A load prediction method and system based on hybrid rice algorithm optimization

Abstract

The embodiment of the application provides a load prediction method and system based on hybrid rice algorithm optimization, the method comprises the following steps: acquiring air conditioner running parameters and environment parameters of a hybrid rice cultivation environment; determining important features through a preset algorithm evaluation; constructing a BP neural network model for load prediction of the cultivation environment with the number of important features as the dimension; acquiring the gene sequence of the hybrid rice and initializing, adjusting the initial weight of the BP neural network model with the gene sequence, and training the input model with the important features; taking the load prediction accuracy of the cultivation environment in the model weight training as the fitness value, grouping the hybrid rice by type, updating and optimizing the hybrid rice, training the initial weight of the BP neural network model with the updated and optimized hybrid rice; when the training stops, outputting the optimal weight parameter configuration of the BP neural network, inputting the important features into the BP neural network model with the optimal weight parameter configuration, and determining an air conditioner running scheme.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of automation control technology, and in particular to a load prediction method and system based on hybrid rice algorithm optimization. BACKGROUND

[0002] With the progress of science and technology, other scientific and technological equipment also needs to be involved in the process of cultivating hybrid rice, such as seed preservation, breeding, greenhouse cultivation, and seed production. At each stage, air conditioning is needed to participate in the process, which can control the load of the environment of the corresponding cultivation environment of hybrid rice, and help to improve the yield, quality and reliability of scientific research results of hybrid rice.

[0003] However, in the traditional hybrid rice cultivation process, the air conditioner running start-stop time designed by artificial has a serious problem of energy waste, because the start-stop time designed by artificial often cannot accurately match the actual environmental demand, resulting in the air conditioner still running when it is not needed, or starting too early when it is needed, causing energy waste. SUMMARY

[0004] In view of the problems in the prior art, the present application provides a load prediction method and system based on hybrid rice algorithm optimization.

[0005] The present application provides a load prediction method based on hybrid rice algorithm optimization, comprising:

[0006] Obtain the associated parameters of the hybrid rice cultivation environment, and preprocess the associated parameters, wherein the associated parameters include air conditioner running parameters and environmental parameters;

[0007] Evaluate the importance of the associated parameters by a preset algorithm, determine the important features in the associated parameters, and construct a BP neural network model for load prediction of the cultivation environment based on the number of important features as the dimension;

[0008] Obtain the gene sequence of hybrid rice and initialize it, adjust the initial weight of the BP neural network model with the gene sequence, and input the important features to train the BP neural network model;

[0009] Take the load prediction accuracy of the cultivation environment in the model weight training as the fitness value, group the hybrid rice by different groups, update and optimize the hybrid rice based on different groupings, and train the initial weight of the BP neural network model with the updated and optimized hybrid rice;

[0010] When the update optimization process detects the termination condition, the update optimization is stopped, the optimal weight parameter configuration of the BP neural network is output, and the important features are input into the BP neural network model with the optimal weight parameter configuration to determine the air conditioning operation scheme of the cultivation environment.

[0011] In one of the embodiments, the line classification groups include:

[0012] The maintainer line, the restorer line, and the sterile line;

[0013] The update optimization of the hybrid rice based on different line classification groups includes:

[0014] The sterile line is crossed with the maintainer line, and whether the update optimization of the sterile line is performed is determined based on the fitness value of the cross result;

[0015] The restorer line is self-crossed, and whether the update optimization of the restorer line is performed is determined based on the maximum self-crossing number.

[0016] In one of the embodiments, the method further includes:

[0017] The sterile line is crossed with the maintainer line by using a cross formula, and whether the sterile line is replaced by the offspring rice seed is determined based on the fitness value of the offspring rice seed in the cross result in the BP neural network model, when the fitness value is greater than the fitness value of the maintainer line, otherwise, the sterile line is retained;

[0018] The cross formula includes:

[0019]

[0020] The above formula is the process of crossing the tth generation, is the dth gene in the gene sequence of the ith rice seed in the sterile line, is the dth gene in the gene sequence of the k2th rice seed selected from the maintainer line, is the dth gene in the gene sequence of the k2th rice seed selected from the sterile line, and r2 and r3 are random numbers in the range of [-1, 1], and r2+r3≠0.

[0021] In one of the embodiments, the method further includes:

[0022] When the restorer line completes self-crossing, whether the self-crossing number of the restorer line reaches the preset maximum self-crossing number is determined, if not, the restorer line is self-crossed by using a self-crossing formula to generate offspring rice seeds;

[0023] The self-crossing formula includes:

[0024]

[0025] wherein, denotes the i-th recovery line new individuals generated by self-crossing, denotes the k-th individual randomly selected from the recovery line, r4 is a uniform random number in the range of [0, 1], x best denotes the current optimal individual;

[0026] Based on the BP neural network model, the fitness value of the offspring rice seeds in the self-crossing result is detected, and the fitness value of the offspring rice seeds and the last generation rice seeds is compared. When the fitness value of the offspring rice seeds is high, the last generation rice seeds is updated and optimized by the offspring rice seeds, and the self-crossing number is set to 0. Otherwise, the last generation rice seeds is retained, and the self-crossing number is increased by 1.

[0027] In one of the embodiments, the method further comprises:

[0028] When it is detected that the self-crossing number reaches the maximum self-crossing number, the rice seeds are reset by the following formula:

[0029]

[0030] wherein, denotes the recovery line new rice seeds generated by the reset operation, x min , x max are the upper and lower limits of the gene quantization, and r5 is a uniform random number in the range of [0, 1].

[0031] In one of the embodiments, the method further comprises:

[0032] The important features are input into the BP neural network model configured with the optimal weight parameter, and the output final load prediction result is obtained.

[0033] The load type and load value of the load prediction result are detected, and the air conditioning operation scheme of the cultivation environment is determined in combination with the environmental parameters.

[0034] The embodiment of the application provides a load prediction system based on hybrid rice algorithm optimization, which comprises:

[0035] An acquisition module is configured to acquire associated parameters of a hybrid rice cultivation environment, and to pre-process the associated parameters, wherein the associated parameters comprise air conditioning operation parameters and environmental parameters.

[0036] An evaluation module is configured to evaluate the importance of the correlation parameters by a preset algorithm, determine important features in the correlation parameters, and construct a BP neural network model for predicting the load of the cultivation environment based on the number of the important features as dimensions.

[0037] A weight module is configured to obtain a gene sequence of the hybrid rice and initialize the BP neural network model, adjust the initial weights of the BP neural network model based on the gene sequence, and input the important features to the BP neural network model for model training.

[0038] A training module is configured to take the prediction accuracy of the load of the cultivation environment in the model weight training as a fitness value, group the hybrid rice by strains, update and optimize the hybrid rice based on different strain groups, and train the initial weights of the BP neural network model based on the hybrid rice after the update and optimization.

[0039] An output module is configured to stop the update and optimization when a termination condition is triggered in the update and optimization process, output the optimal weight parameter configuration of the BP neural network, input the important features into the BP neural network model with the optimal weight parameter configuration, and determine the air conditioning operation scheme of the cultivation environment.

[0040] In one of the embodiments, the system further comprises:

[0041] A first optimization module is configured to cross the sterile line with the maintainer line, and determine whether to update and optimize the sterile line based on the fitness value of the cross result.

[0042] A second optimization module is configured to self-cross the restorer line, and determine whether to update and optimize the restorer line based on the maximum number of self-crossing.

[0043] An embodiment of the present application provides an electronic device, comprising a processor and a memory;

[0044] The processor is connected to the memory.

[0045] The memory is configured to store executable program codes.

[0046] The processor runs a program corresponding to the executable program codes by reading the executable program codes stored in the memory, so as to execute the method described in one or more embodiments.

[0047] An embodiment of the present application provides a non-transitory computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the steps of the load prediction method based on the hybrid rice algorithm optimization.

[0048] In view of the above, in one or more embodiments of the present specification, the correlation parameters of the hybrid rice breeding environment are obtained, and the correlation parameters are preprocessed, including air conditioning operation parameters and environmental parameters; the importance of the correlation parameters is evaluated through a preset algorithm, important features in the correlation parameters are determined, and the number of important features is used as the dimension to construct a BP neural network model for load prediction of the breeding environment; the genetic sequence of the hybrid rice is obtained and initialized, the initial weight of the BP neural network model is adjusted based on the genetic sequence, and the important features are used as input to train the BP neural network model; the correct rate of load prediction of the breeding environment in the model weight training is used as the fitness value, the hybrid rice is grouped by strain, and the hybrid rice is updated and optimized based on different strain groups, and the initial weight of the BP neural network model is trained based on the hybrid rice after the update and optimization; when the update and optimization process triggers the termination condition, the update and optimization are stopped, the optimal weight parameter configuration of the BP neural network is output, and the important features are input into the BP neural network model with the optimal weight parameter configuration to determine the air conditioning operation scheme of the breeding environment. In this way, the load of the corresponding room can be predicted in real time, and the system can automatically adjust the start and stop time of the air conditioner according to the predicted load demand, so as to ensure that the air conditioner starts and stops at the right time and avoids unnecessary energy consumption. BRIEF DESCRIPTION OF DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and those skilled in the art can also obtain other drawings according to these drawings without creative labor.

[0050] Figure 1 is a flowchart of a load prediction method based on hybrid rice algorithm optimization provided by one embodiment of the present specification.

[0051] Figure 2 is a flowchart of a hybrid rice algorithm optimized BP neural network weight parameter configuration provided by one embodiment of the present specification.

[0052] Figure 3 is a structural schematic diagram of a load prediction system based on hybrid rice algorithm optimization provided by one embodiment of the present specification.

[0053] Figure 4 is a structural schematic diagram of an electronic device provided by one embodiment of the present specification. DETAILED DESCRIPTION

[0054] The subject matter described herein will now be discussed with reference to example implementations. It should be understood that these implementations are discussed so that a better appreciation of the subject matter described herein can be attained, and are not intended to limit the scope of protection, applicability, or examples set forth in the claims. Changes in the function and arrangement of elements discussed can be made without departing from the scope of the subject matter covered by the present disclosure. Various examples can omit, substitute, or add various procedures or components as appropriate. For instance, the methods described can be performed in an order different than that described, and other steps can be added, omitted, or combined. Also, features described with respect to some examples can be combined in other examples.

[0055] As used herein, the terms "includes," "including," "has," "having," "contains," "containing," "comprises," "comprising," "including," "contains," "containing" and the like can mean "including, but not limited to." The term "based on" means "based, at least in part, on." The terms "one embodiment," "an embodiment," "example embodiment," and the like can mean "at least one embodiment." The term "another embodiment" means "at least one other embodiment." The terms "first," "second," "third," etc. can refer to different or the same objects. Other definitions can be found throughout this disclosure, whether explicitly stated or implicitly suggested. The definitions are equally applicable to uses of the terms throughout this specification.

[0056] As Figure 1 shown, the embodiment of the present application provides a breeding environment load prediction method based on hybrid rice algorithm optimization, comprising:

[0057] Step S101, obtain the correlation parameters of the hybrid rice breeding environment, and preprocess the correlation parameters, the correlation parameters including air conditioner running parameters and environment parameters.

[0058] Specifically, the associated parameters of the hybrid rice cultivation environment are obtained, wherein the collection of the associated parameters can be achieved by establishing a data integration platform or using an Internet of Things (IoT) platform for collecting, storing and preliminarily processing historical and real-time data collected from various sensors. When using the Internet of Things (IoT), the Lora protocol can be applied for communication. Reliable connection performance and security are provided for various Internet of Things applications. Wireless transmission of real-time data from sensors to edge gateways is achieved. The associated parameters collected by the sensors can include relevant parameters of the space air conditioner and other environmental parameters during the growth of hybrid rice, and the associated parameters at least include indoor temperature and outdoor temperature at several different positions, indoor humidity, air conditioner start time, air conditioner air supply, air conditioner working time, air conditioner working state under the corresponding working time, and production power and steam consumption data. The recording of the working state of the air conditioner includes but is not limited to operating mode (such as refrigeration, heating, dehumidification, etc.), power consumption, fault alarm, and power consumption of the air conditioning system itself and possible auxiliary heating (such as steam) steam consumption. After the associated parameters are determined, the associated parameters need to be sorted and preprocessed. The sorting can be, for example, regular checking of the integrity and accuracy of each data and processing of the abnormal data. The preprocessing can be, for example, cleaning, missing value processing, standardization or normalization, etc.

[0059] In step S102, the importance of the associated parameters is evaluated by a preset algorithm, important features in the associated parameters are determined, and a BP neural network model for predicting the load of the cultivation environment is constructed based on the number of important features as the dimension.

[0060] Specifically, the associated parameters obtained in the above steps need to be screened to determine important features with high importance in the associated parameters. The method for evaluating the importance of the features of the associated parameters can include feature analysis and selection by using random forest and XGBoost (Extreme Gradient Boosting), to obtain optimal feature parameters. The specific steps include: constructing a data set of the associated parameters, then each tree in the random forest is trained using a feature subset in the data set, the "importance" of each feature can be evaluated by observing the frequency of the feature as a splitting point in all trees and its contribution to the reduction of impurity (such as Gini index or entropy) when splitting. XGBoost can calculate the frequency of each feature in all trees to determine the importance of the feature by constructing a tree model through a gradient boosting framework. By combining the above methods, a feature importance list can be generated, in which each feature is associated with an importance score. The list is sorted to identify features with a score higher than a preset threshold, which are the important features in the associated parameters required in the subsequent steps.

[0061] After determining the important features, a BP neural network model for predicting the breeding environment load is constructed, wherein the room load refers to the total amount of heat that needs to be removed or added by an air conditioning system in the field of building environmental control and heating, ventilation and air conditioning (HVAC) in order to maintain the indoor environment within the set comfort range, i.e. the comfort range of hybrid rice in the embodiment. The room load can be divided into cold load and heat load. The BP neural network consists of four parts: input layer, hidden layer, fully connected layer and output layer. The input layer is an n-dimensional feature vector input, and n depends on the number of important features to ensure that the neural network can receive information of all important features. The hidden layer is located after the input layer, and each neuron of the hidden layer is connected to all neurons of the previous layer. The fully connected layer flattens and connects the features extracted by the previous layer to the output layer, thereby mapping the features to the final prediction result. In the prediction of the breeding environment load, the fully connected layer can realize the mapping between the input features and the output prediction by learning the complex relationship between the features, and the output layer can use a linear activation function (such as an identity function) to directly output the predicted value of the room load, because the load prediction is a regression problem, for example, in the load prediction formula F(wx+b), w is the connection weight, b is the bias term, and F is the activation function Relu. The activation function Relu is usually used in the hidden layer, and the output layer may use a linear function to directly output a continuous numerical value. In addition, after the BP neural network completes the preliminary prediction of the room load, a support vector machine (SVM) is introduced to further process the prediction result, specifically for "start-stop time judgment". The SVM is used to analyze the predicted load curve and identify the key points of load change, such as when the predicted load reaches the threshold value that requires starting or stopping the air conditioning system. The SVM extracts features from the neural network prediction result by learning, establishes a classification boundary, and judges whether the condition for starting or stopping the air conditioning is met.

[0062] In step S103, the genetic sequence of hybrid rice is obtained and initialized, the initial weight of the BP neural network model is adjusted based on the genetic sequence, and the important features are input to the BP neural network model for model training.

[0063] Specifically, the corresponding seeds of hybrid rice are obtained, the gene sequence of the hybrid rice seeds is decoded, and the initial configuration of the BP neural network parameters is initialized. The length of the gene sequence corresponds to the number of features in the input data set, that is, the gene sequence of the hybrid rice is decoded into feature weights, each gene in the gene sequence corresponds to the weight of a feature, and the input training set is trained by using the BP neural network algorithm to obtain a cultivation environment load prediction model corresponding to each group of weights. The specific steps can be: encoding, each "gene" or feature weight can be initialized by a random or heuristic method to ensure that the entire possible weight space is covered; decoding: converting these gene sequences into actual weight values and applying them to the connections between the input layer and the hidden layer and between the hidden layer and the output layer of the BP neural network; model training: using the initial weights, training the network using the BP algorithm, inputting the data set of important features in the above steps, obtaining the accuracy of load prediction, and the training target is to minimize the error between the predicted load and the actual load. The load prediction accuracy of each model is used as the fitness function value of the corresponding hybrid rice seed.

[0064] In step S104, the cultivation environment load prediction accuracy in the model weight training is used as the fitness value, the hybrid rice is grouped according to the different lines, and the hybrid rice is updated and optimized based on different line grouping, and the initial weight training of the BP neural network model is performed through the updated and optimized hybrid rice.

[0065] Specifically, the hybrid rice seeds are arranged according to the fitness size according to the fitness function value of the trained hybrid rice seeds, and the hybrid rice seeds are sequentially divided into a maintenance line, a recovery line and a sterile line according to the fitness size. The sterile line has the smallest fitness, is a plant line that cannot produce fertile pollen and can only be pollinated by cross-pollination, and is used as a female parent in hybrid breeding to receive pollen from the recovery line or the maintenance line for hybridization. The recovery line has a medium fitness and is used as a female parent in hybrid breeding to receive pollen from the recovery line or the maintenance line for hybridization, which overcomes the sterility of the sterile line to enable the hybrid offspring to reproduce normally. The maintenance line has the highest fitness, which is used for reproduction and maintenance of the sterile line. Then, the hybrid rice is updated and optimized according to different line grouping. The process of updating and optimizing can be as shown in Figure 2 Figure 2 The flowchart of the hybrid rice algorithm optimizing the BP neural network weight parameter configuration first sets the related parameters of the updating and optimizing process, including the maximum number of iterations I max , the maximum number of self-interactions T max , the current generation number k = 0, and the current self-interaction number t i = 0, then calculates the fitness value of the rice seeds and divides the types by using the weighted integration algorithm, and updates the rice seeds according to different types.

[0066] ​For the sterile line and the maintainer line:

[0067] The sterile line and the maintainer line are crossed according to the following crossing formula. If the fitness value of the new gene sequence generated is greater than the fitness value of the selected maintainer line, the current sterile line is replaced, otherwise the sterile line of the last generation is retained:

[0068]

[0069] The above formula is the process of crossing in the tth generation, is the dth gene in the gene sequence of the ith rice seed in the sterile line, is the dth gene in the gene sequence of the k2th rice seed selected from the maintainer line. is the dth gene in the gene sequence of the k2th rice seed selected from the sterile line. r2, r3 are random numbers in the range of [-1, 1], and r2+r3≠0.

[0070] For the restorer line:

[0071] The restorer line individual is self-crossed, and it is judged whether the individual of the restorer line reaches the maximum self-crossing number. If not, the restorer line is self-crossed to generate the next generation of rice seeds by the following self-crossing formula, wherein the maximum self-crossing number can be set by itself, and the specific value is related to the factors of maintaining genetic diversity and avoiding inbreeding. The specific self-crossing formula is:

[0072]

[0073] wherein, represents the ith restorer line new individual generated by self-crossing, represents the kth individual randomly selected from the restorer line. r4 is a uniform random number in the range of [0, 1]. x best represents the current optimal individual.

[0074] In the next generation of self-crossing, the fitness of the next generation of rice seeds is compared with the original fitness of the last generation. If the seeds after self-crossing are better, the new seeds are retained and the self-crossing number is set to 0, otherwise the self-crossing number is increased by 1. If the maximum self-crossing number is reached, the rice seeds are reset according to the following formula:

[0075]

[0076] wherein, represents the restorer line new rice generated by the reset operation.x min , x max are the upper and lower limits of gene quantization, and r5 is a uniform random number in the range of [0, 1].

[0077] Step S105, when detecting that the update optimization process triggers the termination condition, stopping the update optimization, outputting the optimal weight parameter configuration of the BP neural network, and determining the air conditioning operation scheme of the cultivation environment in combination with the important features.

[0078] Specifically, when detecting that the update optimization of the hybrid rice seeds reaches the termination condition, such as the maximum iteration number reaching I max , and the maximum self-crossing number reaching T max , the gene sequence of the hybrid rice seed with the maximum fitness value is obtained, that is, the optimal weight parameter configuration of the corresponding BP neural network, and then the important features in the associated parameters are input into the BP neural network with the optimal weight parameter configuration to determine the corresponding load prediction result of the hybrid rice, the load prediction result including the load type, such as cold load or heat load, and the load value, that is, the specific numerical value of the corresponding heat, and then based on the result of the load prediction, in combination with the heat value determined in the environmental parameters, the air conditioning operation scheme of the cultivation environment is adjusted to meet the requirements of the load prediction result.

[0079] The load prediction method based on hybrid rice algorithm optimization provided by the embodiment of the application, obtains the associated parameters of the hybrid rice cultivation environment, and pre-processes the associated parameters, the associated parameters including air conditioning operation parameters and environmental parameters; evaluates the importance of the associated parameters through a preset algorithm, determines important features in the associated parameters, and constructs a BP neural network model for load prediction of the cultivation environment based on the number of important features as dimensions; obtains the gene sequence of the hybrid rice and initializes, adjusts the initial weight of the BP neural network model with the gene sequence, and inputs the important features as input to train the BP neural network model; takes the load prediction accuracy of the cultivation environment in the model weight training as the fitness value, groups the hybrid rice by different groups, and updates and optimizes the hybrid rice based on different groupings, and trains the initial weight of the BP neural network model through the hybrid rice after the update optimization; when detecting that the update optimization process triggers the termination condition, stopping the update optimization, outputting the optimal weight parameter configuration of the BP neural network, and inputting the important features into the BP neural network model with the optimal weight parameter configuration to determine the air conditioning operation scheme of the cultivation environment. In this way, the load of the corresponding room can be predicted in real time, and the system can automatically adjust the start and stop time of the air conditioner according to the predicted load demand, so as to ensure that the air conditioner starts and stops at the appropriate time and avoid unnecessary energy consumption.

[0080] Please refer to Figure 3 , Figure 3 is a structural schematic diagram of a load prediction system based on hybrid rice algorithm optimization provided by the embodiment of the application. As shown in Figure 3 , the system comprises:

[0081] The acquisition module S201 is configured to acquire the correlation parameters of the hybrid rice breeding environment, and pre-process the correlation parameters, wherein the correlation parameters include air conditioner operation parameters and environment parameters.

[0082] The evaluation module S202 is configured to evaluate the importance of the correlation parameters by using a preset algorithm, determine important features in the correlation parameters, and construct a BP neural network model for breeding environment load prediction based on the number of important features as dimensions.

[0083] The weight module S203 is configured to acquire a gene sequence of the hybrid rice and initialize, adjust the initial weight of the BP neural network model based on the gene sequence, and input the important features to train the BP neural network model.

[0084] The training module S204 is configured to use the correct rate of breeding environment load prediction in the model weight training as a fitness value, group the hybrid rice by strains, update and optimize the hybrid rice based on different strain groups, and train the initial weight of the BP neural network model based on the hybrid rice after the update and optimization.

[0085] The output module S205 is configured to stop the update and optimization when a termination condition of the detection update and optimization process is triggered, output the optimal weight parameter configuration of the BP neural network, input the important features into the BP neural network model with the optimal weight parameter configuration, and determine the air conditioner operation scheme of the breeding environment.

[0086] In one embodiment, the system further comprises:

[0087] The first optimization module is configured to cross the sterile line with the maintainer line, and determine whether to update and optimize the sterile line based on the fitness value of the cross result.

[0088] The second optimization module is configured to self-cross the restorer line, and determine whether to update and optimize the restorer line based on the maximum self-crossing number.

[0089] Those skilled in the art can clearly understand that the technical solutions of the embodiments of the present application can be implemented by means of software and / or hardware. The "unit" and "module" in the specification refer to software and / or hardware that can independently complete or cooperate with other components to complete a specific function, wherein the hardware can be, for example, a field programmable gate array (FPGA), an integrated circuit (IC), etc.

[0090] The processing units and / or modules of the embodiments of the present application can be implemented by analog circuits that implement the functions of the embodiments of the present application, or can be implemented by software that implements the functions of the embodiments of the present application.

[0091] Referring to Figure 3 , a structural schematic diagram of an electronic device related to the embodiments of the present application is shown, which can be used to implement the method in the embodiments shown in Figure 1 . As shown in Figure 3 , the electronic device 300 can include at least one processor 301, at least one network interface 304, a user interface 303, a memory 305, and at least one communication bus 302.

[0092] The communication bus 302 is used to realize the connection and communication between the components.

[0093] The user interface 303 can include a display screen (Display), a camera (Camera), and can optionally include a standard wired interface, a wireless interface.

[0094] The network interface 304 can optionally include a standard wired interface, a wireless interface (such as a WI-FI interface).

[0095] The processor 301 can include one or more processing cores. The processor 301 connects various parts in the entire electronic device 300 through various interfaces and lines, executes various functions of the terminal 300 and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory 305, and calling data stored in the memory 305. Optionally, the processor 301 can be implemented in at least one of a hardware form of a digital signal processing (Digital Signal Processing, DSP), a field-programmable gate array (Field-Programmable Gate Array, FPGA), and a programmable logic array (Programmable Logic Array, PLA). The processor 301 can be a combination of one or more of a central processing unit (Central Processing Unit, CPU), a graphics processing unit (Graphics Processing Unit, GPU), and a modem. Among them, the CPU is mainly used to process the operating system, user interface, and application programs; the GPU is used to render and draw the content to be displayed on the display screen; and the modem is used to process wireless communication. It can be understood that the above-mentioned modem can also not be integrated into the processor 301, but can be implemented by a separate chip.

[0096] The memory 305 may include random access memory (RAM) or read-only memory. Optionally, the memory 305 may include a non-transitory computer-readable storage medium. The memory 305 can be used to store instructions, programs, code, code sets, or instruction sets. The memory 305 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-described method embodiments, etc.; the data storage area may store data involved in the above-described method embodiments, etc. Optionally, the memory 305 may also be at least one storage device located remotely from the aforementioned processor 301. Figure 3 As shown, the memory 305, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and program instructions.

[0097] exist Figure 3 In the electronic device 300 shown, the user interface 303 is mainly used to provide an input interface for the user and obtain the user input data; while the processor 301 can be used to call the image-based interactive application stored in the memory 305 and specifically perform the following operations: acquire the inspection images collected by the UAV when performing the inspection task, as well as the inspection route corresponding to the inspection task, and determine the image position and environmental information of the image position by combining the marked targets in the inspection image; detect the sharpness score of the inspection image, and when the sharpness score is lower than the preset score, establish a training image dataset based on the sharpness score, construct a first neural network model, take the blurred image as input data and the sharp image as output data, backpropagate to optimize the weights, and obtain the trained first neural network model. The training image dataset includes blurred images and sharp images; input the environmental information and sharpness score into the second neural network model for training, and iteratively train to obtain the trained second neural network model; when a real-time inspection task is received, acquire the real-time environmental information of the route corresponding to the real-time inspection task, input the real-time environmental information into the second neural network model, output the estimated sharpness score, and configure the corresponding first neural network model based on the estimated sharpness score to perform real-time enhancement of the captured image.

[0098] The application further provides a computer readable storage medium, which stores a computer program. The computer program is executed by a processor to implement the steps of the method. The computer readable storage medium can include, but is not limited to, any type of disk, including a floppy disk, an optical disk, a DVD, a CD-ROM, a micro drive, a magneto-optical disk, a ROM, a RAM, an EPROM, an EEPROM, a DRAM, a VRAM, a flash memory device, a magnetic card or an optical card, a nanosystem (including a molecular memory IC), or any type of medium or device suitable for storing instructions and / or data.

[0099] It should be noted that, for the foregoing method embodiments, in order to simply describe, they are all described as a series of action combinations, but those skilled in the art should know that the present application is not limited to the order of the actions described, because according to the present application, certain steps can be performed in other order or simultaneously. Secondly, those skilled in the art should know that the embodiments described in the specification all belong to preferred embodiments, and the actions and modules involved are not necessarily necessary for the present application.

[0100] In the above embodiments, the description of each embodiment is focused on, and the parts not described in detail in a certain embodiment can be referred to the related description of other embodiments.

[0101] In several embodiments provided by the present application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are only schematic. The division of the units is only a logical function division. There can be another division manner in actual implementation, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some services interfaces, devices or units, and can be electrical or other forms.

[0102] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, that is, they can be located in one place, or can be distributed on a plurality of network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment scheme.

[0103] In addition, each functional unit in each embodiment of the present application can be integrated into a processing unit, or each unit can exist physically independently, or two or more units can be integrated into one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0104] The integrated unit, if implemented in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable memory. Based on such understanding, the technical solutions of the present application, essentially or in part, or all or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a memory and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned memory includes various media that can store program codes, such as a flash disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0105] A person of ordinary skill in the art can understand that all or part of the steps of the various methods in the above embodiments can be executed by a program instructing relevant hardware, and the program can be stored in a computer readable memory, which can include a flash disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, etc.

[0106] The above describes specific embodiments of the present specification. Other embodiments are within the scope of the appended claims. In some cases, the acts or steps recited in the claims can be performed in an order other than that in the embodiments and still achieve desirable results. In addition, the processes depicted in the figures do not necessarily require the particular order shown or sequential order to achieve the desired results. In certain implementations, multitasking and parallel processing can be advantageous or possible.

Claims

1. A load prediction method based on hybrid rice algorithm optimization, characterized in that, include: The correlation parameters of the hybrid rice cultivation environment are obtained, and the correlation parameters are preprocessed. The correlation parameters include air conditioning operation parameters and environmental parameters. The importance of the correlation parameters is evaluated by a preset algorithm to determine the important features among the correlation parameters. The number of the important features is used as a dimension to construct a BP neural network model for predicting environmental load. The gene sequence of hybrid rice is obtained and initialized. The initial weights of the BP neural network model are adjusted using the gene sequence. The important features are used as input to train the BP neural network model. Using the prediction accuracy of the cultivation environment load in the model weight training as the fitness value, the hybrid rice is grouped by line, and the hybrid rice is updated and optimized based on different line groups. The updated and optimized hybrid rice is used to train the initial weights of the BP neural network model. When the detection update optimization process triggers the termination condition, the update optimization stops, the optimal weight parameter configuration of the BP neural network is output, the important features are input into the BP neural network model with the optimal weight parameter configuration, and the air conditioning operation scheme of the cultivation environment is determined.

2. The load prediction method based on hybrid rice algorithm optimization according to claim 1, characterized in that, The department grouping includes: Maintainer lines, restorer lines, and sterile lines; The updating and optimization of the hybrid rice based on different line groupings includes: The sterile line and the maintainer line are crossed, and the fitness value of the hybridization result is used to determine whether the sterile line should be updated and optimized. The restorer line is self-crossed, and the maximum number of self-crosses is used to determine whether the restorer line should be updated and optimized.

3. The load prediction method based on hybrid rice algorithm optimization according to claim 2, characterized in that, The step of hybridizing the sterile line with the maintainer line, and determining whether to update and optimize the sterile line based on the fitness value of the hybridization result, includes: The sterile line and the maintainer line were hybridized using a hybridization formula. The fitness value of the offspring rice seeds in the hybridization results was detected based on a BP neural network model. When the fitness value was greater than that of the maintainer line, the sterile line was replaced with offspring rice seeds; otherwise, the sterile line was retained. The hybridization formula includes: ， The above formula describes the hybridization process in the t-th generation. This refers to the d-th gene in the gene sequence of the i-th rice seed in the male-sterile line. This refers to the d-th gene in the k2-th rice seed gene sequence selected from the maintainer line. This refers to the d-th gene in the k2-th rice seed gene sequence selected from the sterile line. , The result is a random number in the range [-1, 1], and .

4. The load prediction method based on hybrid rice algorithm optimization according to claim 2, characterized in that, The step of performing self-crossing on the restorer line and determining whether to update and optimize the restorer line based on the maximum number of self-crossings includes: When the restorer line completes self-pollination, it is determined whether the number of self-pollinations of the restorer line has reached the preset maximum number of self-pollinations. If the preset maximum number of self-pollinations has not been reached, the restorer line produces offspring rice seeds through self-pollination using the self-pollination formula. The self-crossing formula includes: ， in, Represents the i-th recovery system New individuals produced through self-fertilization This represents the k-th individual randomly selected from the restorer line. A uniformly random number within the range [0,1]. This represents the current optimal individual; The fitness values ​​of offspring rice seeds in the self-pollination results are detected using a BP neural network model. The fitness values ​​of offspring rice seeds are compared with those of the previous generation of rice seeds. When the fitness value of the offspring rice seeds is high, the previous generation of rice seeds are updated and optimized using the offspring rice seeds, and the self-pollination count is set to 0. Otherwise, the previous generation of rice seeds are retained and the self-pollination count is incremented by 1.

5. The load prediction method based on hybrid rice algorithm optimization according to claim 4, characterized in that, The method further includes: When the self-pollination count is detected to have reached the maximum self-pollination count, the rice seeds are reset using the following formula: ， in, Indicates the recovery system New rice seeds generated through the reset operation. , These are the upper and lower limits after gene quantification. It is a uniform random number in the range [0,1].

6. The load prediction method based on hybrid rice algorithm optimization according to claim 1, characterized in that, The step of inputting the important features into a BP neural network model with optimal weight parameter configuration to determine the air conditioning operation scheme for the cultivation environment includes: The key features are input into a BP neural network model with optimal weight parameters to obtain the final load prediction result. The load type and load value of the load prediction results are detected, and the air conditioning operation plan for the cultivation environment is determined in combination with the environmental parameters.

7. A load prediction system based on hybrid rice algorithm optimization, characterized in that, The system includes: The acquisition module is used to acquire the associated parameters of the hybrid rice cultivation environment and preprocess the associated parameters, which include air conditioning operation parameters and environmental parameters. The evaluation module is used to evaluate the importance of the correlation parameters through a preset algorithm, determine the important features among the correlation parameters, and construct a BP neural network model for predicting environmental load using the number of the important features as a dimension. The weight module is used to obtain and initialize the gene sequence of hybrid rice, adjust the initial weights of the BP neural network model with the gene sequence, and train the BP neural network model by taking the important features as input. The training module is used to group the hybrid rice into lines using the prediction accuracy of the cultivation environment load in the model weight training as the fitness value, and to update and optimize the hybrid rice based on different line groups. The updated and optimized hybrid rice is used to train the initial weights of the BP neural network model. The output module is used to stop the update optimization when the detection update optimization process triggers the termination condition, output the optimal weight parameter configuration of the BP neural network, input the important features into the BP neural network model with the optimal weight parameter configuration, and determine the air conditioning operation scheme of the cultivation environment.

8. The load prediction system based on hybrid rice algorithm optimization according to claim 7, characterized in that, The system also includes: The first optimization module is used to cross the sterile line with the maintainer line, and determine whether to update and optimize the sterile line based on the fitness value of the crossover result. The second optimization module is used to perform self-crossing on the restorer line and determine whether to update and optimize the restorer line based on the maximum number of self-crossings.

9. An electronic device, comprising a processor and a memory; The processor is connected to the memory; The memory is used to store executable program code; The processor runs a program corresponding to the executable program code stored in the memory to perform the method as described in any one of claims 1-6.

10. A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method as described in any one of claims 1-6.

Images