A multi-parameter monitoring and regulation method of an AI intelligent flowerpot

Abstract

The application relates to the technical field of automatic parameter control, in particular to a multi-parameter monitoring and regulation method of an AI intelligent flowerpot, which comprises the following steps: after the intelligent flowerpot is started, hardware self-checking is completed, plant type information input by a user or automatic plant species identification is acquired through a multi-modal interface, preset suitable growth parameter thresholds corresponding to tropical plants, subtropical plants, succulents / desert plants and temperate plants are loaded, user-defined parameter thresholds are supported, and a plant growth parameter benchmark library containing temperature, water level and light threshold values is established. Through the construction of the exclusive plant parameter benchmark library and the optimization of a multi-source data preprocessing process, the accuracy of monitoring data is improved; the application solves the problems of insufficient accuracy, poor adaptability, lack of emotional interaction and self-optimization capability of traditional intelligent flowerpots, takes into account plant accurate maintenance and user experience, is especially suitable for being used by people living alone, improves the long-term use value of the equipment, and has a wide application scenario.

Description

Technical Field

[0001] This invention relates to the field of automated parameter control technology, specifically to a multi-parameter monitoring and control method for an AI-powered smart flowerpot. Background Technology

[0002] AI smart flower pots are intelligent gardening devices that integrate sensors, IoT, and AI technologies. They can collect plant growth environment parameters in real time and automatically perform maintenance operations. They are widely used in home gardening, companionship for people living alone, and the construction of smart home ecosystems. They can effectively reduce the threshold for plant maintenance and adapt to the needs of modern fast-paced life and special groups.

[0003] However, most existing smart flower pots only have single parameter monitoring and simple control functions, lacking a dedicated parameter configuration system for different plant types. The data preprocessing accuracy is insufficient, and they are easily affected by environmental interference, leading to decision-making biases. At the same time, traditional devices have fixed control strategies, lack dynamic weight analysis capabilities, cannot adapt to plant growth cycles and regional climate differences, and lack emotional interaction design, making it difficult to meet the companionship needs of people living alone.

[0004] In addition, most products lack a self-iterative optimization mechanism, and their adaptability decreases after long-term use, making it impossible to form a closed-loop maintenance ecosystem and difficult to balance the accuracy of plant maintenance with user experience.

[0005] In summary, a multi-parameter monitoring and control method for AI-powered smart flowerpots needs to be proposed to solve the above problems. Summary of the Invention

[0006] The purpose of this invention is to provide a multi-parameter monitoring and control method for AI smart flowerpots to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: This invention proposes a multi-parameter monitoring and control method for AI smart flowerpots, comprising the following steps: S1. Device initialization and parameter configuration: After the smart flowerpot is powered on, it completes a hardware self-test, obtains the plant type information input by the user through a multimodal interface or automatically identifies the plant species, loads the preset suitable growth parameter thresholds corresponding to tropical, subtropical, succulent / desert, and temperate plants, supports user-defined parameter thresholds, and establishes a plant growth parameter benchmark library containing temperature, water level, and light thresholds. S2. Multi-parameter acquisition and data preprocessing: Water level, 0-1000 lux light intensity, temperature, tilt angle, human proximity and touch operation data are collected in real time through non-contact liquid level sensor and light sensor equipment. The moving average filtering algorithm is used to remove abnormal data and normalize the multi-source sensor data to the [0,1] interval to generate a structured monitoring dataset. S3. AI intelligent decision generation, based on the parameter benchmark library of step S1 and the structured monitoring dataset of step S2, inputs the AI ​​large language model after being fine-tuned by the horticultural maintenance corpus, analyzes the influence weight of each monitoring parameter on plant growth through the plant growth influencing factor weight algorithm, and generates control strategies and interactive response instructions by combining festival, weather and spatiotemporal scene information. S4. Control Execution and Interactive Feedback: Based on the control strategy in step S3, the automatic watering, smart home linkage, and power failure protection execution modules are activated. Based on the interactive response command, the LCD screen displays 100+ dynamic expressions and voice broadcasts to achieve multi-dimensional feedback. S5. Data synchronization and model iteration: Monitoring data and control records are synchronized to the APP and cloud via Tuya IoT Cloud API. Based on user feedback and growth cycle data, the AI ​​model decision logic is optimized through plant maintenance strategy reinforcement learning algorithm. The device program and interactive resource library are updated regularly through firmware upgrades.

[0008] Preferably, the implementation steps of step S1 include: S1.1. Hardware self-test: After the device is powered on, it automatically detects the connection validity of the liquid level sensor, light sensor, temperature sensor, tilt sensor, human infrared sensor, and touch sensor, and displays the detection progress and network configuration status. S1.2. First-time use guide: Follow the steps in sequence: welcome ceremony, button instructions, APP download instructions, Bluetooth connection instructions, device addition instructions, WIFI connection instructions, watering instructions, and plant placement instructions to complete the initialization configuration. Long press or multiple taps of the initialization button can be used to restart the initialization process. S1.3. Parameter configuration, loading preset thresholds: tropical plants: temperature 18-30℃, water level threshold corresponding to soil moisture 60%-80%, light 300-600 lux; subtropical plants: temperature 15-28℃, water level threshold corresponding to soil moisture 50%-70%, light 100-300 lux; succulents / desert plants: temperature 15-32℃, water level threshold corresponding to soil moisture 30%-50%, light 600-800 lux; temperate plants: temperature 10-25℃, water level threshold corresponding to soil moisture 40%-60%, light 300-600 lux. Users can customize thresholds through the APP and send them to device storage.

[0009] Preferably, the implementation steps of step S2 include: S2.1. Water level data acquisition: A non-contact liquid level sensor with a sensing thickness of ≤5mm is used to collect water tank water level and soil moisture data. When water shortage is detected and the flag is 0, the water shortage timer is activated. If the timer exceeds 8 hours, the reminder level is upgraded. S2.2. Environmental parameter acquisition: The light sensor classifies light levels into five levels: 0-100 lux (extremely low), 100-300 lux (low), 300-600 lux (medium), 600-800 lux (high), and 800-1000 lux (extremely high); the temperature sensor collects data in real time, with a blue screen theme associated with temperatures below 15℃, a green theme associated with temperatures between 15-28℃, and a gradient red theme associated with temperatures above 28℃; the tilt sensor monitors the tilt angle, and the human infrared sensor captures human proximity signals within 3-5m; S2.3. Data preprocessing: The raw monitoring data of soil moisture, temperature, and light intensity are processed using a moving average filtering algorithm, as shown in equation (1): (1); In the formula, The data is filtered soil moisture, in %RH. When the raw soil moisture content is continuously collected at 45%, 47%, 46%, 48%, and 47%, a sliding window of N=5 is used. That is, (45+47+46+48+47) / 5=46.6%RH; The sliding window size is 5-10; The original monitoring data of soil moisture, temperature and light intensity collected at different times are processed by this algorithm. Outliers exceeding 3 times the standard deviation are removed, and the data of soil moisture 0-100%RH, temperature -10-50℃ and light intensity 0-1000lux are normalized to the range of [0,1] to generate a structured dataset.

[0010] Preferably, the implementation steps of step S3 include: S3.1. Model selection and fine-tuning: Qwen-7B / Doubao-13B / GLM-4 was selected as the base model, and fine-tuning was performed based on a horticultural maintenance corpus containing plant maintenance knowledge and sensor data interpretation to ensure that the plant status recognition accuracy is ≥95%. S3.2. Calculation of plant growth influencing factors weights: The weight values ​​of each monitoring parameter are calculated using the plant growth influencing factor weighting algorithm, as shown in equation (2): (2); In the formula, This is the weight value of the i-th monitoring parameter, specifically the weight value of the water level parameter; The percentage of monitoring frequency for a parameter within one hour; when the water level sensor collects data 30 times within one hour, for a total of 90 monitoring times. =30 / 90≈0.33; The parameters are environmental impact factors: water level 0.8, temperature 0.7, light intensity 0.6, tilt angle 0.4, and human proximity 0.3. The value range is [0.6, 0.8]; The value range is [0.2, 0.4]; The semantic correlation between the parameters and plant growth is 0.9 for water level, 0.85 for temperature, 0.8 for light intensity, 0.5 for tilt angle, and 0.4 for human proximity. S3.3. Strategy generation: Combining the weight analysis results with spatiotemporal information, a watering command is generated when the water level weight is ≥0.7, a temperature adjustment command is generated when the temperature weight is ≥0.65, and a light adaptation prompt is generated when the light weight is ≥0.6. At the same time, corresponding facial expressions and voice commands are matched.

[0011] Preferably, the implementation steps of step S4 include: S4.1. Control and execution: The automatic watering module calculates the watering duration using a precise watering control algorithm based on the soil moisture data collected in step S2 and the suitable water level threshold in step S1, as shown in equation (3): (3); In the formula, ; ; ; This is the soil moisture deviation value. Succulents thrive in soil with a moisture content of 40%. The current filtered moisture content is 25%. ; When the temperature is abnormal, the air conditioner and humidifier will be activated. When the tilt angle exceeds 15°, an alarm will be triggered. If the tilt is completely tilted, the power will be automatically cut off. S4.2. Expression interaction: The expression library contains dynamic expressions for being thirsty, happy, and welcoming. In a natural state, the "drinking water" expression switches every 2 minutes for 5 seconds, and the "lazy" expression switches every 5 minutes. The corresponding scene expression is switched based on the instructions in step S3. S4.3. Voice interaction: In alarm scenarios (water shortage, abnormal temperature), human voice prompts are played; in interactive scenarios (touch, human proximity), simulated animal sounds or white noise are played. 2-3 sets of voice prompts are configured for random playback at the same type of touch point. A 10-minute cooling-off mechanism is activated after triggering.

[0012] Preferably, the implementation steps of step S5 include: S5.1. Data synchronization: Monitoring data is uploaded to the APP in real time through Tuya IoT Cloud API to generate daily, weekly and monthly statistical reports, and supports querying historical water level, temperature, light data and alarm records; S5.2. Firmware upgrade: The APP detects the difference between the current firmware version of the device and the latest version on the cloud, downloads the .bin format firmware package in segments and sends them to the device. After verification, the device restarts and enters bootloader mode, erases the old firmware and writes the new firmware, and then restarts. S5.3. Model iteration: The AI ​​model decision logic is optimized using a plant maintenance strategy reinforcement learning algorithm, as shown in equation (4): (4); In the formula, The value of the current plant care status is specifically the value of the growth status of succulents under conditions of 40% humidity, 25℃ temperature, and 700 lux light. The learning rate; As a reward for user feedback, when a user rates the watering effect 8 out of 10. ; Discount factor; The next maintenance status value is specifically the growth status value under the conditions of 40% humidity, 25℃ temperature, and 700 lux light after watering; after generating 10 batches of control strategies, the parameter weight calculation model is calibrated and the expression and voice triggering logic is optimized.

[0013] Preferably, the alarm threshold in step S1.3 is set as follows: when the temperature of tropical plants is below 12℃, subtropical plants below 8℃, succulents / desert plants below 10℃, and temperate plants below 5℃, an APP alarm is triggered and an audio sound is played. The alarm function can be turned off through the APP.

[0014] Preferably, in step S2.2, the human infrared sensor is deployed in a transparent hole below the screen. A thermal compensation algorithm is used to eliminate the monitoring error caused by the close proximity of summer air temperature and human body temperature. If no human body is detected within 30 minutes, the screen brightness is automatically reduced or the screen is turned off to reduce power consumption.

[0015] Preferably, the environmental impact factors in step S3.2 Water levels in rainy southern regions are dynamically adjusted according to regional climate characteristics. Adjusted to 0.7, and to 0.9 for arid regions in the north; semantic relevance. Assigning values ​​based on plant physiology principles, succulent plant light levels Adjusted to 0.85, humidity for tropical plants Adjusted to 0.95.

[0016] Preferably, in step S4.1, the pipe of the automatic watering module is connected to the bottom of the flowerpot, and the watering time is dynamically adjusted according to the soil moisture deviation value. For every 10% increase in the deviation value, the watering time increases by 3 seconds. When the moisture deviation of the succulent plant is 15%, the watering time is 4.5 seconds, and the longest watering time for a single watering does not exceed 30 seconds.

[0017] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention improves the accuracy of monitoring data by constructing a dedicated plant parameter benchmark library and optimizing the multi-source data preprocessing process, providing a reliable basis for subsequent decision-making; it achieves dynamic multi-parameter analysis and targeted regulation strategy generation by combining plant growth influencing factor weighting algorithms with AI models, adapting to different plant growth characteristics and environmental differences; it ensures a stable plant growth environment and meets the companionship needs of people living alone through smart home linkage and multimodal emotional interaction design; and it continuously improves device performance through data synchronization and reinforcement learning iterative optimization, forming a closed-loop maintenance ecosystem. In summary, this invention solves the problems of insufficient accuracy, poor adaptability, lack of emotional interaction, and self-optimization capabilities of traditional smart flowerpots, balancing precise plant maintenance with user experience, especially suitable for people living alone, enhancing the long-term use value of the device, and has a wide range of application scenarios. Attached Figure Description

[0018] Figure 1 This is a flowchart of a multi-parameter monitoring and control method for an AI smart flowerpot according to the present invention; Detailed Implementation

[0019] 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.

[0020] For examples, please refer to Figure 1 In practical applications, this invention proposes a multi-parameter monitoring and control method for AI smart flowerpots, specifically including the following steps: S1. Device initialization and parameter configuration: After the smart flowerpot is powered on, it completes a hardware self-test, obtains the plant type information input by the user through a multimodal interface or automatically identifies the plant species, loads the preset suitable growth parameter thresholds corresponding to tropical, subtropical, succulent / desert, and temperate plants, supports user-defined parameter thresholds, and establishes a plant growth parameter benchmark library containing temperature, water level, and light thresholds. It should also be noted that the implementation steps of step S1 include: S1.1. Hardware self-test: After the device is powered on, it automatically detects the connection validity of the liquid level sensor, light sensor, temperature sensor, tilt sensor, human infrared sensor, and touch sensor, and displays the detection progress and network configuration status. S1.2. First-time use guide: Follow the steps in sequence: welcome ceremony, button instructions, APP download instructions, Bluetooth connection instructions, device addition instructions, WIFI connection instructions, watering instructions, and plant placement instructions to complete the initialization configuration. Long press or multiple taps of the initialization button can be used to restart the initialization process. S1.3. Parameter configuration, loading preset thresholds: Tropical plants: temperature 18-30℃, water level threshold corresponding to soil moisture 60%-80%, light 300-600 lux; Subtropical plants: temperature 15-28℃, water level threshold corresponding to soil moisture 50%-70%, light 100-300 lux; Succulents / Desert plants: temperature 15-32℃, water level threshold corresponding to soil moisture 30%-50%, light 600-800 lux; Temperate plants: temperature 10-25℃, water level threshold corresponding to soil moisture 40%-60%, light 300-600 lux. Users can customize thresholds via the APP and send them to device storage. It should also be noted that the alarm threshold in step S1.3 is set as follows: when the temperature of tropical plants is below 12℃, subtropical plants below 8℃, succulents / desert plants below 10℃, and temperate plants below 5℃, the APP alarm will be triggered and a sound will be played once. The alarm function can be turned off through the APP. In practical applications, the equipment stores preset parameter thresholds to form a baseline library. Specific values ​​are: suitable temperature 15-32℃, soil moisture 30%-55%, light intensity 600-800 lux, and a low-temperature alarm threshold of 10℃. Parameters are stored in key-value pair format, such as: Plant type: cactus; Lower limit of temperature: 15℃; Upper temperature limit: 32℃; Lower limit of humidity: 30%; Humidity upper limit: 55%; Lower limit of illumination: 600 lux; Maximum illumination: 800 lux; Low temperature alarm threshold: 10℃; Step S1 completes the binding of the device with the network and APP, establishes a dedicated plant growth parameter benchmark library, clarifies the suitable growth threshold of cacti, provides data basis for subsequent monitoring and regulation, and at the same time reduces the operation difficulty for elderly people living alone through visual guidance, ensuring that the initialization process is simple and easy to understand. S2. Multi-parameter acquisition and data preprocessing: Water level, 0-1000 lux light intensity, temperature, tilt angle, human proximity and touch operation data are collected in real time through non-contact liquid level sensor and light sensor equipment. The moving average filtering algorithm is used to remove abnormal data and normalize the multi-source sensor data to the [0,1] interval to generate a structured monitoring dataset. It should also be noted that the implementation steps of step S2 include: S2.1. Water level data acquisition: A non-contact liquid level sensor with a sensing thickness of ≤5mm is used to collect water tank water level and soil moisture data. When water shortage is detected and the flag is 0, the water shortage timer is activated. If the timer exceeds 8 hours, the reminder level is upgraded. S2.2. Environmental parameter acquisition: The light sensor classifies light levels into five levels: 0-100 lux (extremely low), 100-300 lux (low), 300-600 lux (medium), 600-800 lux (high), and 800-1000 lux (extremely high); the temperature sensor collects data in real time, with a blue screen theme associated with temperatures below 15℃, a green theme associated with temperatures between 15-28℃, and a gradient red theme associated with temperatures above 28℃; the tilt sensor monitors the tilt angle, and the human infrared sensor captures human proximity signals within 3-5m; It should also be noted that in step S2.2, the human infrared sensor is deployed in the transparent hole below the screen. A thermal compensation algorithm is used to eliminate the monitoring error caused by the close proximity of summer temperature and human body temperature. If no human body is detected within 30 minutes, the screen brightness is automatically reduced or the screen is turned off to reduce power consumption. S2.3. Data preprocessing: The raw monitoring data of soil moisture, temperature, and light intensity are processed using a moving average filtering algorithm, as shown in equation (1): (1); In the formula, The data is filtered soil moisture, in %RH. When the raw soil moisture content is continuously collected at 45%, 47%, 46%, 48%, and 47%, a sliding window of N=5 is used. That is, (45+47+46+48+47) / 5=46.6%RH; The sliding window size is 5-10; The original monitoring data of soil moisture, temperature and light intensity collected at different times are processed by this algorithm. Outliers exceeding 3 times the standard deviation are removed, and the data of soil moisture 0-100%RH, temperature -10-50℃ and light intensity 0-1000lux are normalized to the range of [0,1] to generate a structured dataset. Through step S2, multiple sensors comprehensively collect key parameters such as water level, temperature, and light intensity. After filtering, outlier removal, and normalization, accurate and uniform structured data is obtained, eliminating the impact of environmental interference and data format differences on subsequent decision-making and ensuring the reliability of input data. S3. AI intelligent decision generation, based on the parameter benchmark library of step S1 and the structured monitoring dataset of step S2, inputs the AI ​​large language model after being fine-tuned by the horticultural maintenance corpus, analyzes the influence weight of each monitoring parameter on plant growth through the plant growth influencing factor weight algorithm, and generates control strategies and interactive response instructions by combining festival, weather and spatiotemporal scene information. It should also be noted that the implementation steps of step S3 include: S3.1. Model selection and fine-tuning: Qwen-7B / Doubao-13B / GLM-4 was selected as the base model, and fine-tuning was performed based on a horticultural maintenance corpus containing plant maintenance knowledge and sensor data interpretation to ensure that the plant status recognition accuracy is ≥95%. S3.2. Calculation of plant growth influencing factors weights: The weight values ​​of each monitoring parameter are calculated using the plant growth influencing factor weighting algorithm, as shown in equation (2): (2); In the formula, This is the weight value of the i-th monitoring parameter, specifically the weight value of the water level parameter; The percentage of monitoring frequency for a parameter within one hour; when the water level sensor collects data 30 times within one hour, for a total of 90 monitoring times. =30 / 90≈0.33; The parameters are environmental impact factors: water level 0.8, temperature 0.7, light intensity 0.6, tilt angle 0.4, and human proximity 0.3. The value range is [0.6, 0.8]; The value range is [0.2, 0.4]; The semantic correlation between the parameters and plant growth is 0.9 for water level, 0.85 for temperature, 0.8 for light intensity, 0.5 for tilt angle, and 0.4 for human proximity. It should also be noted that the environmental impact factors in step S3.2 Water levels in rainy southern regions are dynamically adjusted according to regional climate characteristics. Adjusted to 0.7, and to 0.9 for arid regions in the north; semantic relevance. Assigning values ​​based on plant physiology principles, succulent plant light levels Adjusted to 0.85, humidity for tropical plants Adjusted to 0.95; S3.3. Strategy generation: Combining the weight analysis results with spatiotemporal information, a watering command is generated when the water level weight is ≥0.7, a temperature adjustment command is generated when the temperature weight is ≥0.65, and a light adaptation prompt is generated when the light weight is ≥0.6. At the same time, corresponding facial expressions and voice commands are matched. By combining the S3AI model with a weighting algorithm, multi-parameter intelligent analysis is achieved, accurately judging the plant growth status and user interaction needs, generating targeted strategies, solving the problem of traditional flower pots lacking intelligent decision-making capabilities, and providing precise instruction support for subsequent control and interaction. S4. Control Execution and Interactive Feedback: Based on the control strategy in step S3, the automatic watering, smart home linkage, and power failure protection execution modules are activated. Based on the interactive response command, the LCD screen displays 100+ dynamic expressions and voice broadcasts to achieve multi-dimensional feedback. It should also be noted that the implementation steps of step S4 include: S4.1. Control and execution: The automatic watering module calculates the watering duration using a precise watering control algorithm based on the soil moisture data collected in step S2 and the suitable water level threshold in step S1, as shown in equation (3): (3); In the formula, ; ; ; This is the soil moisture deviation value. Succulents thrive in soil with a moisture content of 40%. The current filtered moisture content is 25%. ; When the temperature is abnormal, the air conditioner and humidifier will be activated. When the tilt angle exceeds 15°, an alarm will be triggered. If the tilt is completely tilted, the power will be automatically cut off. It should also be noted that in step S4.1, the pipe of the automatic watering module is connected to the bottom of the flower pot. The watering time is dynamically adjusted according to the soil moisture deviation value. For every 10% increase in the deviation value, the watering time increases by 3 seconds. When the moisture deviation of the succulent plant is 15%, the watering time is 4.5 seconds. The maximum watering time for a single watering session shall not exceed 30 seconds. S4.2. Expression interaction: The expression library contains dynamic expressions for being thirsty, happy, and welcoming. In a natural state, the "drinking water" expression switches every 2 minutes for 5 seconds, and the "lazy" expression switches every 5 minutes. The corresponding scene expression is switched based on the instructions in step S3. S4.3. Voice interaction: In alarm scenarios (water shortage, abnormal temperature), human voice prompts are played; in interactive scenarios (touch, human proximity), simulated animal sounds or white noise are played. 2-3 sets of voice prompts are configured for random playback at the same type of touch point. A 10-minute cooling-off mechanism is activated after triggering. Step S4 enables precise control such as automatic watering and smart home integration, ensuring a stable growing environment for cacti; through multimodal interaction such as facial expressions, voice, and sound effects, it provides emotional companionship for elderly people living alone, alleviating their loneliness, while reducing the manual operation costs for the elderly, achieving the dual goals of "precise care + emotional companionship"; S5. Data synchronization and model iteration: Monitoring data and control records are synchronized to the APP and cloud via Tuya IoT Cloud API. Based on user feedback and growth cycle data, the AI ​​model decision logic is optimized through plant maintenance strategy reinforcement learning algorithm. The device program and interactive resource library are updated regularly through firmware upgrades. It should also be noted that the implementation steps of step S5 include: S5.1. Data synchronization: Monitoring data is uploaded to the APP in real time through Tuya IoT Cloud API to generate daily, weekly and monthly statistical reports, and supports querying historical water level, temperature, light data and alarm records; S5.2. Firmware upgrade: The APP detects the difference between the current firmware version of the device and the latest version on the cloud, downloads the .bin format firmware package in segments and sends them to the device. After verification, the device restarts and enters bootloader mode, erases the old firmware and writes the new firmware, and then restarts. S5.3. Model iteration: The AI ​​model decision logic is optimized using a plant maintenance strategy reinforcement learning algorithm, as shown in equation (4): (4); In the formula, The value of the current plant care status is specifically the value of the growth status of succulents under conditions of 40% humidity, 25℃ temperature, and 700 lux light. The learning rate; As a reward for user feedback, when a user rates the watering effect 8 out of 10. ; Discount factor; The next maintenance status value is specifically the growth status value under the conditions of 40% humidity, 25℃ temperature, and 700 lux light after watering; after generating 10 batches of control strategies, the parameter weight calculation model is calibrated and the expression and voice triggering logic is optimized. Step S5 enables real-time synchronization and traceability of maintenance data, making it convenient for seniors to check the plant growth status. Firmware upgrades add new functions, and model iterations and optimizations improve decision-making accuracy and interactive adaptability, ensuring long-term stable operation of the equipment and continuous adaptation to plant growth needs and user habits.

[0021] In summary, this invention forms a closed-loop collaboration through steps S1 to S5: S1 establishes a dedicated parameter benchmark library to provide a data foundation for subsequent processes; S2 collects and preprocesses multi-source data to ensure the reliability of input data; S3 intelligently generates decision instructions based on data and the benchmark library, realizing data-driven decision transformation; S4 executes control and interaction, translating decisions into actual actions; S5 synchronizes data and iteratively optimizes, continuously improving equipment performance. The entire process, through a closed-loop mechanism of parameter configuration, data collection, intelligent decision-making, execution feedback, and iterative optimization, achieves precise control of the plant growth environment and emotional interaction with users, while also possessing self-optimization capabilities.

[0022] 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 method for multi-parameter monitoring and control of an AI-powered smart flowerpot, characterized in that, Includes the following steps: S1. After powering on, the smart flowerpot completes a hardware self-test, obtains the plant type information input by the user through a multimodal interface or automatically identifies the plant species, loads the preset suitable growth parameter thresholds corresponding to tropical, subtropical, succulent / desert, and temperate plants, supports user-defined parameter thresholds, and establishes a plant growth parameter benchmark library containing temperature, water level, and light thresholds. S2. Real-time data collection of water level, 0-1000 lux light intensity, temperature, tilt angle, human proximity and touch operation data is carried out using non-contact liquid level sensor and light sensor equipment. Abnormal data is removed by using a moving average filtering algorithm, and the multi-source sensor data is normalized to the [0,1] interval to generate a structured monitoring dataset. S3. Based on the parameter benchmark library of step S1 and the structured monitoring dataset of step S2, input the AI ​​large language model after fine-tuning the horticultural maintenance corpus, analyze the influence weight of each monitoring parameter on plant growth through the plant growth influencing factor weight algorithm, and generate control strategies and interactive response instructions by combining festival and weather spatiotemporal scene information. S4. Based on the control strategy in step S3, the automatic watering, smart home linkage, and power failure protection execution modules are activated. Based on the interactive response command, the LCD screen is triggered to display 100+ dynamic expressions and voice broadcast, realizing multi-dimensional feedback. S5. The monitoring data and control records are synchronized to the APP and cloud via Tuya IoT Cloud API. Based on user feedback and growth cycle data, the AI ​​model decision logic is optimized through plant maintenance strategy reinforcement learning algorithm. The device program and interactive resource library are updated regularly through firmware upgrades.

2. The multi-parameter monitoring and control method for an AI smart flowerpot according to claim 1, characterized in that, The implementation steps of step S1 include: S1.

1. Hardware self-test: After the device is powered on, it automatically detects the connection validity of the liquid level sensor, light sensor, temperature sensor, tilt sensor, human infrared sensor, and touch sensor, and displays the detection progress and network configuration status. S1.

2. First-time use guide: Follow the steps in sequence: welcome ceremony, button instructions, APP download instructions, Bluetooth connection instructions, device addition instructions, WIFI connection instructions, watering instructions, and plant placement instructions to complete the initialization configuration. Long press or multiple taps of the initialization button can be used to restart the initialization process. S1.

3. Parameter configuration, loading preset thresholds: tropical plants: temperature 18-30℃, water level threshold corresponding to soil moisture 60%-80%, light 300-600 lux; subtropical plants: temperature 15-28℃, water level threshold corresponding to soil moisture 50%-70%, light 100-300 lux; succulents / desert plants: temperature 15-32℃, water level threshold corresponding to soil moisture 30%-50%, light 600-800 lux; temperate plants: temperature 10-25℃, water level threshold corresponding to soil moisture 40%-60%, light 300-600 lux. Users can customize thresholds through the APP and send them to device storage.

3. The multi-parameter monitoring and control method for an AI smart flowerpot according to claim 1, characterized in that, The implementation steps of step S2 include: S2.

1. Water level data acquisition: A non-contact liquid level sensor with a sensing thickness of ≤5mm is used to collect water tank water level and soil moisture data. When water shortage is detected and the flag is 0, the water shortage timer is activated. If the timer exceeds 8 hours, the reminder level is upgraded. S2.

2. Environmental parameter acquisition: The light sensor classifies light levels into five levels: 0-100 lux (extremely low), 100-300 lux (low), 300-600 lux (medium), 600-800 lux (high), and 800-1000 lux (extremely high); the temperature sensor collects data in real time, with a blue screen theme associated with temperatures below 15℃, a green theme associated with temperatures between 15-28℃, and a gradient red theme associated with temperatures above 28℃; the tilt sensor monitors the tilt angle, and the human infrared sensor captures human proximity signals within 3-5m; S2.

3. Data preprocessing: The raw monitoring data of soil moisture, temperature, and light intensity are processed using a moving average filtering algorithm, as shown in equation (1): （1）； In the formula, The data is filtered soil moisture, in %RH. When the raw soil moisture content is continuously collected at 45%, 47%, 46%, 48%, and 47%, a sliding window of N=5 is used. That is, (45+47+46+48+47) / 5=46.6%RH; The sliding window size is 5-10; The original monitoring data of soil moisture, temperature and light intensity collected at different times are processed by this algorithm. Outliers exceeding 3 times the standard deviation are removed, and the data of soil moisture 0-100%RH, temperature -10-50℃ and light intensity 0-1000lux are normalized to the range [0,1] to generate a structured dataset.

4. The multi-parameter monitoring and control method for an AI smart flowerpot according to claim 1, characterized in that, The implementation steps of step S3 include: S3.

1. Model selection and fine-tuning: Qwen-7B / Doubao-13B / GLM-4 was selected as the base model, and fine-tuning was performed based on a horticultural maintenance corpus containing plant maintenance knowledge and sensor data interpretation. S3.

2. Calculation of plant growth influencing factors weights: The weight values ​​of each monitoring parameter are calculated using the plant growth influencing factor weighting algorithm, as shown in equation (2): （2）； In the formula, This is the weight value of the i-th monitoring parameter, specifically the weight value of the water level parameter; The percentage of monitoring frequency for a parameter within one hour; when the water level sensor collects data 30 times within one hour, for a total of 90 monitoring times. =30 / 90≈0.33; The parameters are environmental impact factors: water level 0.8, temperature 0.7, light intensity 0.6, tilt angle 0.4, and human proximity 0.

3. The value range is [0.6, 0.8]; The value range is [0.2, 0.4]; The semantic correlation between the parameters and plant growth is 0.9 for water level, 0.85 for temperature, 0.8 for light intensity, 0.5 for tilt angle, and 0.4 for human proximity. S3.

3. Strategy generation: Combining the weight analysis results with spatiotemporal information, a watering command is generated when the water level weight is ≥0.7, a temperature adjustment command is generated when the temperature weight is ≥0.65, and a light adaptation prompt is generated when the light weight is ≥0.

6. At the same time, corresponding facial expressions and voice commands are matched.

5. The multi-parameter monitoring and control method for an AI smart flowerpot according to claim 1, characterized in that, The implementation steps of step S4 include: S4.

1. Control and execution: The automatic watering module calculates the watering duration using a precise watering control algorithm based on the soil moisture data collected in step S2 and the suitable water level threshold in step S1, as shown in equation (3): （3）； In the formula, ; ； ； This is the soil moisture deviation value. Succulents thrive in soil with a moisture content of 40%. The current filtered moisture content is 25%. ; When the temperature is abnormal, the air conditioner and humidifier will be activated. When the tilt angle exceeds 15°, an alarm will be triggered. If the tilt is completely tilted, the power will be automatically cut off. S4.

2. Expression interaction: The expression library contains dynamic expressions for being thirsty, happy, and welcoming. In a natural state, the "drinking water" expression switches every 2 minutes for 5 seconds, and the "lazy" expression switches every 5 minutes. The corresponding scene expression is switched based on the instructions in step S3. S4.

3. Voice interaction: In alarm scenarios, human voice prompts are played; in interactive scenarios, simulated animal sounds or white noise are played. 2-3 sets of voice prompts are configured for random playback at the same type of touchpoint. A 10-minute cooldown mechanism is activated after triggering.

6. The multi-parameter monitoring and control method for an AI smart flowerpot according to claim 1, characterized in that, The implementation steps of step S5 include: S5.

1. Data synchronization: Monitoring data is uploaded to the APP in real time through Tuya IoT Cloud API, generating daily, weekly and monthly statistical reports, and supporting the query of historical water level, temperature, light data and alarm records; S5.

2. Firmware upgrade: The APP detects the difference between the current firmware version of the device and the latest version on the cloud, downloads the .bin format firmware package in segments and sends them to the device. After verification, the device restarts and enters bootloader mode, erases the old firmware and writes the new firmware, and then restarts. S5.

3. Model iteration: The AI ​​model decision logic is optimized using a plant maintenance strategy reinforcement learning algorithm, as shown in equation (4): （4）； In the formula, The value of the current plant care status is specifically the value of the growth status of succulents under conditions of 40% humidity, 25℃ temperature, and 700 lux light. The learning rate; As a reward for user feedback, when a user rates the watering effect 8 out of 10. ; Discount factor; The value of the next maintenance condition is specifically the growth condition value under the conditions of 40% humidity, 25℃ temperature, and 700 lux light after watering. After generating 10 batches of control strategies, the parameter weight calculation model is calibrated, and the expression and voice triggering logic is optimized.

7. The method for multi-parameter monitoring and control of an AI smart flowerpot according to claim 2, characterized in that, In step S1.3, the alarm threshold is set as follows: when the temperature of tropical plants is below 12℃, subtropical plants below 8℃, succulents / desert plants below 10℃, and temperate plants below 5℃, the APP alarm will be triggered and an audio sound will be played. The alarm function can be turned off through the APP.

8. The multi-parameter monitoring and control method for an AI smart flowerpot according to claim 3, characterized in that, In step S2.2, the human infrared sensor is deployed in a transparent hole below the screen. A thermal compensation algorithm is used to eliminate the monitoring error caused by the close proximity of summer air temperature and human body temperature. If no human body is detected within 30 minutes, the screen brightness is automatically reduced or the screen is turned off to reduce power consumption.

9. The multi-parameter monitoring and control method for an AI smart flowerpot according to claim 4, characterized in that, Environmental impact factors in step S3.2 Water levels in rainy southern regions are dynamically adjusted according to regional climate characteristics. Adjusted to 0.7, and to 0.9 for arid regions in the north; semantic relevance. Assigning values ​​based on plant physiology principles, succulent plant light levels Adjusted to 0.85, humidity for tropical plants Adjusted to 0.

95.

10. The multi-parameter monitoring and control method for an AI smart flowerpot according to claim 5, characterized in that, In step S4.1, the automatic watering module's pipe is connected to the bottom of the flowerpot. The watering time is dynamically adjusted according to the soil moisture deviation value. For every 10% increase in the deviation value, the watering time increases by 3 seconds. When the moisture deviation of the succulent plant is 15%, the watering time is 4.5 seconds. The longest watering time for a single watering session shall not exceed 30 seconds.

Images