A fresh flower drying control method and device based on a neural network and a medium
The flower drying method controlled by neural networks automatically selects and dries fresh flowers, solving the problem of poor reliability in traditional manual drying and achieving a highly efficient and accurate flower drying process.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- 浪潮工业互联网股份有限公司
- Filing Date
- 2022-09-06
- Publication Date
- 2026-04-28
AI Technical Summary
Traditional flower drying processes rely on manual monitoring, which leads to poor reliability, high labor costs, and difficulty in guaranteeing the quality of finished products.
A neural network-based fresh flower drying control method is adopted. Fresh flowers to be processed are selected by image feature detection, and the drying process is automatically controlled by pre-generated processing curves and drying parameters. The drying parameters are optimized by training a particle swarm optimization neural network.
Automated drying has been achieved, which has improved the quality and efficiency of finished flower products, reduced labor costs, and ensured the accuracy and precision of drying parameters.
Smart Images
Figure CN115526839B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of fresh flower processing technology, specifically to a method, equipment, and medium for controlling the drying of fresh flowers based on neural networks. Background Technology
[0002] Moisture is the medium for a series of chemical reactions within the leaves during flower processing, and it is the most important influencing factor in flower processing. Drying, as the final step in flower processing, plays a crucial role in the quality of the processed flowers. Traditional flower drying processes rely heavily on manual labor, which is unreliable and incurs high labor costs. Summary of the Invention
[0003] To address the aforementioned issues, this application proposes a neural network-based method for controlling the drying of fresh flowers, comprising: acquiring a fresh flower image and performing feature detection on the fresh flower image to select fresh flowers to be processed from the fresh flower image;
[0004] Based on the intended use of the flowers to be processed, the estimated moisture content of the flowers after drying is determined, and multiple processing stages corresponding to the flowers to be processed are determined by the initial moisture content and the estimated moisture content of the flowers to be processed.
[0005] The target moisture content corresponding to each of the multiple processing stages is determined. Based on the initial moisture content, water supply, and target moisture content of the flowers to be processed, the drying parameters required for the drying process of the flowers to be processed are determined through a pre-generated processing curve. The processing curve is obtained by fitting the input and output parameters of a pre-constructed neural network.
[0006] For the multiple processing stages, the fresh flowers to be processed are dried by controlling the drying parameters in the drying workshop to obtain dried flowers.
[0007] In one implementation of this application, feature detection is performed on the flower image to select flowers to be processed from the flower image, specifically including:
[0008] The flower image is semantically segmented using a pre-trained feature detection model to obtain a segmented feature image; the feature image includes a first region and a second region, and the first region and the second region contain different color features.
[0009] Extract the edge contour of the first region and perform regional growth analysis on the edge contour to determine whether the first region carries lesion features;
[0010] If the first region carries lesion features, determine the proportion of the first region in the feature image, and select flowers from the flower image whose proportion is less than a preset threshold as flowers to be processed.
[0011] In one implementation of this application, semantic segmentation is performed on the flower image to obtain a segmented feature image, specifically including:
[0012] The multiple color channels corresponding to the flower image are respectively feature-encoded to obtain the channel feature maps corresponding to the multiple color channels;
[0013] For the multiple color channels, the channel feature maps are processed to obtain processed channel feature maps; the processing includes multi-scale feature aggregation and spatial correlation enhancement processing;
[0014] The processed channel feature maps are fused to obtain the feature image corresponding to the flower image.
[0015] In one implementation of this application, the flowers to be processed are dried by controlling the drying parameters in the drying workshop, specifically including:
[0016] Determine the processing stage of the fresh flowers to be processed;
[0017] The operating parameters of the drying workshop are collected, and the operating parameters are compared with the drying parameters corresponding to the processing stage to determine whether the drying workshop meets the preset operating requirements.
[0018] If the preset operating requirements are not met, the required compensation amount for the operating parameters is determined based on the difference between the corresponding values of each operating parameter and the corresponding values of each drying parameter. Based on the compensation amount, the corresponding auxiliary equipment is called to perform collaborative drying for the processing stage.
[0019] In one implementation of this application, before determining the required drying parameters for the fresh flowers to be processed during the drying process using a pre-generated processing curve, the method further includes:
[0020] Construct the topology of the neural network;
[0021] Initialize the particle parameters of each particle to obtain an initial population after initialization, and initialize the parameters of the neural network according to each particle in the initial population; the parameters include the weights and thresholds between different layers of the topology.
[0022] The state parameters of the fresh flowers are collected to construct a training set based on the state parameters; the training set includes initial moisture content, feeding amount, temperature, humidity, and drying time.
[0023] The neural network is trained based on the training set, and the target function output value of the trained neural network is determined.
[0024] Based on the fitness of each particle, the parameters of the neural network are optimized to obtain the trained neural network; the fitness corresponds to the output value of the objective function of the neural network.
[0025] The processing curve of the fresh flowers is obtained by fitting the input and output parameters of the neural network; the output parameters include the target moisture content.
[0026] In one implementation of this application, the parameters of the neural network are optimized based on the fitness of each particle to obtain the trained neural network, specifically including:
[0027] Based on the fitness of each particle, the global optimum of the corresponding population is obtained;
[0028] Determine whether the global optimal value meets the preset requirements; if not, update the velocity and position of each particle.
[0029] Based on the updated speed and position, the parameters of the neural network are optimized until the global optimum meets the preset requirements.
[0030] In one implementation of this application, after selecting the flowers to be processed from the flower image, the method further includes:
[0031] Obtain the main image of the flower to be processed, extract the feature points of the main image, and perform three-dimensional reconstruction of the flower to be processed based on the feature points to obtain the corresponding three-dimensional model;
[0032] Based on the three-dimensional model, the main shape of the flower to be processed is obtained, and the first-level parameters corresponding to the main shape are determined;
[0033] The color information of the fresh flowers to be processed is collected using a color measuring instrument.
[0034] The chromaticity information is compared with a preset chromaticity level to determine the second-level parameters of the flowers to be processed;
[0035] The grade corresponding to the fresh flowers to be processed is obtained based on the first grade parameter and the second grade parameter;
[0036] After drying the fresh flowers to be processed to obtain dried flowers, the method further includes:
[0037] The dried flowers are sorted according to their grade and then transported to different types of finished product warehouses for storage; different finished product warehouses correspond to different usage categories.
[0038] In one implementation of this application, the plurality of processing stages are determined based on the temperature in the drying parameters.
[0039] This application provides a neural network-based fresh flower drying control device, characterized in that it includes:
[0040] At least one processor; and,
[0041] A memory communicatively connected to the at least one processor; wherein,
[0042] The memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enable the at least one processor to perform actions such as:
[0043] Acquire flower images, perform feature detection on the flower images, and select flowers to be processed from the flower images;
[0044] Based on the intended use of the flowers to be processed, the estimated moisture content of the flowers after drying is determined, and multiple processing stages corresponding to the flowers to be processed are determined by the initial moisture content and the estimated moisture content of the flowers to be processed.
[0045] The target moisture content corresponding to each of the multiple processing stages is determined. Based on the initial moisture content, water supply, and target moisture content of the flowers to be processed, the drying parameters required for the drying process of the flowers to be processed are determined through a pre-generated processing curve. The processing curve is obtained by fitting the input and output parameters of a pre-constructed neural network.
[0046] For the multiple processing stages, the fresh flowers to be processed are dried by controlling the drying parameters in the drying workshop to obtain dried flowers.
[0047] This application provides a non-volatile computer storage medium storing computer-executable instructions, characterized in that the computer-executable instructions are configured as follows:
[0048] Acquire flower images, perform feature detection on the flower images, and select flowers to be processed from the flower images;
[0049] Based on the intended use of the flowers to be processed, the estimated moisture content of the flowers after drying is determined, and multiple processing stages corresponding to the flowers to be processed are determined by the initial moisture content and the estimated moisture content of the flowers to be processed.
[0050] The target moisture content corresponding to each of the multiple processing stages is determined. Based on the initial moisture content, water supply, and target moisture content of the flowers to be processed, the drying parameters required for the drying process of the flowers to be processed are determined through a pre-generated processing curve. The processing curve is obtained by fitting the input and output parameters of a pre-constructed neural network.
[0051] For the multiple processing stages, the fresh flowers to be processed are dried by controlling the drying parameters in the drying workshop to obtain dried flowers.
[0052] The fresh flower drying control method based on neural networks proposed in this application can bring the following beneficial effects:
[0053] Feature detection of flower images ensures that the selected flowers for processing do not have disease characteristics, effectively improving the quality of the finished product. By using a pre-generated processing curve, the drying parameters required for the current flower drying process can be directly obtained given the initial moisture content and water intake of the flowers. This eliminates the need for manual observation and adjustment of drying parameters, achieving an automated drying process that effectively improves drying efficiency and accuracy. Attached Figure Description
[0054] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments of this application and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:
[0055] Figure 1 A flowchart illustrating a neural network-based fresh flower drying control method provided in this application embodiment;
[0056] Figure 2 A schematic diagram of a neural network topology provided in an embodiment of this application;
[0057] Figure 3 A flowchart illustrating the training process of a neural network, as provided in an embodiment of this application;
[0058] Figure 4 This is a schematic diagram of a flower drying control device based on a neural network, provided as an embodiment of this application. Detailed Implementation
[0059] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0060] The technical solutions provided by the various embodiments of this application are described in detail below with reference to the accompanying drawings.
[0061] like Figure 1 As shown in the embodiments of this application, the method for controlling the drying of fresh flowers includes:
[0062] 101: Acquire flower images, perform feature detection on the flower images, and select the flowers to be processed from the flower images.
[0063] Before drying fresh flowers, they need to be inspected for characteristics to determine whether their quality meets the standards. This allows for the removal of flowers with diseased characteristics, ensuring the quality of the finished product.
[0064] Specifically, a pre-trained feature detection model is used to perform semantic segmentation on the flower image, resulting in segmented feature images. The unsegmented flower image is a three-channel image. Feature encoding is performed on each of the multiple color channels corresponding to the flower image, resulting in channel feature maps for each color channel, with each channel feature map representing different contextual information. After obtaining the channel feature maps, multi-scale feature aggregation and spatial correlation enhancement processing are performed on the corresponding channel feature maps for the three color channels, resulting in processed channel feature maps. These processed channel feature maps are then fused to obtain the final feature image corresponding to the flower image.
[0065] It is understandable that the feature image after semantic segmentation is a binary image with black and white as the base colors. Based on this, the feature image is divided into a first region and a second region with different corresponding color features. The first region represents the white region, which may carry certain lesion features, and the second region represents the black region, which is the normal region.
[0066] Furthermore, after obtaining the feature image of the flower, feature judgment is performed on the first region to determine whether it truly carries disease characteristics. The edge contour of the first region is extracted, and regional growth analysis is performed on the edge contour, that is, the edge contour of the first region at different time periods is obtained to determine whether the edge contour has an expanding trend over time. If so, it indicates that the first region carries disease characteristics, such as lesions, rot, etc.
[0067] Furthermore, if the first region contains disease characteristics, the proportion of the first region in the feature image can be used to determine whether the flower can be considered a qualified product for further processing. If the proportion is less than a preset threshold, it means that although the flower to be dried has a certain degree of disease spots, it will not affect the overall quality after processing. It can also be circulated in the market by classifying its quality. Therefore, for flowers with disease characteristics less than the preset proportion, the next drying process can still be carried out.
[0068] 102: Based on the intended use of the fresh flowers to be processed, determine the estimated moisture content of the fresh flowers after drying, and determine the multiple processing stages corresponding to the fresh flowers to be processed by using the initial moisture content and estimated moisture content of the fresh flowers to be processed.
[0069] After the fresh flowers are selected for processing, they need to be dried. The required moisture content varies depending on the type of flower and its intended use. For example, dried flowers can be used for decoration or have practical value (such as medicinal use or as a tea ingredient). Generally, the required moisture content for decorative dried flowers can be slightly higher than that for practical dried flowers. This is because if the moisture content of decorative dried flowers is too low, they are more likely to be damaged during subsequent distribution, which would significantly affect their value.
[0070] Therefore, for flowers of different uses, their estimated moisture content after drying needs to be determined separately. Based on the initial and estimated moisture content of the flowers to be processed, multiple processing stages can be determined. It should be noted that the processing stages are determined by the temperature of the drying workshop. For example, if the initial moisture content is 50% and the estimated moisture content is 20%, the drying process can be roughly divided into: gradually increasing the temperature to cause significant water loss in the flowers; maintaining a constant temperature after reaching a certain moisture content; and slightly lowering the temperature just before reaching the estimated moisture content to ensure the flowers do not lose too much water. Based on this temperature change process, the flower processing can be divided into multiple stages. The specific stage division can be determined according to actual needs, and this application does not limit it.
[0071] 103: Determine the target moisture content corresponding to each of the multiple processing stages. Based on the initial moisture content, water supply, and target moisture content of the flowers to be processed, determine the drying parameters required for the drying process of the flowers to be processed through a pre-generated processing curve. The processing curve is obtained by fitting the input and output parameters of a pre-constructed neural network.
[0072] To achieve precise drying of fresh flowers, the drying process can be divided into multiple processing stages, thus achieving step-by-step drying. The target moisture content achieved at each processing stage differs. This application addresses the change in moisture content of fresh flowers during the drying process by employing a neural network algorithm. Using drying temperature, humidity, drying time, and initial moisture content as input parameters, a precise moisture content prediction model is established. This model establishes a functional relationship between initial feed rate, initial moisture content, drying temperature, drying humidity, drying time, and target moisture content. Furthermore, batches of fresh flowers with excellent drying results from actual production are recorded, and their moisture content change curves are plotted as a standard reference. By combining the standard curve with the functional relationship between the input parameters, the appropriate temperature and humidity required during the drying process are derived, generating the corresponding processing curves.
[0073] Specifically, the topology of the neural network is constructed; this neural network is a feedback neural network model. The above neural network adds a support layer to the traditional BP neural network, giving the network local memory. For example... Figure 2 The diagram shows a topological structure of a neural network, consisting of an input layer, hidden layers, a receiving layer, and an output layer. The number of neurons in the receiving layer and hidden layers is the same, and the number of neurons in the input layer is the same as the input parameters, which is 5. Since this model is based on solving regression problems, the number of neurons in the output layer is 1. The number of neurons in the hidden layer can be selected according to the actual performance of the model (it is possible to try selecting within the range of 1 to 10 during training). The setting of the number of hidden layers and nodes has a great impact on the performance of the network. Too many will increase the complexity and computational load of the network, and may even lead to overfitting, while too few will result in poor performance. Generally, it is initially set to one hidden layer, and the specific number of layers can be adjusted according to the model training results. This application does not limit this.
[0074] Neurons in each layer are interconnected, transmitting information through different weights and thresholds. The weight between the input layer and the hidden layer is w. ij The weight between the receiving layer and the hidden layer is w jq The threshold is b j The output value of each node in the hidden layer is calculated as follows:
[0075]
[0076] h' j =h j (t-1) (2)
[0077] Where i = 1, 2, 3, ..., n, j = 1, 2, 3, ..., l, q = 1, 2, 3, ..., l, h j Let f(.) be the hidden layer output value, f(.) be the hidden layer activation function, and x be the hidden layer output value.i For input, h' j It is the hidden layer output value of the previous time step, which is fed back by the receiving layer and used together with the input layer as the current output. t is the number of learning steps.
[0078] The output value of each node in the output layer is calculated as follows:
[0079]
[0080] Where j = 1, 2, 3, ..., l, k = 1, 2, 3, ..., m, g(.) is the output layer activation function, b k Let f(.) be the threshold for the k-th node. It should be noted that f(.) and g(.) are not necessarily the same.
[0081] Furthermore, the neural network constructed in this application can store and utilize output information from past moments, realizing the mapping of dynamic systems and directly reflecting the dynamic characteristics of the system. Therefore, it is superior to the BP neural network in terms of network stability and computational power. However, its weight and threshold update method is the same as that of the BP neural network. During the training process of the neural network, it is easy to get trapped in local minima and it is difficult to reach the global optimum. In order to solve this problem and increase the global optimization ability of the network, the parameters of the neural network need to be continuously optimized to avoid the network getting trapped in local minima. That is to say, the global optimum is found through the sharing of group information. In the group activity, each individual benefits from the experience discovered and accumulated by all individuals in the optimization process, thus solving the problem of local convergence.
[0082] After constructing the neural network, the particle parameters of each particle are initialized to obtain the initial population. After the population initialization is completed, the parameters of the neural network are initialized based on each particle in the initial population. These parameters include the weights and thresholds between different levels of the topology.
[0083] It should be noted that the expression for particle swarm optimization is as follows:
[0084] v i =ω·v i +c i ·rand·(pbest i -x i )+c2·rand(gbest i -x i (4)
[0085] x i =x' i +v i (5)
[0086] Among them, v i It is the particle velocity, xi This is the particle's position, x' i It is the previous particle position, ω is the inertia factor (a random number between 0 and 1), and c is the position of the particle. i c2 is the learning factor, which is usually taken as a normal value of 2. (pbest) i For the individual's historical best value, gbest i This is the global historical optimum. The particle parameters that need to be initialized include v. i x i , ω, c i c2 and the number of iterations.
[0087] Furthermore, to adapt to the topology of the neural network, the state parameters of the flowers are collected, thereby constructing a training set based on the state parameters. Table 1 shows some sample data provided in the embodiments of this application:
[0088] Table 1
[0089]
[0090] As shown in Table 1, the training set includes initial moisture content, feed amount, temperature, humidity, and drying time.
[0091] Furthermore, after obtaining the training set, the initialized neural network is trained using the training set, and after completing the current iteration, the output value of the neural network's objective function is calculated. The objective function in this application uses mean squared error, and the expression for the error function is as follows:
[0092]
[0093] Where, d k For the actual value, y k Let E be the predicted value, and E be the error function, where k = 1, 2, 3, ..., m.
[0094] The neural network uses gradient descent to achieve backpropagation of information to update weights and thresholds, thus continuously updating the weights and thresholds. The calculation formula for the update process is as follows:
[0095]
[0096]
[0097] Here, η is the learning rate, which needs to be selected appropriately based on the actual situation during the training process.
[0098] The formulas for updating weights and thresholds between the input layer and the receiving layer and the hidden layer are as follows:
[0099]
[0100]
[0101]
[0102] Furthermore, if the calculated output value of the objective function does not meet the preset requirements, the parameters of the neural network need to be optimized based on the fitness of each particle. Then, the processing curve of the flower can be fitted using the input and output of the optimized neural network. It can be understood that the fitness corresponds to the output value of the neural network's objective function, the input parameters of the neural network are of the same type as the parameters contained in the training set, and the output is the target water content.
[0103] In one embodiment, the parameters of the neural network are optimized through the following steps: First, the fitness of each particle is calculated to obtain the global optimum of the population based on the best individual. This fitness is the value of the neural network's objective function. Then, it is determined whether the global optimum meets the preset requirements. If not, it indicates that the current training result of the neural network is not optimal, and the velocity and position of each particle need to be updated. Next, the parameters of the neural network are adjusted using the updated particle parameters, and the global optimum of the optimized particle population is recalculated until the global optimum meets the preset requirements. This completes the training process of the neural network, resulting in the best prediction performance. By establishing the functional relationship between its input and output parameters and fitting and generating corresponding processing curves, the required drying parameters can be obtained during the actual flower drying process, based on the initial moisture content, feed amount, and target moisture content at each processing stage. These drying parameters include temperature, humidity, and drying time, effectively improving drying efficiency and accuracy.
[0104] Figure 3 A flowchart of a neural network training process is provided for an embodiment of this application, such as... Figure 3As shown, the topology of the neural network is determined, and then each particle is randomly initialized to obtain the initial population. After particle initialization, the parameters of the neural network are initialized according to the particle parameters. After the initialization process of the neural network is completed, it is trained using the training set, and the training effect of the neural network is tested using the test set after each round of training iteration to obtain the test error. Since the fitness function of the particles corresponds to the objective function, the fitness of each particle can be determined according to the error function after obtaining the test error, so as to obtain the global optimum of the current population based on the determined optimal individual. It is determined whether the global optimum meets the preset requirements. If it does, the current iteration training process ends; if it does not, the historical optimum of each particle and the global optimum of its population are updated, and the velocity and position of each particle are updated, the weights and thresholds are output, and then the parameters of the neural network are optimized according to the updated weights and thresholds until the global optimum of the particle swarm can meet the preset requirements. At this point, the training process of the neural network is completed.
[0105] 104: For multiple processing stages, the fresh flowers to be processed are dried by controlling the drying parameters in the drying workshop to obtain dried flowers.
[0106] After obtaining the drying parameters corresponding to each processing stage through step 103, the drying workshop can be controlled to adjust the drying parameters in real time to finally obtain dried flowers with the expected moisture content.
[0107] In one embodiment, during the drying process, the operating parameters of the drying workshop need to be monitored in real time. This is because changes in the external environment during drying, such as the actual operating temperature of the drying equipment failing to reach the preset drying temperature, may affect the drying conditions of the current drying workshop. Alternatively, the drying conditions of the drying workshop may also be affected by the flowers being processed. Therefore, to improve processing accuracy, it is necessary to determine the processing stage of the flowers to be processed in real time, collect the operating parameters of the drying workshop, and compare the operating parameters with the drying parameters corresponding to the current processing stage to determine whether the drying workshop meets the preset operating requirements. If it does not meet the preset operating requirements, then the required compensation amount for the operating parameters is determined based on the difference between the corresponding values of each operating parameter and the corresponding values of each drying parameter. Based on this compensation amount, corresponding auxiliary equipment is called in to assist the drying process. For example, if the actual drying temperature does not reach the preset temperature, air conditioning can be turned on for auxiliary drying.
[0108] After the flowers have been dried, they need to be further sorted according to their grade and then transported to the finished product warehouse for their respective uses for storage.
[0109] Specifically, after selecting the fresh flowers to be processed, a main image of the flowers is obtained, and feature points are extracted from the main image. Based on these feature points, a 3D reconstruction of the flowers is performed to obtain a corresponding 3D model. The main image refers to an image that shows the overall shape of the flowers and is used to evaluate their appearance grade. After obtaining the 3D model, the main shape of the flowers is obtained based on the 3D model, and the first-level parameters corresponding to the main shape are determined. Simultaneously, chromaticity information of the flowers is collected using a color measuring instrument, and then compared with a preset chromaticity grade to determine the second-level parameters. After obtaining the first and second-level parameters, the finished product is graded by comprehensively considering the flower's shape and chromaticity. A weighted sum of the first and second-level parameters is required to obtain the final grade of the flowers. After drying the flowers, they can be packaged according to the predetermined flower grade, facilitating subsequent distribution processes.
[0110] The above are embodiments of the method proposed in this application. Based on the same inventive concept, embodiments of this application also provide a flower drying device based on a neural network, the structure of which is as follows: Figure 4 As shown.
[0111] Figure 4 This is a schematic diagram of a neural network-based flower drying device provided in an embodiment of this application. Figure 4 As shown, the device includes:
[0112] At least one processor 401;
[0113] And a memory 402 that is communicatively connected to at least one processor;
[0114] The memory 402 stores instructions executable by at least one processor, which are executed by at least one processor 401 to enable at least one processor 401 to:
[0115] Acquire flower images, perform feature detection on the flower images, and select the flowers to be processed from the flower images;
[0116] Based on the intended use of the fresh flowers to be processed, determine the estimated moisture content of the fresh flowers after drying, and determine the multiple processing stages corresponding to the fresh flowers to be processed by using the initial moisture content and estimated moisture content of the fresh flowers to be processed.
[0117] The target moisture content corresponding to each of the multiple processing stages is determined. Based on the initial moisture content, water supply, and target moisture content of the fresh flowers to be processed, the drying parameters required for the drying process of the fresh flowers to be processed are determined through a pre-generated processing curve. The processing curve is obtained by fitting the input and output parameters of a pre-constructed neural network.
[0118] For multiple processing stages, the drying parameters in the drying workshop are controlled to dry the fresh flowers to obtain dried flowers.
[0119] This application embodiment also provides a non-volatile computer storage medium storing computer-executable instructions, wherein the computer-executable instructions are configured as follows:
[0120] Acquire flower images, perform feature detection on the flower images, and select the flowers to be processed from the flower images;
[0121] Based on the intended use of the fresh flowers to be processed, determine the estimated moisture content of the fresh flowers after drying, and determine the multiple processing stages corresponding to the fresh flowers to be processed by using the initial moisture content and estimated moisture content of the fresh flowers to be processed.
[0122] The target moisture content corresponding to each of the multiple processing stages is determined. Based on the initial moisture content, water supply, and target moisture content of the fresh flowers to be processed, the drying parameters required for the drying process of the fresh flowers to be processed are determined through a pre-generated processing curve. The processing curve is obtained by fitting the input and output parameters of a pre-constructed neural network.
[0123] For multiple processing stages, the drying parameters in the drying workshop are controlled to dry the fresh flowers to obtain dried flowers.
[0124] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the device and medium embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the description of the method embodiments.
[0125] The devices and media provided in this application are one-to-one with the methods. Therefore, the devices and media also have similar beneficial technical effects as their corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the devices and media will not be repeated here.
[0126] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0127] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0128] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0129] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0130] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.
[0131] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.
[0132] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.
[0133] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0134] The above description is merely an embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of this application should be included within the scope of the claims of this application.
Claims
1. A method for controlling the drying of fresh flowers based on neural networks, characterized in that, The method includes: Acquire flower images, perform feature detection on the flower images, and select flowers to be processed from the flower images; Based on the intended use of the flowers to be processed, the estimated moisture content of the flowers after drying is determined, and multiple processing stages corresponding to the flowers to be processed are determined by the initial moisture content and the estimated moisture content of the flowers to be processed. The target moisture content corresponding to each of the multiple processing stages is determined. Based on the initial moisture content, water supply, and target moisture content of the flowers to be processed, the drying parameters required for the drying process of the flowers to be processed are determined through a pre-generated processing curve. The processing curve is obtained by fitting the input and output parameters of a pre-constructed neural network. For the multiple processing stages, the fresh flowers to be processed are dried by controlling the drying parameters in the drying workshop to obtain dried flowers; Perform feature detection on the flower image to select flowers to be processed from the flower image, specifically including: The flower image is semantically segmented using a pre-trained feature detection model to obtain a segmented feature image; the feature image includes a first region and a second region, and the first region and the second region contain different color features. Extract the edge contour of the first region and perform regional growth analysis on the edge contour to determine whether the first region carries lesion features; If the first region carries lesion features, determine the proportion of the first region in the feature image, and select flowers with the proportion less than a preset threshold from the flower image as flowers to be processed. Before determining the required drying parameters for the fresh flowers during the drying process using a pre-generated processing curve, the method further includes: Construct the topology of the neural network; Initialize the particle parameters of each particle to obtain an initial population after initialization, and initialize the parameters of the neural network according to each particle in the initial population; the parameters include the weights and thresholds between different layers of the topology. The state parameters of the fresh flowers are collected to construct a training set based on the state parameters; the training set includes initial moisture content, feeding amount, temperature, humidity, and drying time. The neural network is trained based on the training set, and the target function output value of the trained neural network is determined. Based on the fitness of each particle, the parameters of the neural network are optimized to obtain the trained neural network; the fitness corresponds to the output value of the objective function of the neural network. The processing curve of the fresh flowers is obtained by fitting the input and output parameters of the neural network; the output parameters include the target moisture content. After selecting the flowers to be processed from the flower images, the method further includes: Obtain the main image of the flower to be processed, extract the feature points of the main image, and perform three-dimensional reconstruction of the flower to be processed based on the feature points to obtain the corresponding three-dimensional model; Based on the three-dimensional model, the main shape of the flower to be processed is obtained, and the first-level parameters corresponding to the main shape are determined; The color information of the fresh flowers to be processed is collected using a color measuring instrument. The chromaticity information is compared with a preset chromaticity level to determine the second-level parameters of the flowers to be processed; The grade corresponding to the fresh flowers to be processed is obtained based on the first grade parameter and the second grade parameter; After drying the fresh flowers to be processed to obtain dried flowers, the method further includes: The dried flowers are sorted according to their grade and then transported to different types of finished product warehouses for storage; different finished product warehouses correspond to different uses.
2. The method for controlling flower drying based on a neural network according to claim 1, characterized in that, Semantic segmentation is performed on the flower image to obtain a segmented feature image, specifically including: The multiple color channels corresponding to the flower image are respectively feature-encoded to obtain the channel feature maps corresponding to the multiple color channels; For the multiple color channels, the channel feature maps are processed to obtain processed channel feature maps; the processing includes multi-scale feature aggregation and spatial correlation enhancement processing; The processed channel feature maps are fused to obtain the feature image corresponding to the flower image.
3. The method for controlling the drying of fresh flowers based on a neural network according to claim 1, characterized in that, The flowers to be processed are dried by controlling the drying parameters in the drying workshop, specifically including: Determine the processing stage of the fresh flowers to be processed; The operating parameters of the drying workshop are collected, and the operating parameters are compared with the drying parameters corresponding to the processing stage to determine whether the drying workshop meets the preset operating requirements. If the preset operating requirements are not met, the required compensation amount for the operating parameters is determined based on the difference between the corresponding values of each operating parameter and the corresponding values of each drying parameter. Based on the compensation amount, the corresponding auxiliary equipment is called to perform collaborative drying for the processing stage.
4. The method for controlling the drying of fresh flowers based on a neural network according to claim 1, characterized in that, Based on the fitness of each particle, the parameters of the neural network are optimized to obtain the trained neural network, specifically including: Based on the fitness of each particle, the global optimum of the corresponding population is obtained; Determine whether the global optimal value meets the preset requirements; if not, update the velocity and position of each particle. Based on the updated speed and position, the parameters of the neural network are optimized until the global optimum meets the preset requirements.
5. The method for controlling flower drying based on a neural network according to claim 1, characterized in that, The multiple processing stages are determined based on the temperature in the drying parameters.
6. A fresh flower drying control device based on neural networks, characterized in that, include: At least one processor; as well as, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enable the at least one processor to perform actions such as: Acquire flower images, perform feature detection on the flower images, and select flowers to be processed from the flower images; Based on the intended use of the flowers to be processed, the estimated moisture content of the flowers after drying is determined, and multiple processing stages corresponding to the flowers to be processed are determined by the initial moisture content and the estimated moisture content of the flowers to be processed. The target moisture content corresponding to each of the multiple processing stages is determined. Based on the initial moisture content, water supply, and target moisture content of the flowers to be processed, the drying parameters required for the drying process of the flowers to be processed are determined through a pre-generated processing curve. The processing curve is obtained by fitting the input and output parameters of a pre-constructed neural network. For the multiple processing stages, the fresh flowers to be processed are dried by controlling the drying parameters in the drying workshop to obtain dried flowers; Perform feature detection on the flower image to select flowers to be processed from the flower image, specifically including: The flower image is semantically segmented using a pre-trained feature detection model to obtain a segmented feature image; the feature image includes a first region and a second region, and the first region and the second region contain different color features. Extract the edge contour of the first region and perform regional growth analysis on the edge contour to determine whether the first region carries lesion features; If the first region carries lesion features, determine the proportion of the first region in the feature image, and select flowers with the proportion less than a preset threshold from the flower image as flowers to be processed. Before determining the required drying parameters for the fresh flowers to be processed during the drying process using a pre-generated processing curve, the process also includes: Construct the topology of the neural network; Initialize the particle parameters of each particle to obtain an initial population after initialization, and initialize the parameters of the neural network according to each particle in the initial population; the parameters include the weights and thresholds between different layers of the topology. The state parameters of the fresh flowers are collected to construct a training set based on the state parameters; the training set includes initial moisture content, feeding amount, temperature, humidity, and drying time. The neural network is trained based on the training set, and the target function output value of the trained neural network is determined. Based on the fitness of each particle, the parameters of the neural network are optimized to obtain the trained neural network; the fitness corresponds to the output value of the objective function of the neural network. The processing curve of the fresh flowers is obtained by fitting the input and output parameters of the neural network; the output parameters include the target moisture content. After selecting the flowers to be processed from the flower images, the process also includes: Obtain the main image of the flower to be processed, extract the feature points of the main image, and perform three-dimensional reconstruction of the flower to be processed based on the feature points to obtain the corresponding three-dimensional model; Based on the three-dimensional model, the main shape of the flower to be processed is obtained, and the first-level parameters corresponding to the main shape are determined; The color information of the fresh flowers to be processed is collected using a color measuring instrument. The chromaticity information is compared with a preset chromaticity level to determine the second-level parameters of the flowers to be processed; The grade corresponding to the fresh flowers to be processed is obtained based on the first grade parameter and the second grade parameter; After drying the fresh flowers to be processed to obtain dried flowers, the process further includes: The dried flowers are sorted according to their grade and then transported to different types of finished product warehouses for storage; different finished product warehouses correspond to different uses.
7. A non-volatile computer storage medium storing computer-executable instructions, characterized in that, The computer-executable instructions are set as follows: Acquire flower images, perform feature detection on the flower images, and select flowers to be processed from the flower images; Based on the intended use of the flowers to be processed, the estimated moisture content of the flowers after drying is determined, and multiple processing stages corresponding to the flowers to be processed are determined by the initial moisture content and the estimated moisture content of the flowers to be processed. The target moisture content corresponding to each of the multiple processing stages is determined. Based on the initial moisture content, water supply, and target moisture content of the flowers to be processed, the drying parameters required for the drying process of the flowers to be processed are determined through a pre-generated processing curve. The processing curve is obtained by fitting the input and output parameters of a pre-constructed neural network. For the multiple processing stages, the fresh flowers to be processed are dried by controlling the drying parameters in the drying workshop to obtain dried flowers; Perform feature detection on the flower image to select flowers to be processed from the flower image, specifically including: The flower image is semantically segmented using a pre-trained feature detection model to obtain a segmented feature image; the feature image includes a first region and a second region, and the first region and the second region contain different color features. Extract the edge contour of the first region and perform regional growth analysis on the edge contour to determine whether the first region carries lesion features; If the first region carries lesion features, determine the proportion of the first region in the feature image, and select flowers with the proportion less than a preset threshold from the flower image as flowers to be processed. Before determining the required drying parameters for the fresh flowers to be processed during the drying process using a pre-generated processing curve, the process also includes: Construct the topology of the neural network; Initialize the particle parameters of each particle to obtain an initial population after initialization, and initialize the parameters of the neural network according to each particle in the initial population; the parameters include the weights and thresholds between different layers of the topology. The state parameters of the fresh flowers are collected to construct a training set based on the state parameters; the training set includes initial moisture content, feeding amount, temperature, humidity, and drying time. The neural network is trained based on the training set, and the target function output value of the trained neural network is determined. Based on the fitness of each particle, the parameters of the neural network are optimized to obtain the trained neural network; the fitness corresponds to the output value of the objective function of the neural network. The processing curve of the fresh flowers is obtained by fitting the input and output parameters of the neural network; the output parameters include the target moisture content. After selecting the flowers to be processed from the flower images, the process also includes: Obtain the main image of the flower to be processed, extract the feature points of the main image, and perform three-dimensional reconstruction of the flower to be processed based on the feature points to obtain the corresponding three-dimensional model; Based on the three-dimensional model, the main shape of the flower to be processed is obtained, and the first-level parameters corresponding to the main shape are determined; The color information of the fresh flowers to be processed is collected using a color measuring instrument. The chromaticity information is compared with a preset chromaticity level to determine the second-level parameters of the flowers to be processed; The grade corresponding to the fresh flowers to be processed is obtained based on the first grade parameter and the second grade parameter; After drying the fresh flowers to be processed to obtain dried flowers, the process further includes: The dried flowers are sorted according to their grade and then transported to different types of finished product warehouses for storage; different finished product warehouses correspond to different uses.
Citation Information
Patent Citations
Grain drying method and system
CN112179125A
Automatic drying control management system
CN114705034A
Method and system for recognizing plant diseases and recording medium
US20130136312A1