Plant lighting system and method based on dynamic light quality adjustment of plant state
By combining image acquisition and real-time analysis of environmental monitoring data, using the Hidden Markov model to dynamically adjust the spectral characteristics of the plant lighting system, the problem that the plant lighting system in the existing technology cannot match the plant growth needs in time, and improve detection accuracy and responsiveness.
Patent Information
- Application Number
- CN202411257853.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-09
- Publication Date
- 2024-11-22
AI Technical Summary
The existing plant lighting systems lack real-time status detection capabilities and are unable to match plant growth needs in time, resulting in incomplete matching of spectrum and demand, affecting plant growth effects.
The data combined with the image acquisition unit and the environmental monitoring unit are used to conduct comprehensive analysis through the server to detect the growth status of the plant in real time, and the spectral characteristics of the lighting scheme are dynamically adjusted according to the detection results. The hidden Markov model is used for growth status prediction and abnormal diagnosis.
It improves the accuracy and timeliness of plant growth status detection, ensures the matching degree of lighting schemes and plant requirements, reduces growth problems caused by delayed state judgment, and enhances the system's response ability.
Smart Images

Figure CN119012461A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of plant lighting, and in particular to a plant lighting system and method for dynamically adjusting light quality based on plant status. Background Art
[0002] Ordinary plant lighting devices have some obvious defects in management. First, they usually lack state recognition and prediction capabilities, cannot adjust lighting conditions according to the specific stage of plant growth, and cannot predict the future growth trend of plants. Second, the lighting strategies provided by ordinary lighting devices are often relatively simple and cannot be dynamically adjusted according to the growth needs of plants.
[0003] In addition, ordinary plant lighting devices often require manual intervention to adjust lighting conditions and lack automated control, which limits management efficiency. They are also weak in response to abnormal situations and cannot promptly identify and respond to problems that may arise during plant growth. Moreover, ordinary lighting devices rarely use data to support decision-making and cannot help managers make more scientific decisions.
[0004] In terms of technical content and intelligence, ordinary plant lighting devices are usually low and cannot meet the needs of modern agricultural development. At the same time, due to the lack of standardization, the plant lighting market is relatively chaotic, and standardization is an important means to promote the industrialization of scientific research results and support the standardized development of the industry. In terms of energy consumption, traditional plant lighting devices usually have problems with high energy consumption and high operating costs compared to new light sources such as LEDs. Finally, ordinary plant lighting devices also lack specificity in spectral regulation and cannot provide the specific spectrum required for plant growth.
[0005] CN115016413A discloses an intelligent environmental control method for crop planting in a plant factory, comprising: modeling the crop planting process with a Markov decision process; constructing a sampling experience pool of an intelligent environmental control algorithm, designing a sampling learning strategy for a plant factory; designing a reward function to guide the intelligent environmental control algorithm; using the intelligent environmental control algorithm to learn the environmental control strategy, and outputting the environmental control strategy through a dual-path mechanism. However, if the state space definition is not comprehensive enough when using Markov decision process modeling to control the plant environment, some important environmental factors may be ignored, resulting in unsatisfactory control strategy effects. The intelligent environmental control algorithm relies on a sampling experience pool, and if the sample size in the experience pool is insufficient or unrepresentative, poor learning results may result. In addition, when the environment changes, old experience may no longer be applicable, resulting in a lag in model update.
[0006] CN118077450A discloses a plant growth lamp spectrum adjustment system based on wireless networking, including an air environment monitoring module, an air state analysis module, a soil environment monitoring module, a soil state analysis module, a growth spectrum analysis module, a wireless communication module, a control center and a cloud platform. The growth spectrum adjustment parameters of the growing plant corresponding to the current monitoring period are analyzed by the air environment state impact index and the soil environment state impact index, and the growth spectrum adjustment coefficient of the growing plant corresponding to the current monitoring period is obtained. Targeted matching analysis is performed based on the air environment factors at different positions of the plant during the growth process, and the influence of the soil environment at different positions of the plant during the growth process on the plant growth is also comprehensively considered, so that the plant growth lamp spectrum adjustment process is more accurate, so that the plant is exposed to the spectrum of the growth lamp to the best state. However, the system relies on the monitoring module to collect environmental data, and the environment may change very quickly, which may be insufficient in real-time adaptation, resulting in a lag in adjustment. The system also mainly relies on preset spectrum adjustment parameters, and lacks the ability to flexibly adjust for specific plant needs, making it difficult to make instant adjustments based on real-time feedback.
[0007] Therefore, how to improve the timeliness and flexibility of light regulation of plant lighting so as to provide light that is more reasonably matched with the environment while meeting the specific needs of plants is a problem that the present invention hopes to solve.
[0008] In addition, on the one hand, there are differences in understanding among those skilled in the art; on the other hand, the applicant studied a large number of documents and patents when making the present invention, but due to space limitations, not all details and contents are listed in detail. However, this does not mean that the present invention does not have the characteristics of these prior arts. On the contrary, the present invention already has all the characteristics of the prior art, and the applicant reserves the right to add relevant prior art to the background technology. Summary of the invention
[0009] Currently, the existing technology has a delay in the real-time detection of plant status, resulting in the spectrum of lighting applied to plants not fully matching the needs of the plants. In particular, when the insufficient growth of plants is caused by environmental factors, the current plant lighting control system cannot determine the environmental factors in a timely manner, and the effect of improving the growth status of plants by adjusting the light quality of lighting parameters is poor. Therefore, how to monitor or detect the growth status of plants in real time and accurately, and to determine in a timely manner whether the abnormal growth status is caused by light factors or environmental factors, is a problem that cannot be solved by the current existing technology.
[0010] In view of the deficiencies of the prior art, the present invention provides a plant lighting system with dynamic light quality adjustment based on plant status from a first aspect, comprising an image acquisition unit, an environmental monitoring unit, a lighting unit and a server. The image acquisition unit is used to acquire time-related growth images and spectral characteristics of plants; the environmental monitoring unit is used to monitor environmental parameters in the process of plant growth in real time; the lighting unit is used to illuminate the plants according to a first lighting scheme; the server forms plant growth sequence data according to the growth images, spectral characteristics and environmental parameters, determines the growth status of the plants in real time based on the growth sequence data and a first model formed by pre-training, determines a second lighting scheme based on a second model including mapping data of plant growth status and lighting scheme, and adjusts the spectral characteristics of the first lighting scheme based on the second lighting scheme.
[0011] The plant lighting system combines the data of the image acquisition unit and the environmental monitoring unit and uses the server for comprehensive analysis to improve the accuracy of plant growth status detection. The system can update the plant growth image and environmental parameters in real time, ensuring the timeliness of the detection results, thereby providing a more accurate growth status assessment for the plant.
[0012] According to a preferred embodiment, a method for determining the growth status of a plant in real time based on growth sequence data and a first model formed by pre-training includes: decoding an input growth sequence to obtain a hidden state sequence that can derive the input growth sequence with maximum probability; classifying the hidden states in the hidden state sequence to obtain the plant growth level corresponding to each point in the growth sequence; calculating the time series characteristic parameters of the growth sequence, comparing them with the threshold of the time series characteristic parameters, and determining the plant growth state of the growth sequence; and judging the health status of the plant based on a combination of the plant growth level and growth state of the growth sequence.
[0013] By decoding the growth sequence data and using the hidden Markov model, the system can determine the growth status of the plant in real time, including the plant growth level and health status. This method further enhances the accuracy of the plant growth status judgment by comparing it with the threshold of the time series characteristic parameters.
[0014] According to a preferred embodiment, the method for constructing the first model includes: preprocessing the historical data of plant growth data, converting the input growth sequence into a signal energy sequence; inputting the signal energy sequence into a hidden Markov model for training and establishing an energy sequence model, wherein the hidden states in the energy sequence model are classified to correspond to different plant growth levels; calculating and determining the threshold of the time series characteristic parameters related to the detection of plant growth status; inputting the plant growth sequence to be detected and detecting the growth status of the plant.
[0015] The construction method of the first model improves the accuracy of plant growth state detection by preprocessing historical data and converting it into a signal energy sequence, and then inputting it into a hidden Markov model for training. The hidden state classification of the model corresponds to different plant growth levels, making the growth state detection more detailed and specific.
[0016] According to a preferred embodiment, the method of converting an input plant growth sequence into a signal energy sequence includes: calculating the energy value and average energy of historical data; judging the plant growth level based on the average energy within a time window; wherein the plant growth sequence includes soil environmental parameters, temperature parameters, spectral parameters, normalized vegetation index and leaf characteristic parameters.
[0017] The method of converting plant growth sequences into signal energy sequences improves the accuracy of judging plant growth levels by calculating the energy value and average energy of historical data and combining time window analysis. This method integrates multiple environmental parameters such as soil, temperature, and spectrum, providing comprehensive data support for accurate assessment of plant growth status.
[0018] According to a preferred embodiment, the method of inputting a signal energy sequence into a hidden Markov model for training includes: expressing the signal energy sequence as an observation probability using a probability mass function of a Poisson distribution; setting initial parameters; calculating a maximum likelihood estimation function of the observation probability; maximizing the maximum likelihood estimation function to obtain model parameters for the next iteration, and iterating until the model parameters converge or a preset number of iterations is reached.
[0019] Through the training of the hidden Markov model, the present invention can express the observation probability with the probability mass function of the Poisson distribution, and optimize the model parameters through the maximum likelihood estimation function. This iterative training method improves the accuracy of the model's prediction of plant growth status and ensures the convergence and reliability of the model parameters.
[0020] According to a preferred embodiment, the lighting scheme includes light intensity parameters, light quality parameters, light cycle parameters and fill light parameters, and the server generates fixed lighting parameters of the lighting device and fill light parameters of the fill light device based on the physical characteristics of the immovable lighting device, the physical characteristics of the fill light device and the second lighting scheme, wherein the fill light parameters include lighting coordinates and dynamic lighting parameters.
[0021] The formulation of the lighting plan takes into account multi-dimensional parameters such as light intensity, light quality, photoperiod and fill light. The server generates more accurate lighting parameters based on the physical characteristics of the lighting device and fill light device, as well as the second lighting plan. This improves the matching degree of the lighting plan to the plant growth requirements.
[0022] According to a preferred embodiment, the first model is based on the plant growth state of the regional unit and a hidden state sequence related to the plant growth state; the first model determines the cause of the abnormal plant growth state based on the hidden state sequence.
[0023] The server determines the relevant hidden state sequence by analyzing the plant growth state and the first model, and then determines the cause of the abnormal plant growth state. This analysis method not only improves the accuracy of diagnosing plant growth problems, but also makes the judgment of environmental factors more timely and accurate, which helps to quickly take countermeasures.
[0024] From a second aspect, the present invention provides a plant lighting method for dynamic light quality adjustment based on plant status, the method comprising: collecting time-related growth images and spectral characteristics of plants; real-time monitoring of environmental parameters during plant growth; irradiating plants according to a first lighting scheme; forming plant growth sequence data according to the growth images, spectral characteristics and environmental parameters, determining the growth status of the plants in real time based on the growth sequence data and a first model formed by pre-training, determining a second lighting scheme based on a second model including mapping data of plant growth status and lighting schemes, and adjusting the spectral characteristics of the first lighting scheme based on the second lighting scheme.
[0025] The plant lighting method collects the growth images and spectral characteristics of plants in real time through the image acquisition unit, ensuring the timely update of plant growth detection data. Combined with the environmental parameters monitored in real time by the environmental monitoring unit, the server can quickly form growth sequence data reflecting the current state, providing a data basis for the detection of plant growth status. In addition, through the server's comprehensive analysis of growth images, spectral characteristics and environmental parameters, the method can determine the growth status of plants in real time based on the first model, avoiding the response lag problem caused by data collection and processing delays in traditional detection methods.
[0026] According to a preferred embodiment, a method for determining the growth status of a plant in real time based on growth sequence data and a first model formed by pre-training includes: decoding an input growth sequence to obtain a hidden state sequence that can derive the input growth sequence with maximum probability; classifying the hidden states in the hidden state sequence to obtain the plant growth level corresponding to each point in the growth sequence; calculating the time series characteristic parameters of the growth sequence, comparing them with the threshold of the time series characteristic parameters, and determining the plant growth state of the growth sequence; and judging the health status of the plant based on a combination of the plant growth level and growth state of the growth sequence.
[0027] This method decodes the growth sequence data in real time to quickly obtain the hidden state sequence of the plant growth level, thus achieving instant evaluation of the plant growth status. The calculation of the time series feature parameters and comparison with their thresholds further ensures the timeliness of the plant growth status detection, providing data support for timely adjustment of the lighting scheme. In addition, by classifying the hidden state sequence and combining it with the plant growth level, this method can timely judge the health status of the plant and reduce plant growth problems caused by delayed status judgment. This fast detection and evaluation process significantly reduces the time delay in the detection process.
[0028] According to a preferred embodiment, the first model is based on the plant growth state of the regional unit and a hidden state sequence related to the plant growth state; the first model determines the cause of the abnormal plant growth state based on the hidden state sequence.
[0029] The server uses the plant growth state and the first model to promptly determine the hidden state sequence related to the plant growth state, providing a basis for quickly identifying abnormal plant growth. This real-time analysis method based on regional units ensures a rapid response to the cause of abnormal plant growth state. By analyzing the hidden state sequence, the server can quickly determine the cause of abnormal plant growth state, avoiding the problem of missing the best adjustment time due to delayed diagnosis. This method improves the response speed of plant growth detection and helps to take necessary measures to improve the plant growth environment in a timely manner. BRIEF DESCRIPTION OF THE DRAWINGS
[0030] Figure 1 It is a simplified schematic diagram of module connection relationship of a plant lighting system for dynamic light quality adjustment based on plant status provided by the present invention;
[0031] Figure 2 It is a logical schematic diagram of a plant lighting system for dynamic light quality adjustment based on plant status provided by the present invention;
[0032] Figure 3 It is a schematic diagram of the process of the plant lighting method for dynamic light quality adjustment based on plant state provided by the present invention;
[0033] Figure 4 It is a schematic diagram of the process of detecting the growth status of a plant provided by the present invention;
[0034] Figure 5 It is a schematic diagram of the process of converting a growth sequence into a signal energy sequence provided by the present invention.
[0035] Reference numerals list
[0036] 100: image acquisition unit; 200: environment monitoring unit; 300: lighting unit; 310: lighting device; 320: fill light device; 400: server; 410: first model; 420: second model; 430: adjustment module; 500: terminal. DETAILED DESCRIPTION
[0037] The following is a detailed description with reference to the accompanying drawings.
[0038] Example 1
[0039] Currently, the existing technology has a delay in the real-time detection of plant status, resulting in the spectrum of lighting applied to plants not fully matching the needs of the plants. In particular, when the insufficient growth of plants is caused by environmental factors, the current plant lighting control system cannot determine the environmental factors in a timely manner, and the effect of improving the growth status of plants by adjusting the light quality of lighting parameters is poor. Therefore, how to detect the growth status of plants in real time and accurately, and to determine in a timely manner whether the abnormal growth status is caused by light factors or environmental factors, is a problem that cannot be solved by the current existing technology.
[0040] In view of the shortcomings of the prior art, the present invention provides a plant lighting system with dynamic light quality adjustment based on plant status, such as Figure 1 As shown, it includes an image acquisition unit 100, an environment monitoring unit 200, a lighting unit 300 and a server 400. The image acquisition unit 100 is connected to the server 400 via a wired or wireless network, and the collected images and spectral data are transmitted to the server 400 in real time. The environment monitoring unit 200 is also connected to the server 400, and the monitored environmental parameters such as soil moisture, temperature, etc. are transmitted in real time. The lighting unit 300 is connected to the server 400, receives the lighting scheme adjusted by the server 400 according to the plant growth status, and adjusts the lighting parameters in real time.
[0041] The image acquisition unit 100 is used to collect time-related growth images and spectral characteristics of plants. Preferably, the image acquisition unit 100 can be mounted on a drone, or fixed in the space above the plant, such as hanging on the ceiling, mounted on a bracket, etc. Further preferably, the image acquisition unit 100 is mounted on a drone, which has more advantages. In a plant factory, the area in the planting shed is also large. If the image acquisition unit 100 is set in a fixed manner, even if the image acquisition angle is adjustable, the acquisition angle of the plant growth image and spectral characteristics collected by the image acquisition unit 100 is limited. The acquisition of plant growth images in some areas will be affected by light and the acquisition distance, resulting in large errors in the acquisition data of plant images and their spectral characteristics. The image acquisition unit 100 carried by a drone is used to collect the growth image and spectral characteristics of the plant. With the advantage of the drone in dynamic monitoring and flexible adjustment of the flight path, the acquisition distance and acquisition angle between the image acquisition unit 100 and the plant can be relatively small, and the flight altitude can be adjusted, which optimizes the acquisition conditions, so that the image acquisition unit 100 has a small difference in the acquisition of the growth image and spectral characteristics of the plants in each unit area, thereby reducing or even eliminating the acquisition error of the plant growth. Preferably, the benefits of using a drone to carry the image acquisition unit 100 also include that the image acquisition unit 100 can flexibly collect the growth image and spectral characteristics of the plants in the same unit area at different heights and different angles, and even collect the growth image and spectral characteristics of a single plant. Preferably, the image acquisition unit 100 installed on the drone does not require wiring.
[0042] The environmental monitoring unit 200 is used to monitor the environmental parameters of the plant growth process in real time. The environmental monitoring unit 200 includes a first temperature sensor, a first humidity sensor, and a first gas sensor for monitoring the carbon dioxide concentration in the air for monitoring the plant surrounding environment; and also includes a second temperature sensor and a second humidity sensor, a nitrogen, phosphorus and potassium NPK sensor, a soil pH sensor, etc. for monitoring the soil environment.
[0043] The lighting unit 300 is used to illuminate the plants according to the first lighting scheme. The lighting unit 300 includes a fixed lighting device 310 and a movable supplementary light device 320.
[0044] The lighting device 310 uses an LED light source that provides a variety of spectra to illuminate the plants. Specifically, a variety of LED light sources are distributed in different positions of the lighting device 310 to ensure uniform illumination. The fixing bracket is used to firmly fix the lighting device 310 in a specific position to ensure its stability and safety. The power control system is electrically connected to the LED light source in the lighting device 310, and is used to control the switch and brightness of the lighting device 310, including a timing switch and a dimming function. The lighting device 310 is also provided with a heat dissipation element. For example, if a high-power lamp is used, a heat dissipation element such as a heat dissipation fan or a heat sink needs to be provided to prevent overheating from affecting the life of the LED light source.
[0045] Preferably, the fixed position lighting device 310 is installed above the plant growth area to ensure that it is at an appropriate height from the plants. The lighting intensity and time are set by the first control module to ensure the implementation of the first lighting solution.
[0046] Preferably, the movable fill-light device 320 includes a movable bracket, a fill-light lamp, a driving device, and a second control module. The movable bracket includes a track or a slide rail system, which can move the fill-light device 320 horizontally or vertically within a certain range. The fill-light lamp can be a lamp of the same type as the fixed lamp, which is used to supplement the light at a specific angle or position. The driving device can be an electric motor or a servo motor to control the movement of the fill-light device 320. The second control module is linked to the first control module of the fixed lighting device 310, and can automatically adjust the fill-light position and light intensity according to the growth status of the plant.
[0047] Specifically, a track is installed above the plant growth area, and the fill light device 320 is fixed on a movable bracket. The moving path and irradiation timing of the fill light device 320 are set to ensure that necessary light is provided at different stages of plant growth.
[0048] The server 400 is used to process the data, determine the growth status of the plant and adjust the spectral characteristics. Specifically, the processing steps include:
[0049] S410: Forming plant growth sequence data according to the growth image, spectral characteristics and environmental parameters, and determining the growth status of the plant.
[0050] The growth sequence includes biomass indicators, photosynthetic capacity, health indicators, soil environment, gas environment, growth cycle information and growth rate.
[0051] Biomass indicators include plant height, crown width, dry weight and wet weight.
[0052] The height of plants is measured regularly to assess their growth rate and status. The canopy area of the plant is measured to reflect the growth status of the plant. The dry matter weight and wet weight of the plant are assessed to assess the growth status. For example, the wet weight of the plant is estimated using a humidity sensor. Near infrared spectroscopy (NIRS), laser scanning, etc. can be used to estimate the dry matter weight of the plant without destroying the plant.
[0053] Photosynthetic capacity includes chlorophyll content and photosynthetic rate.
[0054] The photosynthesis rate and the health status of the plant are indirectly inferred based on the spectral image collected by the image collection unit 100. Preferably, by analyzing the reflectivity of the plant to light of different wavelengths, its chlorophyll content can be inferred and the photosynthesis capacity can be further evaluated.
[0055] The health index includes leaf color and pest index. The color change of the leaf (such as yellowing, fading, etc.) can be obtained by analyzing the image of the plant by the server 400. The server 400 can also evaluate the plant's pest situation and its impact on growth based on the plant's image.
[0056] Soil environment includes soil moisture and soil nutrient parameters. Soil nutrient parameters include the concentration of major nutrients such as nitrogen, phosphorus, potassium, etc. Gas environment includes air temperature and humidity.
[0057] Growth cycle information includes growth stage information. Time series data is formed by regularly observing and recording the growth stages of plants (such as germination, flowering, fruiting, etc.).
[0058] By recording the growth rate of plants over a specific period of time (such as daily growth height), growth rate information can be inferred.
[0059] S420: Determine a second lighting scheme based on the second model 420 including mapping data of plant growth status and lighting schemes.
[0060] S430: Adjusting the spectral characteristics of the first lighting scheme based on the second lighting scheme.
[0061] The image acquisition unit 100 sends the plant growth image and spectral characteristic data to the server 400. The environmental monitoring unit 200 sends the real-time monitored environmental parameter data to the server 400. The preprocessing module in the server 400 sends the processed plant growth sequence data to the first model 410. The first model 410 sends the decoded hidden state sequence to the second model 420. The second model 420 sends the second lighting scheme generated according to the plant growth state and mapping data to the adjustment module 430. The adjustment module 430 sends the adjusted spectral characteristic parameters to the fill light device 320 for real-time fill light control. The first model 410 also sends the determined cause of the abnormal plant growth state to the terminal 500 for management personnel to analyze and make decisions. Through the above information transmission processes, the system can efficiently collect, process, analyze and make decisions on data, and achieve accurate control and timely response to the plant growth environment.
[0062] The plant lighting system improves the accuracy of plant growth status detection by combining the data of the image acquisition unit 100 and the environmental monitoring unit 200 and performing comprehensive analysis using the server 400. The system can update the growth image and environmental parameters of the plant in real time, ensuring the timeliness of the detection results, thereby providing a more accurate growth status assessment for the plant.
[0063] Preferably, if Figure 2 As shown, the growth status of the plant is determined in real time based on the growth sequence data and the first model 410 formed by pre-training.
[0064] S411: Decode the input growth sequence to obtain a hidden state sequence that can derive the input growth sequence with the maximum probability. Classify the hidden states in the hidden state sequence to obtain the plant growth level corresponding to each point in the growth sequence. Calculate the time series feature parameters of the growth sequence, compare them with the threshold of the time series feature parameters, and determine the plant growth state of the growth sequence. Determine the health status of the plant based on the combination of the plant growth level and growth state of the growth sequence.
[0065] By decoding the growth sequence data and using the hidden Markov model, the system can determine the growth status of the plant in real time, including the plant growth level and health status. This method further enhances the accuracy of the plant growth status judgment by comparing it with the threshold of the time series characteristic parameters.
[0066] S412: Figure 4 As shown, the method for constructing the first model 410 includes:
[0067] S4121: Preprocess the historical data of plant growth data.
[0068] Specifically, the historical data of the growth sequence of plant growth is cleaned and preprocessed, for example, outliers are deleted, zero values in the historical data are removed, and the data type is adjusted so that the processed data meets the requirements of the energy sequence model for input data.
[0069] S4122: If Figure 5 As shown, the input growth sequence is converted into a signal energy sequence. The construction method of the first model 410 pre-processes the historical data and converts it into a signal energy sequence, and evaluates the plant growth level based on the average energy in the time window. Then the hidden Markov model is input for training, thereby improving the accuracy of plant growth state detection.
[0070] Specifically, for plant growth sequences, since the trained hidden Markov model needs to model the growth state of the plant, the training data needs to reflect the growth level of the sample points. Plant growth data may fluctuate greatly, and the growth value of the sample points in a single area may not accurately represent its actual growth level. Therefore, it is necessary to refer to the growth data of the previous window time to comprehensively evaluate the growth level. This requires the introduction of the energy characteristics of the time series signal. The energy characteristics of the time series signal can describe the energy distribution of the signal in the time domain, and are often used to analyze the intensity of plant growth, growth periodicity, and overall growth energy.
[0071] The calculation formula for the average energy of discrete time series signal is:
[0072]
[0073] Where N represents the length of the time series signal (the number of data points), and x[n] represents the value of the time series signal at the nth time point. |x[n]| 2 Represents the square of its absolute value. This formula is obtained by summing the square of the amplitude of the time series signal at each time point and then dividing it by the signal length to obtain the average energy of the signal.
[0074] Use the average energy E in the time window avg To evaluate the growth level of plants. By calculating the average energy value in each sample window, the original growth load sequence is converted into a signal energy sequence, and these data will be used as the training input of the hidden Markov model. In this way, the hidden Markov model can learn the distribution characteristics of plant growth states and the transition probabilities between states.
[0075] According to a preferred embodiment, the time window can be set to include ten sample points. A sample point is a unit of growth data corresponding to each timestamp in the time series data. If the time window is set too small, the average energy may not accurately reflect the growth state and only represent the growth load value of a single sample point; if it is set too large, it may cause a time lag in the calculated energy value, affecting the model's real-time response to changes in plant growth status. By properly setting the time window, the average energy can be calculated more accurately, providing accurate training data for the hidden Markov model.
[0076] S4123: Input the signal energy sequence into the hidden Markov model for training and establish an energy sequence model.
[0077] The hidden Markov model consists of a set of parameters that describe the state transition and observation generation process in the model, including the initial state probability vector π, the state transition probability matrix A, and the observation probability matrix B. The hidden Markov model is generally represented as a triple: HMM = (π, A, B).
[0078] Among them, π represents the initial state probability vector, A represents the state transition probability matrix, and B represents the observation probability matrix.
[0079] (1) The plant signal energy sequence is expressed as the observation probability using the probability mass function of the Poisson distribution.
[0080] The growth state sequence of the plant can be regarded as the observation sequence of the hidden Markov model. Since the observation sequence is a discrete count time series, it can be assumed that its distribution is Poisson distribution, and the observation probability can be expressed by the probability mass function of Poisson distribution:
[0081]
[0082] Among them, b i (ot) indicates that it is in state S at time step t i The probability of observing data ot under the condition λ i represents the mean of the Poisson distribution, that is, the average number of events (count data) observed in this state. During the training process, the hidden Markov model adjusts λ according to the observed data. i to better fit the data.
[0083] (2) Set the initial parameters of the hidden Markov model.
[0084] Initialize state probability π0, state transition probability A0 and observation probability B0.
[0085] (3) Use the forward-backward algorithm to calculate the maximum likelihood estimation function Q of the observed data.
[0086] (4) Use the expectation maximization algorithm to maximize the maximum likelihood estimation function Q and obtain the model parameters for the next iteration: the initial state probability vector π, the state transition probability matrix A, and the observation probability matrix B.
[0087] (5) Iterate until the model parameters converge or the preset number of iterations is reached.
[0088] Repeat steps (3) and (5) until the parameters converge or the preset number of iterations is reached. The hyperparameters during training include the number of hidden states n, π0, A0, B0 and the maximum number of iterations I of the hidden Markov model.
[0089] (6) Merge similar hidden states.
[0090] If the Poisson distribution parameters corresponding to different hidden states are similar, these similar hidden states need to be merged to ensure the accuracy and discrimination of the model.
[0091] (7) Calculate the distribution probability and transition direction of each hidden state.
[0092] The training sequence is decoded using the trained hidden Markov model to obtain the hidden state corresponding to each sample point, and the distribution of the hidden state is statistically analyzed to determine the stable state and transition state, and calculate the transition probability of the transition state.
[0093] S4124: Classify hidden states in the energy sequence model to correspond to different plant growth levels.
[0094] Specifically, the hidden states are classified into different plant growth states according to their distribution probability and transition direction.
[0095] The stable state indicates the stable period of plant growth, and the transition state indicates the transition period of plant growth. The growth state of the plant can be accurately judged by the size and transfer direction of the λ value. For example, an increase in load may indicate that the plant has entered a rapid growth period, and a decrease in load may indicate that the plant has entered a mature period or has been affected by adversity. In this way, the hidden Markov model can provide a powerful tool for the detection and analysis of plant growth status.
[0096] In the above content, "load" refers to the various biophysical or physiological stresses that plants endure during their growth, such as water, nutrient requirements, light intensity, temperature changes, etc. These load factors affect the growth state and growth rate of plants. When using the hidden Markov model (HMM) to analyze the growth state of plants, converting these load factors into signal energy sequences can help the model more accurately identify and predict the growth state of plants. In short, "load" here represents the sum of various environmental and physiological conditions that affect the normal growth of plants.
[0097] S4125: Calculate and determine the threshold of the time series characteristic parameter related to the plant growth state detection. Input the plant growth sequence to be detected and detect the growth state of the plant.
[0098] (1) In order to determine the growth status of plants, it is necessary to calculate the time series characteristics of plant growth data and compare them with the preset threshold. The selected time series characteristics should be able to reflect the dynamic characteristics of plant growth from both statistical and temporal aspects, and have the ability to distinguish different growth states.
[0099] The thresholds of the time series characteristic parameters include the coefficient of variation (CV X ), standard deviation (STD), and zero crossing rate (ZCR).
[0100] Coefficient of variation (CV X ) is a statistic used to describe the variability of a sequence, and its calculation formula is:
[0101] CV X =(σ / μ)×100%.
[0102] X represents the time series, μ represents the mean, and σ represents the standard deviation. The larger the value of the coefficient of variation, the greater the variability of the time series relative to the mean. In the present invention, the size of the coefficient of variation is used to test the fluctuation of the time series.
[0103] The standard deviation (STD) is used to assess the magnitude of series fluctuations.
[0104] The zero crossing rate (ZCR) is used to describe the characteristics of the frequency of change of the signal waveform and is calculated as:
[0105]
[0106] Where I(·) represents the indicator function, which is 1 when the condition in the brackets is met, otherwise it is 0. ZCR represents the number of positive and negative changes in the signal between adjacent samples.
[0107] (3) The coefficient of variation threshold is calculated as follows:
[0108] For each sample point in the growth time series used for training, take a subsequence of a time window from the sample point forward, and calculate the coefficient of variation of the subsequence to obtain the coefficient of variation sequence. The coefficient of variation sequence is averaged to obtain the coefficient of variation threshold CV thresh .
[0109] (4) Calculate the standard deviation threshold.
[0110] For the plant growth sequence in the training data, select a subsequence within a time window, calculate its standard deviation, and then find the average value to obtain the standard deviation threshold STD thresh .
[0111] (5) Calculate the zero-crossing rate threshold.
[0112] For the plant growth sequence in the training data, an improved zero-crossing rate calculation method is used, including Gaussian filtering smoothing and weighted calculation, to detect high-frequency vibration in the sequence. The improved ZCR is calculated for the subsequence to obtain the zero-crossing rate sequence, and then the average is calculated to obtain the zero-crossing rate threshold ZCR. thresh .
[0113] Through these thresholds, the growth status of plants can be effectively judged. For example, by comparing the actual observation value with the threshold, it can be identified whether the plant growth is normal or whether there is an abnormality. These time series features and thresholds help the hidden Markov model to more accurately evaluate the growth status of plants.
[0114] As described above, after the hidden Markov model is trained to form the first model 410, various parameters of the plant are input into the first model 410, and the growth state of the plant can be determined.
[0115] Preferably, the lighting scheme includes light intensity parameters, light quality parameters, light cycle parameters and fill light parameters. The server 400 generates fixed lighting parameters of the lighting device 310 and fill light parameters of the fill light device 320 based on the physical characteristics of the immovable lighting device 310, the physical characteristics of the fill light device 320 and the second lighting scheme, wherein the fill light parameters include lighting coordinates and dynamic lighting parameters.
[0116] The formulation of the lighting scheme takes into account multi-dimensional parameters such as light intensity, light quality, light cycle and fill light. The server 400 generates more accurate lighting parameters according to the physical characteristics of the lighting device 310 and the fill light device 320 and the second lighting scheme. This method improves the matching degree of the lighting scheme to the plant growth requirements.
[0117] For example, the tomato plant growth status analysis data is shown below.
[0118] Biomass indicators: height was 60 cm (10 cm lower than expected); crown width was 40 cm (normal range); dry weight estimated by NIRS was 150 g (slightly lower than expected); wet weight estimated by humidity sensor was 1.2 kg (normal range).
[0119] Photosynthetic capacity: Through spectral analysis, the chlorophyll content (SPAD value) was 35 (lower than the ideal value of 40-50); according to spectral image analysis, the photosynthetic rate was 20% lower than expected.
[0120] Health indicators: Image analysis shows that some leaves are slightly yellowing; the pest and disease index is 0.
[0121] Soil environment: Soil moisture is 65% (normal range); soil nutrients are: nitrogen is slightly low, phosphorus and potassium are normal.
[0122] Gas environment: temperature is 25°C (normal range), humidity is 70% (normal range).
[0123] Growth cycle information: In the early stages of flowering.
[0124] Growth rate: Height increase of approximately 1 cm per day (lower than the expected 1.5 cm / day).
[0125] Current lighting scheme: light intensity is 250 μmol / m 2 / s; illumination time is 16 hours / day; spectrum composition is: red light 60%, blue light 30%, green light 10%.
[0126] Based on the above data, the first model 410 determines that the plant growth state is not good, and the main problems are low photosynthesis efficiency and slow growth. The first model 410 recommends adjusting the lighting scheme to promote growth. The first model 410 sends the plant growth state and mapping data to the second model 420.
[0127] The second lighting scheme determined by the second model 420 is: increasing the light intensity to 300 μmol / m 2 / s; keep the lighting time unchanged at 16 hours / day; adjust the spectral composition to: red light 65%, blue light 25%, far-red light 10%.
[0128] Specifically, the steps of the second model 420 generating the second lighting scheme are as follows.
[0129] S421: Use the following formula to standardize the input growth parameters to the range of 0 to 1.
[0130] X norm =(XX min ) / (X max -X min ).
[0131] Where X is the original value, X min and X ma are the minimum and maximum possible values for the parameter, respectively. norm is the normalized parameter value.
[0132] S422: Assign a weight to each growth parameter, as shown in Table 1, the sum of which is 1.
[0133] Table 1: Examples of growth parameter assignment weights.
[0134]
[0135] S423: Calculate the comprehensive growth condition index using the weighted average method:
[0136] Growth Index =Σ(W i *X inorm ).
[0137] Among them, W i is the weight of each parameter, X inorm is the normalized parameter value.
[0138] S424: Determine the light intensity adjustment.
[0139] Based on the comprehensive growth condition index, the adjustment of light intensity is determined using the following formula:
[0140] Intensity new =Intensity current *(1+(Target Index -Growth Index )*K)
[0141] =Intensity current ×(1+(0.85-Growth Index )×0.2).
[0142] Among them, Intensity new is the adjusted light intensity, Intensity current is the current light intensity; Growth Index It is a comprehensive growth index. Index is the target growth index (usually 0.8-0.9), and K is the adjustment factor (e.g. 0.2).
[0143] S425: Spectral composition adjustment.
[0144] Far-red light is mainly used to regulate plant height and influence plant morphogenesis. According to the deviation of specific growth parameters, the spectral composition is adjusted. If the plant height is lower than the target, the proportion of far-red light is increased to promote stem elongation.
[0145] Far_Red new =Far_Red current +0.05*(Target Height -Current Height ) / Target Height .
[0146] Far_Red new is the adjusted far-red spectral component ratio. current is the current far-red light spectrum component ratio. HeightIndicates the target height, Current Height Indicates the current altitude.
[0147] Blue light mainly affects chlorophyll content and stomatal opening and closing, regulating plant morphology and photosynthesis. If the chlorophyll content is lower than the target, increase the proportion of blue light to promote chlorophyll synthesis.
[0148] Blue new =Blue current +0.03*(Target Chlorophyll -Current Chlorophyll ) / Target Chlorophyll .
[0149] Blue new is the adjusted blue light spectrum component ratio. current is the current blue light spectrum component ratio. Chlorophyll Indicates chlorophyll content, Current Chlorophyll Indicates the current chlorophyll content.
[0150] Red light has a significant effect on photosynthesis and biomass accumulation and is the main light source for plant growth. If biomass is lower than the target, increase the proportion of red light to promote photosynthesis and biomass accumulation.
[0151] Red new =Red current +0.04*(Target Biomass -Current Biomass ) / Target Biomass .
[0152] Red new is the adjusted red light spectrum component ratio. current is the current red light spectrum component ratio. Biomass is the target bioequivalent, Current Biomass is the current bioequivalent.
[0153] Green light can penetrate the canopy and improve overall photosynthetic efficiency, especially when densely planted or large plants are used. If the overall photosynthetic efficiency is lower than the target, increase the proportion of green light appropriately to improve photosynthetic efficiency inside the canopy.
[0154] Green new =Green current +0.02*(Target Photosynthesis -
[0155] Current Photosynthesis ) / Target Photosynthesis .
[0156] Green new is the adjusted green light spectrum component ratio. current Is the current green light spectrum component ratio. Photosynthesis is the target photosynthetic efficiency, Current Photosynthesis is the current photosynthetic efficiency.
[0157] S426: Adjust the illumination time. The illumination time usually remains unchanged, but if the total illumination is insufficient, it can be appropriately extended.
[0158] Duration new =Duration current *(Target DLI / Current DLI ).
[0159] Duration new Duration is the adjusted illumination time. current Is the current lighting time. DLI Current is the target daily accumulated light quantum flux density. DLI It is the current day's accumulated light quantum flux density. DLI is the daily light integral.
[0160] S427: Check constraints to ensure that the adjusted parameters are within a reasonable range.
[0161] Light intensity is 100~1000μmol / m 2 / s; the spectral composition is the sum of each component is 100%; the illumination time is 12 to 18 hours / day.
[0162] S428: Based on the above adjustments, a final second lighting solution is generated.
[0163] Light intensity is Intensity new ; The spectrum composition is the new percentage of each band; The illumination time is Duration new .
[0164] S430: Adjusting the spectral characteristics of the first lighting scheme based on the second lighting scheme.
[0165] S431: The original first lighting scheme is as follows.
[0166] Light intensity: 250 μmol / m 2 / s;
[0167] Lighting time: 16 hours / day;
[0168] Spectral composition: red light 60%, blue light 30%, green light 10%.
[0169] S432: The adjusted first lighting scheme is as follows.
[0170] Light intensity: gradually increased to 300 μmol / m 2 / s (daily increase of 10 μmol / m 2 / s).
[0171] Light exposure time: Keep 16 hours / day unchanged.
[0172] The spectral composition is adjusted to:
[0173] Red light: gradually increase from 60% to 65% (1% increase per day);
[0174] Blue light: gradually reduce from 30% to 25% (1% reduction per day);
[0175] Green light: gradually changes from 10% to 0% (reduced by 2% every day);
[0176] Far-red light: gradually increase from 0% to 10% (increase 2% per day).
[0177] Reasons for adjustment include: increasing light intensity to promote photosynthesis; increasing the proportion of red light to promote stem and leaf growth; increasing far-red light to promote plant height increase; reducing the proportion of blue light to balance growth; removing green light to increase light energy utilization efficiency.
[0178] This case demonstrates how to determine a second lighting scheme using the second model 420 according to the specific growth status of the tomato plant, and adjust the original first lighting scheme accordingly.
[0179] The dynamic light regulation scheme for the mid-stage of tomato growth is shown below.
[0180] The basic spectrum setting of the non-movable lighting device 310 (remains unchanged) is shown in Table 2.
[0181] Table 2: Basic spectral settings of lighting device 310.
[0182] parameter Numeric Red light (600~700nm) 50% Blue light (400~500nm) 30% Green light (500~600nm) 15% Far red light (700~800nm) 5% Light intensity <![CDATA[600μmol / m 2 / s]]> Lighting time 16 hours / day
[0183] The dynamic adjustment process of the fill light device 320 is shown in Table 3.
[0184] Table 3: Dynamic adjustment process of the fill light device 320.
[0185]
[0186] The position adjustment of the fill light device 320 is shown in Table 4.
[0187] Table 4: Position adjustment parameters of the fill light device 320.
[0188]
[0189] The comprehensive light environment (immovable lighting device 310 + supplementary light device 320) is shown in Table 5.
[0190] Table 5: Comprehensive light environment parameters.
[0191] parameter Initial state Target state Red light 50% 51.25% Blu-ray 30% 28.75% Green Light 15% 1 4.25% Far red light 5% 5.75% Total light intensity <![CDATA[Approximately 800 μmol / m 2 / s]]> <![CDATA[Approximately 800 μmol / m 2 / s]]>
[0192] The data collection frequency of each sensor is shown in Table 6.
[0193] Table 6: Data collection frequency of each sensor.
[0194]
[0195] Through this dynamic adjustment scheme, the supplementary light device 320 can adapt to the different stages and needs of tomato growth in a more refined and flexible manner. This gradual adjustment can not only avoid the stress caused by sudden changes in the light environment on the plants, but also continuously optimize based on real-time feedback. Combined with the stable basic light provided by the immovable lighting device 310, this system will provide the most optimized growth environment for tomatoes, while providing managers with detailed data support and flexible control options.
[0196] According to a preferred embodiment, the first model 410 determines the cause of abnormal plant growth state according to the plant growth state of the area unit and the hidden state sequence related to the plant growth state.
[0197] For example, check the state sequence, especially the time points when abnormal states appear continuously. For example, if the "disease and insect pest" state appears for many consecutive days, this may indicate that the plant is suffering from continuous pest and disease infestation. For each abnormal state, calculate the time it lasts in the state sequence. Long-term abnormal states may require more urgent management measures. Set thresholds for different abnormal states and trigger alarms when these thresholds are exceeded. For example, if the "nutrient deficiency" state appears for more than 3 consecutive days, fertilization may be required. For each abnormal state, evaluate its probability in the state sequence. Abnormal states with high probability may need to be handled first.
[0198] The first model 410 determines the relevant hidden state sequence and then determines the cause of the abnormal plant growth state. This analysis method not only improves the accuracy of diagnosing plant growth problems, but also makes the judgment of environmental factors more timely and accurate, which helps to quickly take countermeasures.
[0199] Example 2
[0200] This embodiment is a further improvement on Embodiment 1, and the repeated contents will not be repeated here.
[0201] The present invention provides a plant lighting method for dynamically adjusting light quality based on plant status, such as Figure 3 As shown, the method includes: S1: collecting time-related growth images and spectral characteristics of plants.
[0202] S2: Real-time monitoring of environmental parameters during plant growth.
[0203] S3: Illuminating the plants according to the first lighting scheme.
[0204] S4: Generate plant growth sequence data based on growth images, spectral characteristics and environmental parameters.
[0205] S5: Determine the growth status of the plant in real time based on the growth sequence data and the first model 410 formed by pre-training.
[0206] S6: Determine a second lighting scheme based on the second model 420 including mapping data of plant growth status and lighting scheme.
[0207] S7: Adjusting the spectral characteristics of the first lighting scheme based on the second lighting scheme.
[0208] The plant lighting method collects the growth image and spectral characteristics of the plant in real time through the image acquisition unit 100, ensuring the timely update of the plant growth detection data. Combined with the environmental parameters monitored in real time by the environmental monitoring unit 200, the server 400 can quickly form growth sequence data reflecting the current state, providing a data basis for real-time detection of the plant growth state. In addition, through the comprehensive analysis of the growth image, spectral characteristics and environmental parameters by the server 400, the method can determine the growth state of the plant in real time based on the first model 410, avoiding the response lag problem caused by data collection and processing delays in traditional detection methods.
[0209] According to a preferred embodiment, a method for determining the growth state of a plant in real time based on growth sequence data and a first model 410 formed by pre-training includes: decoding an input growth sequence to obtain a hidden state sequence that can derive the input growth sequence with maximum probability. Classifying the hidden states in the hidden state sequence to obtain the plant growth level corresponding to each point in the growth sequence. Calculating the time series characteristic parameters of the growth sequence, comparing them with the threshold of the time series characteristic parameters, and determining the plant growth state of the growth sequence. Judging the health status of the plant based on the combination of the plant growth level and growth state of the growth sequence.
[0210] This method decodes the growth sequence data in real time to quickly obtain the hidden state sequence of the plant growth level, thus achieving instant evaluation of the plant growth status. The calculation of the time series feature parameters and comparison with their thresholds further ensures the timeliness of the plant growth status detection, providing data support for timely adjustment of the lighting scheme. In addition, by classifying the hidden state sequence and combining it with the plant growth level, this method can timely judge the health status of the plant and reduce plant growth problems caused by delayed status judgment. This fast detection and evaluation process significantly reduces the time delay in the detection process.
[0211] According to a preferred embodiment, the server 400 determines a hidden state sequence related to the plant growth state based on the plant growth state of the regional unit and the first model 410. According to the hidden state sequence, the cause of the abnormal plant growth state is determined. The server 400 uses the plant growth state and the first model 410 to promptly determine the hidden state sequence related to the plant growth state, providing a basis for quickly identifying abnormal plant growth. This real-time analysis method based on regional units ensures a rapid response to the cause of abnormal plant growth state. By analyzing the hidden state sequence, the server 400 can quickly determine the cause of the abnormal plant growth state, avoiding the problem of missing the best adjustment time due to delayed diagnosis. This method improves the response speed of plant growth detection and helps to take necessary measures in a timely manner to improve the plant growth environment.
[0212] It should be noted that the above-mentioned specific embodiments are exemplary, and those skilled in the art can come up with various solutions inspired by the disclosure of the present invention, and these solutions also belong to the disclosure scope of the present invention and fall within the protection scope of the present invention. Those skilled in the art should understand that the present invention specification and its drawings are illustrative and do not constitute a limitation on the claims. The scope of protection of the present invention is defined by the claims and their equivalents. The present invention specification contains multiple inventive concepts, such as "preferably" and "according to a preferred embodiment", which means that the corresponding paragraph discloses an independent concept, and the applicant reserves the right to file a divisional application based on each inventive concept.
Claims
1. A plant lighting system with dynamic light quality adjustment based on plant status, characterized in that: include: An image acquisition unit (100) for acquiring time-related growth images and spectral characteristics of plants; An environmental monitoring unit (200) monitors environmental parameters during plant growth in real time; A lighting unit (300) is configured to illuminate the plants according to a first lighting scheme; The server (400) forms plant growth sequence data according to the growth image, the spectral characteristics and the environmental parameters, determines the growth state of the plant in real time based on the growth sequence data and a first model (410) formed by pre-training, determines a second lighting scheme based on a second model (420) including mapping data of the plant growth state and the lighting scheme, and adjusts the spectral characteristics of the first lighting scheme based on the second lighting scheme.
2. The plant lighting system with dynamic light quality adjustment based on plant status according to claim 1, characterized in that: The method for determining the growth state of a plant in real time based on the growth sequence data and a first model (410) formed by pre-training comprises: Decode the input growth sequence to obtain the hidden state sequence that can derive the input growth sequence with the maximum probability; Classifying the hidden states in the hidden state sequence to obtain the plant growth level corresponding to each point in the growth sequence; Calculating a time series characteristic parameter of the growth sequence, comparing it with a threshold value of the time series characteristic parameter, and determining a plant growth state of the growth sequence; The health status of the plant is determined according to a combination of the plant growth level of the growth sequence and the growth status.
3. The plant lighting system with dynamic light quality adjustment based on plant status according to claim 1 or 2, characterized in that: The method for constructing the first model (410) includes: Preprocess the historical data of plant growth data. Convert the input growth sequence into a signal energy sequence; Inputting the signal energy sequence into a hidden Markov model for training and establishing an energy sequence model, wherein the hidden states in the energy sequence model are classified to correspond to different plant growth levels; calculating and determining the threshold of the time series characteristic parameter related to plant growth state detection; Input the plant growth sequence to be detected and detect the growth status of the plant.
4. The plant lighting system with dynamic light quality adjustment based on plant status according to any one of claims 1 to 3, characterized in that: The method of converting the input plant growth sequence into a signal energy sequence includes: Calculating energy values and average energy of the historical data; Determining the plant growth level based on the average energy within a time window; Among them, the plant growth sequence includes soil environmental parameters, temperature parameters, spectral parameters, normalized vegetation index and leaf characteristic parameters.
5. The plant lighting system with dynamic light quality adjustment based on plant status according to any one of claims 1 to 4, characterized in that: The method of inputting the signal energy sequence into a hidden Markov model for training comprises: Representing the signal energy sequence as an observation probability using a probability mass function of a Poisson distribution; Set initial parameters; Calculate the maximum likelihood estimation function of the observation probability; maximize the maximum likelihood estimation function to obtain the model parameters for the next iteration, and iterate until the model parameters converge or reach a preset number of iterations.
6. The plant lighting system with dynamic light quality adjustment based on plant status according to any one of claims 1 to 5, characterized in that: The lighting scheme includes light intensity parameters, light quality parameters, light cycle parameters and fill light parameters. The server (400) generates fixed lighting parameters of the lighting device (310) and the fill light parameters of the fill light device (320) based on the physical characteristics of the immovable lighting device (310), the physical characteristics of the fill light device (320), and the second lighting scheme, The fill light parameters include lighting coordinates and dynamic lighting parameters.
7. The plant lighting system with dynamic light quality adjustment based on plant status according to any one of claims 1 to 6, characterized in that: The first model (410) is based on the plant growth state of the area unit and the hidden state sequence related to the plant growth state; The first model (410) determines the cause of the abnormal growth state of the plant according to the hidden state sequence.
8. A plant lighting method for dynamic light quality adjustment based on plant status, characterized in that: The method comprises: Collect time-dependent growth images and spectral characteristics of plants; Real-time monitoring of environmental parameters during plant growth, The plants are illuminated according to a first lighting scheme, The growth sequence data of the plant is formed according to the growth image, the spectral characteristics and the environmental parameters, the growth state of the plant is determined in real time based on the growth sequence data and a first model (410) formed by pre-training, a second lighting scheme is determined based on a second model (420) including mapping data of the plant growth state and the lighting scheme, and the spectral characteristics of the first lighting scheme are adjusted based on the second lighting scheme.
9. The plant lighting method for dynamic light quality adjustment based on plant status according to claim 8, characterized in that: The method for determining the growth state of a plant in real time based on the growth sequence data and a first model (410) formed by pre-training comprises: Decode the input growth sequence to obtain the hidden state sequence that can derive the input growth sequence with the maximum probability; Classifying the hidden states in the hidden state sequence to obtain the plant growth level corresponding to each point in the growth sequence; Calculating a time series characteristic parameter of the growth sequence, comparing it with a threshold value of the time series characteristic parameter, and determining a plant growth state of the growth sequence; The health status of the plant is determined according to a combination of the plant growth level of the growth sequence and the growth status.
10. The plant lighting method for dynamic light quality adjustment based on plant status according to claim 8 or 9, characterized in that: The first model (410) is based on the plant growth state of the area unit and the hidden state sequence related to the plant growth state; The first model (410) determines the cause of the abnormal growth state of the plant according to the hidden state sequence.
Citation Information
Cited By
AI plant lamp spectrum adjusting method for plant photosynthesis optimization
CN119946951A
An AI plant light spectrum adjustment method for optimizing plant photosynthesis
CN119946951B
Plant growth whole process total factor intelligent control method and system
CN120491725A
Garden landscape design system and method based on computer vision
CN121038055A
Garden landscape design system and method based on computer vision
CN121038055B
