidirectional communication between the smart energy unit and the power load management system, which is used to upload equipment operation data and receive control commands; the second network transmission module (RS-485, Ethernet, LoRa, Bluetooth, HPLC and HPLC+HRF dual mode) is responsible for the connection between the smart energy unit and the user load system to ensure data interaction.

[0008] 2) Load regulation technology: The air conditioning resource regulation submodule in the load flexible regulation module can read parameters such as the air conditioning set temperature and operating status to realize flexible control of system soft start and stop and temperature regulation; the smart energy unit issues power control commands to the equipment according to the regulation needs to limit the operating power of the equipment and achieve the purpose of load regulation; closed-loop regulation is realized by monitoring the power supply circuit of the air conditioning equipment to ensure the regulation effect.

[0009] 3) Data processing and analysis technology: The smart energy unit collects and monitors load data in real time, and analyzes and processes various data such as the current operating power, setting mode, voltage and current curves of the collected air conditioning equipment; it has the ability to monitor and identify adjustable resources on the user side in real time, and classifies and manages them according to response speed, adjustability and other factors to explore their potential; it can also carry out equipment-level and system-level load safety analysis to realize early warning of abnormal power consumption and diagnosis of hidden dangers.

[0010] While the aforementioned existing technologies have made some progress in the remote control of flexible and adjustable loads, they still have some shortcomings:

[0011] 1) Limited Control Mode: Most existing power load management terminals can only regulate electricity consumption through administrative means, such as simply pulling loads according to a pre-set fixed sequence. This fails to meet the diverse energy service needs of users and makes it difficult to achieve refined and differentiated control of different types of flexible loads. This "one-size-fits-all" approach not only fails to fully tap the adjustment potential on the user side but may also cause power outages for some load circuits that are sensitive to power quality, leading to user dissatisfaction.

[0012] 2) Limited Communication Capabilities: Existing equipment in flexible load control systems faces numerous challenges in data transmission. For instance, when establishing data channels for controlling equipment like air conditioning units, data frames contain a large amount of air conditioning parameter information, resulting in excessive data volume. This often necessitates frame-by-frame transmission, which easily leads to data errors during framing, transmission, and parsing, reducing the reliability of debugging. Furthermore, the overall flexible load system lacks an efficient and convenient operating environment for remote communication, failing to meet the needs of real-time remote program upgrades, parameter modifications, and online monitoring and scheduling.

[0013] 3) Weak user-side response capability: The existing power load control system makes it almost impossible for end users to actively respond to power rationing requirements. Even if the load control terminal issues a "notification" message, it often relies on property management personnel to manually notify end users. This process has a significant delay and poor information transmission, which can easily lead to forced load rationing and cause inconvenience to users' normal production and life.

[0014] 4) Low level of control precision: Existing technologies struggle to achieve precise control at the load equipment level and cannot regulate power consumption for specific electrical equipment within a user's premises. For example, in the industrial sector, participation in grid interaction typically involves rigid regulation, requiring the shutdown of some equipment or production lines, which significantly impacts the daily production of electricity customers, resulting in low customer willingness to participate. Summary of the Invention

[0015] In view of the problems existing in the prior art, the present invention proposes a flexible load control remote control terminal, which solves the technical problems of single control mode, limited communication capability, weak user-side response capability and low control precision in the prior art.

[0016] Another objective of this invention is to provide a method for remote control of flexible load regulation.

[0017] The present invention adopts the following technical solution:

[0018] A flexible load control remote control terminal includes an intelligent communication adaptation module, a coupling modeling module, a multi-scenario adaptive control module, and a user interaction feedback module. The modules are interconnected through an internal data bus to achieve real-time data interaction and collaborative work.

[0019] The intelligent communication adaptation module collects load communication signals to achieve multi-protocol adaptive conversion and reliable transmission in extreme environments;

[0020] The coupled modeling module collects multi-dimensional parameters of load, environment, and users, and establishes a coupled model of multi-dimensional parameters of load, environment, and users.

[0021] The multi-scenario adaptive control module receives the prediction result data from the coupled modeling module, identifies the current scenario, and generates a dynamically optimized control strategy.

[0022] The user interaction feedback module is based on the dynamic optimization control strategy generated by the multi-scenario adaptive control module to construct a feedback and adjustment closed loop.

[0023] Furthermore, the intelligent communication adaptation module has the following specific structure: the input end of the protocol adaptive conversion unit is connected to the user load system, and the output end of the protocol adaptive conversion unit is connected to the microprocessor through an internal data bus; the protocol adaptive conversion unit includes a protocol feature library and a fuzzy matching algorithm module; the protocol adaptive conversion unit extracts the feature values ​​of frame structure features and instruction interaction features by real-time parsing of the communication signals of the access user load, matches them with the protocol feature library, and automatically generates conversion rules to establish a communication connection.

[0024] The first network transmission module is connected to the power grid master station, and the second network transmission module is connected to the user-side local equipment. The anti-interference transmission unit communicates bidirectionally with the first network transmission module and the second network transmission module to realize signal transmission between the terminal and the power grid master station and the user side. The anti-interference transmission unit includes a dynamic channel evaluation subunit, a beam tracking antenna, and an electromagnetic shielding layer.

[0025] Furthermore, the coupled modeling module includes a multidimensional data acquisition subunit and a coupled model training subunit. The multidimensional data acquisition subunit collects multidimensional data on load, environment, and users through sensors and inputs the collected multidimensional data into the coupled model training subunit. The coupled model training subunit is based on a long short-term memory network (LSTM) and trains the model with historical data on load, environment, user parameters, and control effects. It outputs the predicted value of load adjustable potential and the environmental interference coefficient to the multi-scenario adaptive control module.

[0026] Furthermore, the multi-dimensional data acquisition subunit acquires load parameters by using a flexible current sensor to collect real-time load power, voltage curves, and start-stop loss data; and reads inherent equipment parameters through a load digital twin interface.

[0027] The multi-dimensional data acquisition subunit collects environmental parameters by using distributed temperature and humidity sensors to collect ambient temperature and humidity; and by using electromagnetic sensors to monitor grid voltage fluctuations and new energy output data.

[0028] The multidimensional data acquisition subunit acquires user parameters by obtaining user comfort preferences and real-time demand data through the user interaction feedback module.

[0029] Furthermore, the coupled model training subunit includes,

[0030] The historical data acquisition module collects minute-level data samples over a continuous period of 12 months. Each sample contains 18-dimensional input features and 2-dimensional labels.

[0031] The data preprocessing module uses forward imputation to fill in missing values ​​for sporadic sensor disconnection data; it performs Min-Max normalization on the input features and scales the labels according to the actual physical range; and it divides the data into training, validation, and test sets according to time sequence to maintain temporal continuity.

[0032] The model training module has an initial learning rate of 0.001, which decays by 10% every 50 rounds.

[0033] Loss function formula:

[0034]

[0035] Where y1 represents the actual adjustable potential. y1 is the predicted value; y2 is the actual environmental interference coefficient. This is a predicted value;

[0036] The training run consists of 200 rounds, and the process stops if the validation set loss does not decrease for 10 consecutive rounds.

[0037] The model tuning module uses a grid search method to optimize the number of neurons in the LSTM layer and adds a Dropout layer to suppress overfitting.

[0038] Furthermore, the multi-scenario adaptive control module includes a scenario identification subunit, a strategy generation subunit, and an execution monitoring subunit. The input end of the scenario identification subunit is connected to the coupled modeling module and the power grid master station, and the output end of the scenario identification subunit is connected to the strategy generation subunit. The strategy generation subunit sends instructions to the intelligent communication adaptation module and the execution monitoring subunit through the data bus. The intelligent communication adaptation module sends instructions to the load. The execution monitoring subunit feeds back deviation data to the strategy generation subunit.

[0039] Furthermore, the scene recognition subunit presets 5 core scenes and automatically identifies the current scene by matching the power grid status, user commands, and environmental parameters in real time; the execution monitoring subunit compares the control commands with the actual load response in real time, and if the deviation is greater than the set value, it triggers a secondary correction to achieve closed-loop control.

[0040] Furthermore, the strategy generation subunit is based on an improved particle swarm optimization algorithm, with the goals of grid stability, user satisfaction, and minimum energy consumption, and combines the adjustable potential output of the coupled model to generate dynamic control curves.

[0041] Particle velocity update formula:

[0042]

[0043] Particle position update formula:

[0044]

[0045] In the formula,

[0046] The velocity of the i-th particle in the (t+1)-th generation in d-dimensional space;

[0047] The position of the i-th particle in the (t+1)-th generation in d-dimensional space;

[0048] w(t): Dynamic inertia weight, which is adaptively adjusted with the number of iterations;

[0049] c1c2: Learning factors, representing the weights of a particle learning from its own historical best and the group's best, respectively;

[0050] r1r2: Random numbers within the interval [0,1], increasing the randomness of the search;

[0051] pbest i,d The historical best position of the i-th particle;

[0052] gbest d The global optimal position of the entire particle swarm;

[0053] λ: Scene weighting factor, dynamically determined based on the current scene to enhance the scene's impact on the control target;

[0054] S d Scene adaptation deviation value: the deviation between the current particle position and the scene requirements in the d-th dimension space.

[0055] Furthermore, the user interaction feedback module includes a feedback input subunit and a demand transformation subunit. The input end of the feedback input subunit is connected to the user terminal, and the output end of the feedback input subunit is connected to the multi-scenario adaptive control module via the demand transformation subunit.

[0056] This invention discloses a method for remote control of flexible loads, comprising the following steps:

[0057] The system collects load communication signals, matches them with a protocol feature library, generates conversion rules, and establishes communication connections. Simultaneously, it collects initial data on load, environment, and users to establish a multi-dimensional parameter coupling basic model of load, environment, and users.

[0058] Collect real-time load, environment, and user multi-dimensional parameters, update the multi-dimensional parameter coupled basic model based on new data, and output prediction results;

[0059] Based on the predicted data, the current scenario is identified, and a dynamically optimized control strategy is generated.

[0060] Based on the dynamic optimization control strategy generated by the multi-scenario adaptive control module, a feedback and adjustment closed loop is constructed.

[0061] The present invention has the following beneficial effects:

[0062] 1. This invention constructs a multi-scenario adaptive control framework, realizing a breakthrough from static preset to dynamic optimization of control mode. It solves the problem that fixed parameter or single-dimensional control logic cannot dynamically adapt control strategies according to the real-time state of the power grid, differences in load type, and changes in scenarios, and overcomes the limitations of a single control mode, thereby improving control flexibility.

[0063] 2. This invention enhances the multi-protocol adaptive conversion capability and extreme environment anti-interference capability of the communication module, achieving full protocol compatibility and high-reliability transmission. It solves the problem that existing technologies rely on fixed mapping for communication protocol conversion and have poor compatibility with new non-standard protocols or hybrid protocol scenarios. At the same time, it solves the problem of insufficient communication stability in complex environments such as strong electromagnetic interference and signal blockage, breaking through the bottleneck of communication capability adaptation and anti-interference.

[0064] 3. This invention constructs a closed-loop interactive mechanism of user feedback and strategy adjustment, which enables dynamic balance between control objectives and user experience. It solves the problems in the prior art where users passively receive information and lack convenient real-time feedback channels, and the control strategies do not fully integrate the dynamic needs of users, thereby improving the real-time response and interaction depth of the user side.

[0065] 4. This invention establishes a multi-dimensional coupling model of load, environment, and user, realizing refined control based on precise perception. It solves the problems of existing technologies where the control accuracy depends on simple parameter settings, fails to combine the coupling relationship between load physical characteristics and environmental parameters, and leads to deviations between the control effect and actual needs, thereby improving the refinement and accuracy of control. Attached Figure Description

[0066] Figure 1 This is a schematic block diagram of the overall system structure of the present invention;

[0067] Figure 2 This is a schematic block diagram of the intelligent communication adapter module structure of the present invention;

[0068] Figure 3 This is a schematic block diagram illustrating the multi-scenario adaptive control workflow of the present invention. Detailed Implementation

[0069] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0070] Example 1

[0071] A flexible load control remote control terminal includes an intelligent communication adaptation module, a coupling modeling module, a multi-scenario adaptive control module, and a user interaction feedback module. The modules are interconnected through an internal data bus to achieve real-time data interaction and collaborative work.

[0072] The intelligent communication adaptation module collects load communication signals to achieve multi-protocol adaptive conversion and reliable transmission in extreme environments;

[0073] The coupled modeling module collects multi-dimensional parameters of load, environment, and users, and establishes a coupled model of multi-dimensional parameters of load, environment, and users.

[0074] The multi-scenario adaptive control module receives the prediction result data from the coupled modeling module, identifies the current scenario, and generates a dynamically optimized control strategy.

[0075] The user interaction feedback module is based on the dynamic optimization control strategy generated by the multi-scenario adaptive control module to construct a feedback and adjustment closed loop.

[0076] The intelligent communication adaptation module has the following structure: the input end of the protocol adaptive conversion unit is connected to user load systems such as air conditioners and charging piles via hardware interfaces such as RS-485 / Ethernet; the output end of the protocol adaptive conversion unit is connected to a microprocessor via an internal data bus. The protocol adaptive conversion unit includes a protocol feature library and a fuzzy matching algorithm module. The protocol feature library pre-stores feature values ​​of 15 mainstream protocols such as Modbus / 698 / ZigBee and non-standard protocols. The protocol adaptive conversion unit extracts the feature values ​​of frame structure features and instruction interaction features from the communication signals such as frame headers / checksums / data lengths of the access user load in real time, matches them with the protocol feature library, and achieves a matching accuracy of ≥98%. It automatically generates conversion rules to establish a communication connection without manual configuration.

[0077] The first network transmission module, such as fiber optic / 5G, connects to the power grid master station and is responsible for long-distance, high-bandwidth data interaction between the terminal and the power grid master station. The second network transmission module, such as LoRa / Bluetooth, connects to the user-side local device. The anti-interference transmission unit communicates bidirectionally with the first and second network transmission modules to realize signal transmission between the terminal, the power grid master station, and the user side. The anti-interference transmission unit includes a dynamic channel evaluation subunit, a beam tracking antenna, and an electromagnetic shielding layer. The dynamic channel evaluation subunit monitors signal strength, bit error rate, and latency. The beam tracking antenna is designed for mobile scenarios. The electromagnetic shielding layer uses nanocrystalline alloy material to attenuate interference signals by ≥40dB.

[0078] The coupled modeling module includes a multidimensional data acquisition subunit and a coupled model training subunit. The multidimensional data acquisition subunit collects multidimensional data on load, environment, and users through sensors and inputs the collected multidimensional data into the coupled model training subunit. The coupled model training subunit is based on a Long Short-Term Memory (LSTM) network and trains the model with historical data on load, environment, user parameters, and control effects. It outputs the load adjustable potential prediction value and environmental interference coefficient to the multi-scenario adaptive control module. The adjustable potential prediction value is such as "the air conditioner can reduce power by 5% in the next 10 minutes without the user noticing", and the environmental interference coefficient is such as "the air conditioner's adjustment sensitivity decreases by 20% in a high-temperature environment".

[0079] The multi-dimensional data acquisition subunit acquires load parameters by using a flexible current sensor to collect real-time power, voltage curves, and start-stop loss data of the load; and reads inherent parameters of the equipment, such as the heat dissipation coefficient of the air conditioner and the charging and discharging efficiency curve of the charging pile, through the load digital twin interface.

[0080] The multi-dimensional data acquisition subunit collects environmental parameters by using distributed temperature and humidity sensors with a sampling frequency of 1Hz to collect ambient temperature and humidity; and monitors grid voltage fluctuations (sampling frequency of 50Hz) and new energy output (such as photovoltaic power) by using electromagnetic sensors.

[0081] The multidimensional data acquisition subunit acquires user parameter data such as temperature sensitivity thresholds and real-time demand data through the user interaction feedback module, as well as commands such as "temporarily pause control".

[0082] The coupled model training subunit includes,

[0083] The historical data acquisition module collects minute-level data over 12 months, totaling approximately 520,000 samples. Each sample contains 18-dimensional input features and 2-dimensional labels, which are manually labeled as "actual adjustable potential" and "actual environmental interference coefficient".

[0084] The data preprocessing module uses forward imputation to fill in missing values ​​for sporadic sensor disconnection data; it performs Min-Max normalization on the input features, mapping them to the [0,1] interval, and scales the labels according to the actual physical range, such as mapping adjustable potential to [0,20%] and interference coefficient to [0,50%]; it divides the data into training set (70%), validation set (20%), and test set (10%) according to time sequence to maintain temporal continuity.

[0085] The model training module uses Adam as the optimizer, with an initial learning rate of 0.001, which decays by 10% every 50 rounds.

[0086] Loss function: Mean Squared Error (MSE), formula:

[0087]

[0088] Where y1 represents the actual adjustable potential. y1 is the predicted value; y2 is the actual environmental interference coefficient. This is a predicted value;

[0089] The training rounds consist of 200 rounds, with an early stopping strategy: the training stops if the validation set loss does not decrease for 10 consecutive rounds.

[0090] The model tuning module uses a grid search method to optimize the number of neurons in the LSTM layer, with 64-32-16 being the optimal combination; a Dropout layer (probability 0.2) is added to suppress overfitting and ensure that the accuracy on the test set is ≥92% (the deviation between predicted and actual values ​​is ≤2%).

[0091] The multi-scenario adaptive control module includes a scenario identification subunit, a strategy generation subunit, and an execution monitoring subunit. The input end of the scenario identification subunit is connected to the coupled modeling module and the power grid master station, and the output end of the scenario identification subunit is connected to the strategy generation subunit. The strategy generation subunit sends instructions to the intelligent communication adaptation module and the execution monitoring subunit through the data bus. The intelligent communication adaptation module sends instructions to the load. The execution monitoring subunit feeds back deviation data to the strategy generation subunit.

[0092] The scenario recognition subunit presets five core scenarios: power grid peak / valley, industrial / civilian scenarios, extreme weather, special user needs, and equipment fault warning. It automatically identifies the current scenario by matching the power grid status, user commands, and environmental parameters in real time, with a recognition delay of <100ms. The power grid status is judged as peak if the frequency deviation is >0.2Hz, the user command is "night mode", and the environmental parameter is temperature >35℃. The execution monitoring subunit compares the control command with the actual load response in real time. For example, "command to reduce power by 3%, actual reduction is 2.8%". If the deviation is >5%, a secondary correction is triggered, such as "adjust by 0.2%", to achieve closed-loop control.

[0093] The strategy generation subunit is based on an improved particle swarm optimization algorithm, with the goals of grid stability with frequency deviation <0.1Hz, user satisfaction greater than or equal to 90%, and minimum energy consumption. It combines the adjustable potential output of the coupled model to generate dynamic control curves, which are not fixed commands.

[0094] Particle velocity update formula:

[0095]

[0096] Particle position update formula:

[0097]

[0098] In the formula,

[0099] The velocity of the i-th particle in the (t+1)-th generation in d-dimensional space;

[0100] The position of the i-th particle in the (t+1)-th generation in d-dimensional space;

[0101] w(t): Dynamic inertia weight, which is adaptively adjusted with the number of iterations. The initial value is 0.9, and it decreases linearly to 0.4 with each iteration, balancing the ability to explore the world and find local optimization.

[0102] c1c2: Learning factor, usually taken as 0.2, representing the weights of a particle learning from its own historical best and the group's best, respectively;

[0103] r1r2: Random numbers within the interval [0,1], increasing the randomness of the search;

[0104] pbest i,d The historical optimal position and the optimal self-regulation strategy of the i-th particle;

[0105] gbest d The global optimal position of the entire particle swarm, and the optimal control strategy for the swarm;

[0106] λ: Scenario weighting factor, dynamically determined based on the current scenario. For example, λ = 0.3 in a peak power grid scenario and λ = 0.6 in a user-sensitive scenario, thus enhancing the impact of the scenario on the control target.

[0107] S d Scene adaptation deviation value: In the d-th dimension, the deviation between the current particle position and the scene requirements. For example, if the user reports that the temperature is uncomfortable, S... d This is a positive correction value.

[0108] For example, in peak power grid scenarios: prioritize reducing load power, with a single-step adjustment range of ≤1% of rated power (to avoid impact);

[0109] User sleep scenario: Limit temperature adjustment range to ≤0.3℃ / time, and interval to ≥5 minutes.

[0110] The user interaction feedback module includes a feedback input subunit and a demand conversion subunit. The input end of the feedback input subunit connects to the user terminal, and the output end connects to the multi-scenario adaptive control module via the demand conversion subunit. The feedback input subunit supports voice commands (recognition accuracy ≥95%), touch screen input, and mobile APP interaction (latency <1s), and receives user feedback (such as "temperature too low" or "temporarily cancel control"). The demand conversion subunit quantifies the user feedback into control constraints (such as "user feedback 'cold' → increase temperature adjustment limit by 0.5℃") and sends it to the multi-scenario adaptive control module.

[0111] Terminal hardware carrier: It adopts a flexible polyimide substrate and integrates a microprocessor (such as ARM Cortex-M7), a multi-protocol communication chip (supporting RS-485 / 5G / LoRa / HPLC, etc.), a flexible sensor array (temperature / vibration / electromagnetic interference sensors) and a piezoelectric actuator. It has the characteristics of being flexible (curvature radius ≥5cm) and resistant to wide-band vibration (10-2000Hz), and is suitable for curved surface load equipment (such as pipeline heating load) and mobile scenarios (such as vehicle air conditioners).

[0112] Example 2

[0113] This invention discloses a method for remote control of flexible loads, comprising the following steps:

[0114] The system collects load communication signals, matches them with a protocol feature library, generates conversion rules, and establishes communication connections. Simultaneously, it collects initial data on load, environment, and users to establish a multi-dimensional parameter coupling basic model of load, environment, and users.

[0115] Collect real-time load, environment, and user multi-dimensional parameters, update the multi-dimensional parameter coupled basic model based on new data, and output prediction results;

[0116] Based on the predicted data, the current scenario is identified, and a dynamically optimized control strategy is generated.

[0117] Based on the dynamic optimization control strategy generated by the multi-scenario adaptive control module, a feedback and adjustment closed loop is constructed.

[0118] The working principle and process are as follows:

[0119] 1. Initialization phase (debugging and adaptation): After the terminal is connected to the load system (such as air conditioner, charging pile), the protocol adaptive conversion unit of the intelligent communication adaptation module automatically collects the load communication signals (such as frame structure, baud rate), matches them with the protocol feature library (such as matching "Modbus-RTU protocol"), generates conversion rules, and establishes a communication connection within 10 seconds (without the need for manual writing of data frames).

[0120] The load-environment-user coupling modeling module starts initial data acquisition (lasting 5 minutes) and generates a basic model (e.g., "The power of this air conditioner is 1.2kW at 26℃").

[0121] 2. During the real-time operation phase, taking "civilian air conditioning control during peak summer power grid periods" as an example, the working process is as follows:

[0122] 1) Data acquisition and model update:

[0123] The multi-dimensional data acquisition subunit collects the following data every second: real-time air conditioner power 1.5kW, ambient temperature 32℃, user-set temperature 26℃, and power grid frequency 50.1Hz (slightly high, judged as peak).

[0124] The coupled model training sub-unit updates the model based on new data, outputting "the air conditioner can reduce power by 8% (to 1.38kW) without the user noticing".

[0125] 2) Scene recognition and policy generation:

[0126] The scene recognition subunit matches "power grid peak + civilian scenario" to trigger peak control mode;

[0127] The multi-objective optimization strategy generates sub-units with the goal of "reducing power by 8%". Combined with user preferences (historical data shows that users can accept 26.3℃), the control curve is generated as follows: "Increase temperature by 0.1℃ in the first minute → power decrease by 0.05kW; increase temperature by 0.2℃ in the third minute → power decrease to 1.38kW".

[0128] 3) Command transmission and execution:

[0129] The intelligent communication adaptation module converts the control curve into protocol commands (such as Modbus commands) that the air conditioner can recognize, and sends them to the air conditioner through the anti-interference transmission unit (at this time, the evaluation channel is 5G, the signal strength is -70dBm, and beam tracking is enabled);

[0130] The air conditioner performs adjustments, and the piezoelectric actuator ensures that the adjustment response delay is less than 50ms.

[0131] 4) User feedback and dynamic adjustments:

[0132] The user provides voice feedback saying "It's a bit hot." After the feedback input subunit recognizes the feedback, the demand conversion subunit updates the constraint to "temperature adjustment 0.1℃".

[0133] The multi-scenario adaptive control module immediately corrected its strategy, instructing the air conditioner to cool down by 0.1℃ (power rebounded to 1.42kW), still meeting the peak control needs of the power grid.

[0134] 3. In case of communication interruption (e.g., loss of 5G signal), the intelligent communication adaptation module automatically switches to the LoRa channel (switching time < 500ms) and caches unsent commands. After the connection is restored, the commands are retransmitted through hash verification.

[0135] If the load parameters are abnormal (such as a sudden 20% increase in current), the coupled model training subunit will trigger an early warning, and the multi-scenario adaptive control module will immediately suspend control and report to the main station to avoid equipment damage.

[0136] This invention also provides a storage medium storing a computer program. When executed by a processor, the computer program implements some or all of the steps in various embodiments of the flexible load control remote control method provided by this invention. The storage medium may be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0137] Those skilled in the art will clearly understand that the techniques in the embodiments of the present invention can be implemented using software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solutions in the embodiments of the present invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in various embodiments or certain parts of the embodiments of the present invention.

[0138] The core innovation of this invention is manifested in the following technical aspects:

[0139] 1) Intelligent communication adaptation module: Through protocol fuzzy matching and anti-interference transmission, it can automatically adapt to more than 95% of non-standard protocols, and the communication packet loss rate is <0.1% in extreme environments (approximately 5% in existing technologies);

[0140] 2) Coupled modeling module: For the first time, user comfort is quantified into model parameters, which improves the control accuracy by 30% (and reduces the user complaint rate by 30% under the same adjustment amount);

[0141] 3) Multi-scenario adaptive control: Upgraded from "fixed instructions" to "dynamic curves", adapting to 10+ scenarios, and improving control flexibility by 50%;

[0142] 4) User interaction and feedback: Achieve "second-level feedback and instant adjustment", increasing user satisfaction from 75% of the existing technology to 92%.

[0143] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A flexible load control remote control terminal, characterized in that, It includes an intelligent communication adaptation module, a coupled modeling module, a multi-scenario adaptive control module, and a user interaction feedback module. These modules are interconnected through an internal data bus to achieve real-time data interaction and collaborative work. The intelligent communication adaptation module collects load communication signals to achieve multi-protocol adaptive conversion and reliable transmission in extreme environments; The coupled modeling module collects multi-dimensional parameters of load, environment, and users, and establishes a coupled model of multi-dimensional parameters of load, environment, and users. The multi-scenario adaptive control module receives the prediction result data from the coupled modeling module, identifies the current scenario, and generates a dynamically optimized control strategy. The user interaction feedback module is based on the dynamic optimization control strategy generated by the multi-scenario adaptive control module to construct a feedback and adjustment closed loop.

2. The flexible load control remote control terminal according to claim 1, characterized in that, The intelligent communication adaptation module has the following structure: the input end of the protocol adaptive conversion unit is connected to the user load system, and the output end of the protocol adaptive conversion unit is connected to the microprocessor through an internal data bus; the protocol adaptive conversion unit includes a protocol feature library and a fuzzy matching algorithm module; the protocol adaptive conversion unit extracts the feature values ​​of frame structure features and instruction interaction features by real-time parsing of the communication signals of the access user load, matches them with the protocol feature library, and automatically generates conversion rules to establish a communication connection; The first network transmission module is connected to the power grid master station, and the second network transmission module is connected to the user-side local equipment. The anti-interference transmission unit communicates bidirectionally with the first network transmission module and the second network transmission module to realize signal transmission between the terminal and the power grid master station and the user side. The anti-interference transmission unit includes a dynamic channel evaluation subunit, a beam tracking antenna, and an electromagnetic shielding layer.

3. The flexible load control remote control terminal according to claim 1, characterized in that, The coupled modeling module includes a multidimensional data acquisition subunit and a coupled model training subunit. The multidimensional data acquisition subunit collects multidimensional data on load, environment, and users through sensors and inputs the collected multidimensional data into the coupled model training subunit. The coupled model training subunit is based on a long short-term memory network (LSTM) and trains the model with historical data on load, environment, user parameters, and control effects. It outputs the predicted value of load adjustable potential and the environmental interference coefficient to the multi-scenario adaptive control module.

4. The flexible load control remote control terminal according to claim 3, characterized in that, The multi-dimensional data acquisition subunit acquires load parameters by collecting real-time power, voltage curves, and start-stop loss data of the load through a flexible current sensor; and reads the inherent parameters of the equipment through a load digital twin interface. The multi-dimensional data acquisition subunit collects environmental parameters by using distributed temperature and humidity sensors to collect ambient temperature and humidity; and by using electromagnetic sensors to monitor grid voltage fluctuations and new energy output data. The multidimensional data acquisition subunit acquires user parameters by obtaining user comfort preferences and real-time demand data through the user interaction feedback module.

5. The flexible load control remote control terminal according to claim 3, characterized in that, The coupled model training subunit includes, The historical data acquisition module collects minute-level data samples over a continuous period of 12 months. Each sample contains 18-dimensional input features and 2-dimensional labels. The data preprocessing module uses forward imputation to fill in missing values ​​for sporadic sensor disconnection data; it performs Min-Max normalization on the input features and scales the labels according to the actual physical range. The data is divided into training set, validation set and test set according to time sequence to maintain the continuity of time sequence; The model training module has an initial learning rate of 0.001, which decays by 10% every 50 rounds. Loss function formula: Where y1 represents the actual adjustable potential. y1 is the predicted value; y2 is the actual environmental interference coefficient. This is a predicted value; The training run consists of 200 rounds, and the process stops if the validation set loss does not decrease for 10 consecutive rounds. The model tuning module uses a grid search method to optimize the number of neurons in the LSTM layer and adds a Dropout layer to suppress overfitting.

6. The flexible load control remote control terminal according to claim 1, characterized in that, The multi-scenario adaptive control module includes a scenario identification subunit, a strategy generation subunit, and an execution monitoring subunit. The input end of the scenario identification subunit is connected to the coupled modeling module and the power grid master station, and the output end of the scenario identification subunit is connected to the strategy generation subunit. The strategy generation subunit sends instructions to the intelligent communication adaptation module and the execution monitoring subunit through the data bus. The intelligent communication adaptation module sends instructions to the load. The execution monitoring subunit feeds back deviation data to the strategy generation subunit.

7. A flexible load control remote control terminal according to claim 6, characterized in that, The scenario recognition subunit presets 5 core scenarios and automatically identifies the current scenario by matching the power grid status, user commands, and environmental parameters in real time. The execution monitoring subunit compares the control commands with the actual load response in real time. If the deviation is greater than the set value, it triggers a secondary correction to achieve closed-loop control.

8. A flexible load control remote control terminal according to claim 6, characterized in that, The strategy generation subunit is based on an improved particle swarm optimization algorithm, with the goals of grid stability, user satisfaction, and minimum energy consumption. It combines the adjustable potential output by the coupled model to generate a dynamic control curve. Particle velocity update formula: Particle position update formula: In the formula, The velocity of the i-th particle in the (t+1)-th generation in d-dimensional space; The position of the i-th particle in the (t+1)-th generation in d-dimensional space; w(t): Dynamic inertia weight, which is adaptively adjusted with the number of iterations; c1c2: Learning factors, representing the weights of a particle learning from its own historical best and the group's best, respectively; r1r2: Random numbers within the interval [0,1], increasing the randomness of the search; pbest i,d The historical best position of the i-th particle; gbest d The global optimal position of the entire particle swarm; λ: Scene weighting factor, dynamically determined based on the current scene to enhance the scene's impact on the control target; S d Scene adaptation deviation value: the deviation between the current particle position and the scene requirements in the d-th dimension space.

9. A flexible load control remote control terminal according to claim 1, characterized in that, The user interaction feedback module includes a feedback input subunit and a demand transformation subunit. The input end of the feedback input subunit is connected to the user terminal, and the output end of the feedback input subunit is connected to the multi-scenario adaptive control module via the demand transformation subunit.

10. A control method for a flexible load remote control terminal according to any one of claims 1-9, characterized in that, Includes the following steps: The system collects load communication signals, matches them with a protocol feature library, generates conversion rules, and establishes communication connections. Simultaneously, it collects initial data on load, environment, and users to establish a multi-dimensional parameter coupling basic model of load, environment, and users. Collect real-time load, environment, and user multi-dimensional parameters, update the multi-dimensional parameter coupled basic model based on new data, and output prediction results; Based on the predicted data, the current scenario is identified, and a dynamically optimized control strategy is generated. Based on the dynamic optimization control strategy generated by the multi-scenario adaptive control module, a feedback and adjustment closed loop is constructed.

Images