CN-122018611-A - Greenhouse environment self-adaptive regulation and control system and method based on vegetable growth stage identification

CN122018611ACN 122018611 ACN122018611 ACN 122018611ACN-122018611-A

Abstract

The application relates to the technical field of intelligent regulation and control of facility agriculture, in particular to a greenhouse environment self-adaptive regulation and control system and method based on vegetable growth stage identification, comprising four modules of growth stage identification, environment parameter monitoring, regulation and control strategy generation, regulation and control execution and feedback correction, wherein the system acquires vegetable canopy images through a high-definition RGB camera, fuses static visual characteristics and dynamic time sequence characteristics, adopts ResNet and transducer fusion models and a confidence verification mechanism, and accurately identifies the growth stage from a seedling stage to a mature stage; and calling a dynamically updated parameter reference library, generating a regulation strategy by combining deviation classification and priority dynamic regulation logic, controlling multiple types of equipment to execute stepless regulation, and correcting and optimizing a regulation effect through closed-loop feedback. The application solves the problems of misjudgment, fixed reference and stiffness response in the traditional regulation and control stage, improves the regulation and control accuracy and robustness, reduces the resource waste and improves the yield and quality of the assisted vegetables.

Inventors

FAN XUELIAN
GE FURONG
Xie Kankai
GUO HUANRU
LU YAN
He Dongchoulin
WANG QI
TAN YUMEI

Assignees

宁波市北仑侃宝果木专业合作社

Dates

Publication Date: 20260512
Application Date: 20260410

Claims (10)

1. Greenhouse environment self-adaptive regulation and control system based on vegetable growth stage identification, which is characterized by comprising: The growth stage identification module is configured to acquire a target greenhouse vegetable canopy image, extract static visual characteristics and dynamic time sequence characteristics of vegetables after preprocessing operations of denoising, segmentation and target region extraction are carried out on the target greenhouse vegetable canopy image, and lock the current vegetable growth stage by adopting a multi-feature fusion identification model in combination with confidence verification, wherein the multi-feature fusion identification model realizes stage judgment through weighted fusion calculation of the static characteristics and the dynamic characteristics, the confidence verification determines a stage locking mode according to a confidence value interval output by the model, and outputs a final growth stage result only when the confidence meets a preset threshold; An environmental parameter monitoring module configured to collect environmental data of the target greenhouse, the environmental data including temperature, air humidity, illumination intensity, soil-related parameters, and CO 2 concentration; The regulation and control strategy generation module is configured to call a preset growth stage environment parameter reference library, match the monitored environment data with an optimal parameter interval corresponding to the current growth stage, calculate a parameter deviation grade and generate an adaptive regulation and control strategy according to a preset regulation and control priority; and the regulation and control execution and feedback correction module is configured to control the greenhouse regulation and control equipment to execute corresponding operation according to the self-adaptive regulation and control strategy, and dynamically correct the regulation and control strategy by re-acquiring environmental data and vegetable canopy images to verify regulation and control effects.
2. The greenhouse environment adaptive control system based on vegetable growth stage identification according to claim 1, wherein the growth stage identification module is configured to collect a target greenhouse vegetable canopy image, extract static visual features and dynamic time sequence features of vegetables after performing preprocessing operations of denoising, segmentation and target region extraction on the target greenhouse vegetable canopy image, and lock a current vegetable growth stage by adopting a multi-feature fusion identification model in combination with confidence verification, wherein the multi-feature fusion identification model realizes stage judgment by weighted fusion calculation of the static features and the dynamic features, the confidence verification determines a stage locking mode according to a model output confidence value interval, and only outputs a final growth stage result when the confidence meets a preset threshold, and comprises: the vegetable growth stage comprises a seedling stage, a flowering stage, a fruiting stage and a maturation stage; When the number of the target vegetable leaves is detected to be smaller than or equal to a preset first number threshold value, and the area increase rate of the target vegetable leaves is larger than or equal to a preset first increase rate threshold value, judging that the vegetable leaves are in the seedling stage; When the number of the target vegetable leaves is larger than the first number threshold and smaller than a preset second number threshold, and the target vegetable plant height growth rate is larger than or equal to a preset second growth rate, judging that the seedling stage is the seedling stage; When the number of the target vegetable flowers is detected to be larger than or equal to a preset third number threshold value, and the confidence coefficient is detected to be larger than or equal to a preset first threshold value, judging that the flowering period is generated; When the diameter of the target vegetable fruits is detected to be larger than or equal to a preset second threshold value, judging the fruiting period; and when the ratio of the coloring pixels of the target vegetable fruits is detected to be larger than or equal to a preset third threshold value, and the leaf area increase rate is smaller than or equal to the preset third increase rate threshold value, judging the mature period.
3. The greenhouse environment adaptive control system based on vegetable growth stage identification according to claim 2, wherein the growth stage identification module is configured to collect a target greenhouse vegetable canopy image, extract static visual features and dynamic time sequence features of vegetables after performing preprocessing operations of denoising, segmentation and target region extraction on the target greenhouse vegetable canopy image, and lock a current vegetable growth stage by adopting a multi-feature fusion identification model in combination with confidence verification, wherein the multi-feature fusion identification model realizes stage determination by weighted fusion calculation of the static features and the dynamic features, the confidence verification determines a stage locking mode according to a model output confidence value interval, and only when the confidence meets a preset threshold, outputs a final growth stage result, and further comprises: The method for acquiring the target greenhouse vegetable canopy image through the high-definition RGB camera comprises the following steps of: when the target vegetable is in the seedling stage or the seedling stage, the acquisition frequency is a preset first frequency value; when the target vegetable is in the flowering period, fruiting period or maturity period, the acquisition frequency is a preset second frequency value; Performing a three-layer screening after the acquisition is completed, including: when the standard deviation of the gray value of the image is larger than or equal to a preset first standard deviation threshold value, the definition reaches the standard; when the occupancy ratio of the image vegetable canopy pixels is greater than or equal to a preset first occupancy ratio threshold value, the target area reaches the standard; when the image shielding pixel duty ratio is smaller than or equal to a preset second duty ratio threshold value, the shielding is up to the standard; When the definition, the target area and the shielding of the acquired image reach the standard, the acquired image is an effective image, otherwise, the camera is triggered to re-acquire; and when the number of the re-acquisition times is equal to a preset third threshold value and the effective image is not acquired, maintaining the last recognition result.
4. The greenhouse environment adaptive control system based on vegetable growth stage recognition according to claim 3, wherein the growth stage recognition module is configured to collect a target greenhouse vegetable canopy image, extract static visual features and dynamic time sequence features of vegetables after performing preprocessing operations of denoising, segmentation and target region extraction on the target greenhouse vegetable canopy image, and lock a current vegetable growth stage by adopting a multi-feature fusion recognition model in combination with confidence verification, wherein the multi-feature fusion recognition model realizes stage determination by weighted fusion calculation of the static features and the dynamic features, the confidence verification determines a stage locking mode according to a model output confidence value interval, and only when the confidence meets a preset threshold, the system further comprises: the static visual features comprise the number of blades, the area of the blades, the plant height, the color of the blades and the number of flowers and fruits, wherein the number of the blades is extracted through a communication area counting method, and only the blades with the area larger than or equal to a preset fourth threshold value are counted, and the area of the blades is converted through the pixel occupation ratio and the camera calibration coefficient; The dynamic time sequence features comprise a leaf area daily growth rate and a plant height daily growth rate which are calculated based on an image sequence in a preset first time threshold, and abnormal data exceeding a preset multiple standard deviation are removed when the growth rate is calculated.
5. The greenhouse environment adaptive control system based on vegetable growth stage recognition according to claim 4, wherein the growth stage recognition module is configured to collect a target greenhouse vegetable canopy image, extract static visual features and dynamic time sequence features of vegetables after performing preprocessing operations of denoising, segmentation and target region extraction on the target greenhouse vegetable canopy image, and lock a current vegetable growth stage by adopting a multi-feature fusion recognition model in combination with confidence verification, wherein the multi-feature fusion recognition model realizes stage determination by weighted fusion calculation of the static features and the dynamic features, the confidence verification determines a stage locking mode according to a model output confidence value interval, and only when the confidence meets a preset threshold, the system further comprises: the multi-feature fusion recognition model is a fusion model of ResNet and a transducer, the model training dataset comprises a preset number of labeled images of different vegetables and different growth stages, and the recognition accuracy is greater than or equal to a preset fifth threshold; Directly locking the current growth stage when the model output confidence is greater than or equal to a preset sixth threshold; when the model output confidence is smaller than the sixth threshold and larger than or equal to a preset seventh threshold, the stage with highest occurrence frequency in the near 3 recognition results is taken; and when the confidence coefficient of the model output is smaller than the seventh threshold value, the last effective recognition result is kept and a manual review prompt is triggered.
6. The greenhouse environment adaptive regulation and control system based on vegetable growth stage identification according to claim 5, wherein the regulation and control strategy generation module is configured to call a preset growth stage environment parameter reference library, match the monitored environment data with an optimal parameter interval corresponding to a current growth stage, calculate a parameter deviation level, and generate an adaptive regulation and control strategy according to a preset regulation and control priority, and comprises: The preset growth stage environment parameter reference library comprises growth stage parameters corresponding to tomatoes, cucumbers and lettuce, and each parameter is divided into an optimal interval, an early warning interval and a safety threshold value; when the same vegetable variety is regulated for 3 times continuously in the same growth stage and has abnormal growth, the upper limit or the upper regulation lower limit of the optimal interval of the parameter is automatically regulated.
7. The greenhouse environment adaptive regulation and control system based on vegetable growth stage identification according to claim 6, wherein the regulation and control strategy generation module is configured to call a preset growth stage environment parameter reference library, match the monitored environment data with an optimal parameter interval corresponding to a current growth stage, calculate a parameter deviation level, and generate an adaptive regulation and control strategy according to a preset regulation and control priority, and comprises: the deviation grade is determined through deviation rate calculation, and when the deviation rate is smaller than or equal to a preset first deviation threshold value, the deviation grade is micro-deviation; when the deviation rate is larger than the first deviation threshold and smaller than or equal to a preset second deviation threshold, the deviation grade is small deviation; when the deviation rate is larger than the second deviation threshold and smaller than or equal to a preset third deviation threshold, the deviation grade is middle deviation; When the deviation rate is greater than the third deviation threshold, the deviation level is a large deviation; The regulation priority is ordered according to illumination, temperature, CO 2 concentration, water and fertilizer parameters and air humidity by default; When the temperature or illumination exceeds the safety threshold, the regulation priority of the temperature or illumination automatically rises to the first level.
8. The adaptive regulation and control system for greenhouse environment based on identification of vegetable growth stage according to claim 7, wherein the regulation and control execution and feedback correction module is configured to control the greenhouse regulation and control device to execute corresponding operations according to the adaptive regulation and control strategy, and to dynamically correct the regulation and control strategy by re-acquiring environmental data and vegetable canopy images to verify regulation and control effects, comprising: The regulating and controlling equipment comprises an LED light supplementing lamp, a sunshade net, a ventilation fan, a heating pipe, a wet curtain cooling system, a CO 2 generator and a water and fertilizer integrated device; The operation state of the equipment is required to be inquired before the regulation operation is executed, if no fault exists, the operation is executed according to the regulation strategy, if the fault exists, the standby equipment is switched and the alarm is triggered, and if no standby equipment exists, the regulation level is reduced and the regulation duration is prolonged.
9. The greenhouse environment adaptive regulation system based on vegetable growth stage identification of claim 8, wherein the regulation execution and feedback correction module is configured to control the greenhouse regulation device to execute corresponding operations according to the adaptive regulation strategy, verify the regulation effect by re-acquiring environmental data and vegetable canopy images, dynamically correct the regulation strategy, and further comprise: when the deviation grade is micro-deviation, re-acquiring data within a preset first time threshold after regulation; when the deviation grade is small deviation or medium deviation, re-acquiring data within a preset second time threshold after regulation; when the deviation level is large deviation, re-acquiring data within a preset third time threshold after regulation; When the parameters return to the optimal interval and the vegetable canopy image is not abnormal, maintaining the current strategy; when the parameters are not regressed, the regulation and control intensity is upgraded according to the deviation grade; And when the image display is abnormal, immediately stopping current regulation, regulating parameters downwards, and updating parameter intervals corresponding to the reference library.
10. A greenhouse environment self-adaptive regulation and control method based on vegetable growth stage identification, which is characterized by being applied to the greenhouse environment self-adaptive regulation and control system based on vegetable growth stage identification as claimed in any one of claims 1 to 9, and comprising the following steps: Collecting a target greenhouse vegetable canopy image, extracting static visual characteristics and dynamic time sequence characteristics of vegetables after the preprocessing operation of denoising, segmentation and target region extraction is carried out on the target greenhouse vegetable canopy image, and locking a current vegetable growth stage by adopting a multi-feature fusion recognition model in combination with confidence level verification, wherein the multi-feature fusion recognition model realizes stage judgment by weighting fusion calculation of the static characteristics and the dynamic characteristics, the confidence level verification determines a stage locking mode according to a confidence level numerical value interval output by the model, and only when the confidence level meets a preset threshold value, outputs a final growth stage result; Collecting environmental data of the target greenhouse, wherein the environmental data comprise temperature, air humidity, illumination intensity, soil related parameters and CO 2 concentration; invoking a preset growth stage environment parameter reference library, matching the monitored environment data with an optimal parameter interval corresponding to the current growth stage, calculating a parameter deviation grade and generating an adaptive regulation strategy according to a preset regulation priority; And controlling the greenhouse regulation and control equipment to execute corresponding operation according to the self-adaptive regulation and control strategy, verifying the regulation and control effect by re-acquiring environmental data and vegetable canopy images, and dynamically correcting the regulation and control strategy.

Description

Greenhouse environment self-adaptive regulation and control system and method based on vegetable growth stage identification Technical Field The invention relates to the technical field of intelligent regulation and control of facility agriculture, in particular to a greenhouse environment self-adaptive regulation and control system and method based on vegetable growth stage identification. Background In facility agriculture large-scale planting, precise regulation and control of greenhouse environmental parameters (temperature, illumination, humidity and the like) directly influence the growth efficiency and quality of vegetables. The existing greenhouse environment regulation and control technology mostly depends on fixed preset parameters or artificial experience to formulate a regulation and control strategy, lacks accurate perception of the real-time growth stage of vegetables, causes mismatching of regulation and control actions and growth requirements of different stages of vegetables, often causes problems of excessive irrigation in a seedling stage, insufficient illumination in a fruiting stage and the like, causes waste of water resources and energy sources, and has difficulty in guaranteeing the yield and quality of the vegetables. While some technologies attempt to introduce image recognition auxiliary regulation, the problem of insufficient recognition precision generally exists, only single visual features are relied on, dynamic time sequence features are not fused, a confidence verification mechanism is lacked, misjudgment in a growth stage is easy to occur, meanwhile, the existing regulation reference library is mostly fixed parameters, cannot be dynamically updated according to the actual growth state of vegetables, regulation priority is stiff, quick response is difficult to occur when the temperature and illumination exceed emergency conditions of a safety threshold, and the risk of regulation mismatch is further aggravated. Therefore, a greenhouse environment self-adaptive regulation and control system with dynamic adaptation capability based on precise growth stage identification is needed, and the core pain point in the prior art is solved. Disclosure of Invention In view of the above, the present invention provides a greenhouse environment adaptive control system and method based on vegetable growth stage recognition, which aims to solve at least one of the problems of the background art. The invention provides a greenhouse environment self-adaptive regulation and control system based on vegetable growth stage identification, which comprises a growth stage identification module, a control module and a control module, wherein the growth stage identification module is configured to acquire a target greenhouse vegetable canopy image, extract static visual characteristics and dynamic time sequence characteristics of vegetables after preprocessing operations of denoising, segmentation and target region extraction are carried out on the target greenhouse vegetable canopy image, and adopt a multi-feature fusion identification model to combine with confidence level verification to lock the current vegetable growth stage, wherein the multi-feature fusion identification model realizes stage judgment through weighted fusion calculation of the static characteristics and the dynamic characteristics, the confidence level verification determines a stage locking mode according to a confidence level value interval output by the model, and outputs a final growth stage result only when the confidence level meets a preset threshold value; An environmental parameter monitoring module configured to collect environmental data of the target greenhouse, the environmental data including temperature, air humidity, illumination intensity, soil-related parameters, and CO 2 concentration; The regulation and control strategy generation module is configured to call a preset growth stage environment parameter reference library, match the monitored environment data with an optimal parameter interval corresponding to the current growth stage, calculate a parameter deviation grade and generate an adaptive regulation and control strategy according to a preset regulation and control priority; and the regulation and control execution and feedback correction module is configured to control the greenhouse regulation and control equipment to execute corresponding operation according to the self-adaptive regulation and control strategy, and dynamically correct the regulation and control strategy by re-acquiring environmental data and vegetable canopy images to verify regulation and control effects. In some embodiments, the growth stage identification module is configured to collect a target greenhouse vegetable canopy image, perform preprocessing operations of denoising, segmentation and target region extraction on the target greenhouse vegetable canopy image, extract static visual features and dynamic time sequence features o