Method for killing stored tobacco powder borers by feeding adult cotesia testis in batches

Abstract

The invention relates to the technical field of storage of tobacco powder borers, and particularly discloses a method for killing stored tobacco powder borers by feeding adult cotesia testis in batches, which comprises the following steps of: adding bee sweet water into a feeding box of host larvae of the cotesia testis for feeding; meanwhile, the temperature, the humidity and the bee sugar water liquid level in the feeding box are subjected to time sequence analysis by adopting a data processing technology based on deep learning, so that the time sequence consumption trend of the bee sugar water liquid level and a time sequence correlation mode of the temperature and the humidity in the box are extracted; the dynamic association relationship between bee sugar water consumption and the environmental condition in the box is disclosed by carrying out interactive analysis on the time sequence association characteristics of the temperature and humidity in the box and the time sequence consumption characteristics of the bee sugar water, so that the intelligent recommendation of the bee sugar water flow under the current environmental condition is realized. By means of the mode, optimal control over bee sugar water supply can be achieved, the breeding conditions of the adult cotesia gigas can be improved, and therefore the survival rate of the cotesia gigas and the quality of the adult cotesia gigas are improved, and the biological control effect is enhanced.

Description

Technical Field

[0001] The present application relates to the technical field of tobacco powder borer storage, and more specifically, to a method for mass-rearing adult Coleoptera parasitica to kill tobacco powder borers in storage. Background Art

[0002] With the development of storage management technology, the demand for controlling storage pests is increasing. Especially for the storage management of high-value crops such as tobacco, effective pest control measures are particularly important. Although traditional chemical control methods can effectively control pests, long-term use will lead to environmental pollution and increased pest resistance. Therefore, biological control as an environmentally friendly and sustainable method has received widespread attention.

[0003] For example, the invention patent with publication number CN110150233A discloses a method for mass breeding of adult wasps of the Gnaphalium spp. to kill tobacco borers in stored tobacco. The method uses honey sugar water as nutrient, cultivates the Gnaphalium spp. to parasitize and lay eggs on host larvae, and after the Gnaphalium spp. eggs have evolved from eggs, larvae, and pupae and finally become adult bees, they are anesthetized with low temperature until they cannot move, and transported to a warehouse where tobacco leaves are stored, where the adult wasps of the Gnaphalium spp. parasitize and kill tobacco borers on the stored tobacco leaves.

[0004] In the process of rearing the wasp, the supply management of honey sugar water is a key link. The proper supply of honey sugar water is directly related to the growth status, survival rate and final quality of adult bees of the wasp larvae. However, traditional rearing management methods often rely on manual observation and empirical judgment. This method is not only time-consuming and labor-intensive, but also difficult to accurately control the supply of honey sugar water, which can easily lead to waste of resources or insufficient honey sugar water, which may lead to the growth stunting or death of the wasp, thereby reducing the effectiveness of biological control.

[0005] Therefore, an optimized method for mass-rearing adult Psoralea corylidae to kill stored tobacco powder borer is desired. Summary of the invention

[0006] The present application provides a method for raising adult bees of the wheat moth braconid wasp in batches to kill stored tobacco powder borers, which can achieve optimized control of the supply of honey sugar water, improve the rearing conditions of the adult bees of the wheat moth braconid wasp, thereby increasing the survival rate and quality of the adult bees of the wheat moth braconid wasp, and enhancing the effect of biological control.

[0007] In a first aspect, a method for mass-rearing adult wasps of the wasp moth to kill stored tobacco powder borers is provided, comprising: placing host larvae of the wasp moth into a first rearing box, and introducing pairs of adult wasps of the first generation of the wasp moth into the first rearing box, adding honey water with a concentration of 10-70% to the first rearing box every day, and placing the first rearing box in an environment with a temperature of 27-29°C, a relative humidity of 70-80% and a photoperiod of 16L, and after the first generation of the wasp moth parasitizes and lays eggs on the host larvae, the second generation of the wasp moth can be obtained; after the eggs of the second generation of the wasp moth are passed through the process of eggs, larvae, , after the pupae finally emerge as adult bees, the second rearing box is transferred to an environment with a temperature of 2-6°C until the adult bees are anesthetized by low temperature and cannot move, and the adult bees of the lycopodiella are transferred from the first rearing box to the transport box for loading; the transport box is transported to a warehouse where tobacco leaves are stored, and the adult bees of the lycopodiella in the transport box are transferred to a bee release device, and when tobacco powder borer larvae appear on the tobacco bales, the bee release device can be opened for bee release, and the adult bees of the lycopodiella parasitize and kill the tobacco powder borer on the stored tobacco leaves, wherein honey water with a concentration of 10-70% is added to the first rearing box every day, including:

[0008] Obtaining the time queue of the temperature in the box and the time queue of the humidity in the box collected by the temperature sensor and the humidity sensor, and obtaining the time queue of the honey water level value in the box collected by the liquid level sensor;

[0009] Performing a time series consumption analysis on the time queue of the honey syrup water level value in the box to obtain an implicit correlation feature vector of the time series of the honey syrup water consumption in the box;

[0010] Performing time series correlation analysis on the time series of the temperature inside the box and the time series of the humidity inside the box to obtain an implicit feature vector of the time series correlation between the temperature inside the box and the humidity inside the box;

[0011] Interactively fusing the implicit correlation feature vector of the honey water consumption time series in the box and the implicit correlation feature vector of the temperature-humidity time series in the box based on feature mapping deviation compensation to obtain the feature vector of the honey water consumption time series pattern under the modulation of the box environment;

[0012] Based on the characteristic vector of the time series pattern of honey syrup water consumption in the box under the modulation of the environment in the box, an optimization instruction for the flow rate of honey syrup water is generated.

[0013] In a possible implementation, performing a time series consumption analysis on the time queue of the honey syrup water level value in the box to obtain an implicit correlation feature vector of the time series of the honey syrup water consumption in the box includes:

[0014] Calculate the difference between the values ​​of the honey syrup water level in the box at each two adjacent predetermined time points in the time queue of the honey syrup water level in the box to obtain a time queue representing the consumption of the honey syrup water in the box;

[0015] The time sequence representing the consumption of honey water in the box is input into a sequence encoder based on a forward LSTM model to obtain a temporal implicit correlation feature vector of the consumption of honey water in the box.

[0016] In a possible implementation, performing a time series correlation analysis on the time series of the temperature inside the box and the time series of the humidity inside the box to obtain an implicit feature vector of the time series correlation between the temperature inside the box and the humidity inside the box includes:

[0017] After arranging the time queue of the temperature inside the box and the time queue of the humidity inside the box into a time series input vector of the temperature inside the box and a time series input vector of the humidity inside the box according to the time dimension, calculating the association matrix between the time series input vector of the temperature inside the box and the time series input vector of the humidity inside the box to obtain a time series association matrix of the temperature-humidity inside the box;

[0018] The in-box temperature-humidity temporal association matrix is ​​input into the in-box environment temporal pattern feature extractor based on the hole convolutional neural network model to obtain the in-box temperature-humidity temporal association implicit feature vector.

[0019] In a possible implementation, the implicit correlation feature vector of the honey water consumption time series in the box and the implicit correlation feature vector of the temperature-humidity time series in the box are interactively fused based on feature mapping deviation compensation to obtain the time series pattern feature vector of the honey water consumption in the box under the modulation of the box environment, including:

[0020] Performing co-space mapping modulation on the honey water consumption time series implicit correlation feature vector in the box and the temperature-humidity time series implicit correlation feature vector in the box to obtain the modulated honey water consumption time series implicit correlation feature vector in the box and the modulated temperature-humidity time series implicit correlation feature vector in the box;

[0021] Based on the feature mapping deviation of the implicit correlation feature vector of the time series of syrup water consumption in the box after modulation and the implicit correlation feature vector of the time series of temperature-humidity in the box after modulation relative to the implicit correlation feature vector of the time series of syrup water consumption in the box and the implicit correlation feature vector of the time series of temperature-humidity in the box, the implicit correlation feature vector of the time series of syrup water consumption in the box after modulation and the implicit correlation feature vector of the time series of temperature-humidity in the box after modulation are interactively fused in a compensatory manner to obtain the time series pattern feature vector of syrup water consumption in the box under the modulation of the in-box environment.

[0022] In a possible implementation, the honey water consumption time series implicit correlation feature vector in the box and the temperature-humidity time series implicit correlation feature vector in the box are subjected to co-space mapping modulation to obtain the modulated honey water consumption time series implicit correlation feature vector in the box and the modulated temperature-humidity time series implicit correlation feature vector in the box, including:

[0023] The implicit correlation feature vector of the time series of syrup water consumption in the box and the implicit correlation feature vector of the time series of temperature-humidity in the box are input into a co-space mapper based on a fully connected layer to obtain the implicit correlation feature vector of the time series of syrup water consumption in the box after modulation and the implicit correlation feature vector of the time series of temperature-humidity in the box after modulation, wherein the implicit correlation feature vector of the time series of syrup water consumption in the box after modulation and the implicit correlation feature vector of the time series of temperature-humidity in the box after modulation have the same characteristic scale.

[0024] In a possible implementation, based on the feature mapping deviation of the implicit correlation feature vector of the time series of syrup water consumption in the box after modulation and the implicit correlation feature vector of the time series of temperature-humidity in the box after modulation relative to the implicit correlation feature vector of the time series of syrup water consumption in the box and the implicit correlation feature vector of the time series of temperature-humidity in the box, the implicit correlation feature vector of the time series of syrup water consumption in the box after modulation and the implicit correlation feature vector of the time series of temperature-humidity in the box after modulation are compensated interactively fused to obtain the time series pattern feature vector of syrup water consumption in the box under the modulation of the box environment, including:

[0025] Performing interpolation processing on the implicit correlation feature vector of the honey water consumption time series in the box and the implicit correlation feature vector of the temperature-humidity time series in the box to obtain the implicit correlation feature vector of the honey water consumption time series in the box after interpolation processing and the implicit correlation feature vector of the temperature-humidity time series in the box after interpolation processing, wherein the implicit correlation feature vector of the honey water consumption time series in the box after interpolation processing, the implicit correlation feature vector of the temperature-humidity time series in the box after interpolation processing, the implicit correlation feature vector of the honey water consumption time series in the box after modulation and the implicit correlation feature vector of the temperature-humidity time series in the box after modulation have the same characteristic scale;

[0026] Input the implicit correlation feature vector of the honey water consumption time series after the interpolation processing and the implicit correlation feature vector of the honey water consumption time series after the modulation, the implicit correlation feature vector of the temperature-humidity time series after the interpolation processing and the implicit correlation feature vector of the temperature-humidity time series after the modulation into the feature mapping deviation measurement network respectively to obtain the honey water consumption time series feature mapping deviation compensation factor and the temperature-humidity time series correlation feature mapping deviation compensation factor;

[0027] Based on the honey water consumption time series feature mapping deviation compensation factor and the box temperature-humidity time series correlation feature mapping deviation compensation factor, the modulation box honey water consumption time series implicit correlation feature vector and the modulation box temperature-humidity time series implicit correlation feature vector are compensated to obtain the compensated box honey water consumption time series implicit correlation feature vector and the compensated box temperature-humidity time series implicit correlation feature vector;

[0028] The implicit correlation feature vector of the time series of honey water consumption in the box after compensation and the implicit correlation feature vector of the time series of temperature-humidity in the box after compensation are input into the multi-dimensional significant interaction network between features to obtain the time series pattern feature vector of honey water consumption in the box under the modulation of the box environment.

[0029] In a possible implementation, the interpolated implicit correlation feature vector of the honey syrup water consumption time series in the box and the modulated implicit correlation feature vector of the honey syrup water consumption time series in the box are input into a feature map deviation measurement network to obtain a honey syrup water consumption time series feature map deviation compensation factor, including:

[0030] Calculate the positional dotted division vector between the time series implicitly associated feature vector of the honey syrup water consumption in the box after modulation and the time series implicitly associated feature vector of the honey syrup water consumption in the box after interpolation processing, and calculate the base-two logarithmic function value of the absolute value of each feature value in the dotted division vector to obtain a honey syrup water consumption time series feature mapping deviation representation vector;

[0031] Calculate the dot product vector between the honey water consumption timing feature map deviation representation vector and the honey water consumption timing implicit correlation feature vector in the modulated box, and calculate the exponential function value with the natural constant as the base and the sum of the eigenvalues ​​of the dot product vector as the exponent to obtain the honey water consumption timing feature map deviation compensation factor.

[0032] In a possible implementation, the time series implicit correlation feature vector of the syrup water consumption in the box after compensation and the time series implicit correlation feature vector of the temperature-humidity in the box after compensation are input into a multi-dimensional significant interaction network between features to obtain the time series pattern feature vector of the syrup water consumption in the box under the modulation of the box environment, including:

[0033] Respectively calculating the position point subtraction, position point multiplication, position point addition and position point division between the compensated in-box syrup water consumption time series implicit correlation feature vector and the compensated in-box temperature-humidity time series implicit correlation feature vector to obtain a first in-box environment-syrup water consumption time series interaction representation vector, a second in-box environment-syrup water consumption time series interaction representation vector, a third in-box environment-syrup water consumption time series interaction representation vector and a fourth in-box environment-syrup water consumption time series interaction representation vector;

[0034] The first in-box environment-honey sugar water consumption timing interaction representation vector, the second in-box environment-honey sugar water consumption timing interaction representation vector, the third in-box environment-honey sugar water consumption timing interaction representation vector and the fourth in-box environment-honey sugar water consumption timing interaction representation vector are cascaded and fused, and then input into a one-dimensional convolutional layer including a maximum pooling layer to obtain a characteristic vector of the in-box honey sugar water consumption timing pattern under the modulation of the in-box environment.

[0035] In a possible implementation, based on the characteristic vector of the time series pattern of honey syrup water consumption in the box under the modulation of the box environment, an optimization instruction for the flow rate of honey syrup water is generated, including:

[0036] The characteristic vector of the time series pattern of the syrup water consumption in the box under the modulation of the box environment is input into the syrup water flow dynamic optimizer based on the decoder to obtain the optimization instruction, and the optimization instruction includes the recommended flow value of the syrup water.

[0037] The present application provides a method for batch rearing of adult wheat moth braconids to kill stored tobacco powder borers, which adds honey sugar water to the feeding box of the host larvae of wheat moth braconids for feeding, and uses a data processing technology based on deep learning to perform time series analysis on the temperature, humidity and honey sugar water level in the feeding box, so as to extract the time series consumption trend of the honey sugar water level and the time series correlation mode of the temperature and humidity in the box, and then, by interactively analyzing the time series correlation characteristics of the temperature and humidity in the box and the time series consumption characteristics of honey sugar water, the dynamic correlation relationship between honey sugar water consumption and the environmental conditions in the box is revealed, and thereby the intelligent recommendation of the honey sugar water flow rate under the current environmental conditions is realized. In this way, the supply of honey sugar water can be optimized and controlled, and the feeding conditions of the adult wheat moth braconids can be improved, thereby improving the survival rate and adult quality of wheat moth braconids, and enhancing the effect of biological control. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings of the embodiments of the present application are briefly introduced below. Obviously, the drawings described below only relate to some embodiments of the present application, and are not intended to limit the present application.

[0039] Figure 1 This is a schematic flow chart of a method for mass-rearing adult Lycopodiella spp. to kill stored tobacco powder borers in an embodiment of the present application.

[0040] Figure 2 This is a schematic flow chart of adding 10-70% honey water into a first rearing box every day in a method for batch rearing of adult Lysimachia spp. to kill stored tobacco powder borers.

[0041] Figure 3 This is a schematic diagram of data flow in which 10-70% honey water is added to the first rearing box every day in the method for batch rearing of adult Lysimachia spp. to kill stored tobacco powder borer in an embodiment of the present application.

[0042] Figure 4 In the method of batch rearing adult B. pygmaeae to kill stored tobacco powder borer in the embodiment of the present application, a time series consumption analysis is performed on the time queue of the honey syrup water level value in the box to obtain a schematic flow chart of the implicit correlation feature vector of the honey syrup water consumption time series in the box.

[0043] Figure 5 In the method of batch rearing adult B. pygmaeae bees to kill stored tobacco powder borers in an embodiment of the present application, a time series correlation analysis is performed on the time queue of the temperature inside the box and the time queue of the humidity inside the box to obtain a schematic flow chart of the implicit feature vector of the time series correlation between the temperature and humidity inside the box. DETAILED DESCRIPTION

[0044] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work also fall within the scope of protection of the present application.

[0045] Patent CN110150233A proposes a method for mass-rearing adult wasps of the genus Psoralea to kill stored tobacco borers. Figure 1 This is a schematic flow chart of a method for batch rearing adult wasps of the present invention to kill stored tobacco powder borers. Figure 1 As shown, the method for batch rearing adult bees of the wheat moth braconid wasp to kill stored tobacco powder borer comprises: S1, placing the host larvae of the wheat moth braconid wasp into a first rearing box, and introducing pairs of first-generation wheat moth braconid wasp adult bees into the first rearing box, adding honey water with a concentration of 10-70% to the first rearing box every day, and placing the first rearing box in an environment with a temperature of 27-29°C, a relative humidity of 70-80% and a photoperiod of 16L, and after the first-generation wheat moth braconid wasp parasitizes and lays eggs on the host larvae, the second-generation wheat moth braconid wasp can be obtained; S2, waiting for the After the second generation of eggs of the wasp have evolved from eggs, larvae, and pupae to adult bees, the second breeding box is transferred to an environment with a temperature of 2-6°C until the adult bees are anesthetized by low temperature and cannot move, and the adult wasp is transferred from the first breeding box to a transport box for loading; S3, transporting the transport box to a warehouse where tobacco leaves are stored, and transferring the adult wasp in the transport box to a bee release device, and when tobacco powder borer larvae appear on the tobacco bales, the bee release device can be opened for releasing the bees, and the adult wasp parasitizes and kills the tobacco powder borer on the stored tobacco leaves.

[0046] In the process of rearing the wasp, the proper supply of honey sugar water is directly related to the growth status, survival rate and final quality of adult bees of the wasp larvae. However, traditional rearing management methods often rely on manual observation and empirical judgment. This method is not only time-consuming and labor-intensive, but also difficult to accurately control the supply of honey sugar water, which can easily lead to waste of resources or insufficient honey sugar water, which may lead to growth stagnation or death of the wasp, thereby reducing the effectiveness of biological control.

[0047] In view of the above technical problems, the technical concept of the present application is: adding honey sugar water to the feeding box of the host larvae of the wheat moth braconid wasp for feeding, and using the data processing technology based on deep learning to perform time series analysis on the temperature, humidity and honey sugar water level in the feeding box, so as to extract the time series consumption trend of the honey sugar water level and the time series correlation pattern of the temperature and humidity in the box, and then, through the interactive analysis of the time series correlation characteristics of the temperature and humidity in the box and the time series consumption characteristics of honey sugar water, to reveal the dynamic correlation relationship between honey sugar water consumption and the environmental conditions in the box, and thereby realize the intelligent recommendation of the honey sugar water flow rate under the current environmental conditions. In this way, the supply of honey sugar water can be optimized and controlled, and the feeding conditions of the adult bees of the wheat moth braconid wasp can be improved, thereby improving the survival rate and quality of the adult bees of the wheat moth braconid wasp, and enhancing the effect of biological control.

[0048] Optionally, in one embodiment of the present application, Figure 2 This is a schematic flow chart of adding 10-70% honey water into a first rearing box every day in a method for batch rearing of adult Lysimachia spp. to kill stored tobacco powder borers. Figure 3 This is a data flow diagram of adding 10-70% honey water to the first feeding box every day in the method of mass-rearing adult lycopodidae to kill stored tobacco powder borers in the embodiment of the present application. Figure 1 and Figure 2 As shown, 10-70% of honey water is added to the first feeding box every day, including: S11, obtaining the time queue of the temperature in the box and the time queue of the humidity in the box collected by the temperature sensor and the humidity sensor, and obtaining the time queue of the honey water level value in the box collected by the liquid level sensor; S12, performing a time series consumption analysis on the time queue of the honey water level value in the box to obtain the implicit correlation feature vector of the honey water consumption time series in the box; S13, performing a time series correlation analysis on the time queue of the temperature in the box and the time queue of the humidity in the box to obtain the implicit feature vector of the time series correlation of the temperature-humidity in the box; S14, performing interactive fusion based on feature mapping deviation compensation on the implicit correlation feature vector of the honey water consumption time series in the box and the implicit feature vector of the temperature-humidity time series correlation in the box to obtain the feature vector of the time series pattern of the honey water consumption in the box under the modulation of the box environment; S15, generating an optimization instruction for the honey water flow rate based on the feature vector of the time series pattern of the honey water consumption in the box under the modulation of the box environment.

[0049] In the above-mentioned method of batch-rearing adult wheat moth braconids to kill stored tobacco powder borers, the S11, obtains the time queue of the temperature in the box collected by the temperature sensor and the humidity sensor and the time queue of the humidity in the box, and takes the time queue of the honey sugar water level value in the box collected by the liquid level sensor. It should be understood that temperature and humidity are important environmental factors that affect the growth and development of wheat moth braconids and their host larvae. Specifically, higher temperatures will accelerate the metabolism of wheat moth braconids and host larvae, resulting in an increase in the demand for energy, so that the consumption of honey sugar water increases. In a high humidity environment, due to the sufficient water in the surrounding environment, the water demand of wheat moth braconids and host larvae is relatively low, which may reduce the demand for water from honey sugar water, resulting in a decrease in the consumption of honey sugar water. Therefore, in the technical solution of the present application, it is expected to optimize the supply of honey sugar water by analyzing the influence of temperature and humidity in the box on the consumption of honey sugar water, and ensure that the wheat moth braconid larvae obtain a suitable growth environment.

[0050] In the above-mentioned method of batch-rearing adult bees of the wheat moth braconid to kill the stored tobacco powder borer, the S12 performs a time series consumption analysis on the time queue of the honey sugar water level value in the box to obtain the implicit correlation feature vector of the honey sugar water consumption time series in the box. It should be understood that through the time series consumption analysis, the consumption of honey sugar water in different time periods can be accurately understood, which is very important for optimizing the feeding conditions. Specifically, through the analysis of the consumption of honey sugar water, the changes in the nutritional needs of the wheat moth braconid in different time periods can be understood, so as to formulate a more reasonable supply strategy to ensure that the adult bees can obtain sufficient nutrition throughout the transportation process and improve their survival rate. Time series consumption analysis can help identify the laws and patterns of honey sugar water consumption, such as whether the consumption in the early and late stages of transportation is different, or the changing trend of consumption under specific conditions (such as temperature changes), and then accurately control the supply of honey sugar water to avoid waste of resources or insufficient honey sugar water. The consumption of honey sugar water can also indirectly reflect the health of adult bees. If the consumption of honey sugar water in a transport box is found to be abnormal, it may mean that the adult bees in the box are in poor health and need to take timely measures.

[0051] Optionally, in one embodiment of the present application, Figure 4 In the method of batch rearing adult lycopodidae to kill stored tobacco powder borers in the embodiment of the present application, a time series consumption analysis is performed on the time queue of the honey sugar water level value in the box to obtain a schematic flow chart of the implicit correlation feature vector of the honey sugar water consumption time series in the box. Figure 4As shown, a time series consumption analysis is performed on the time queue of the honey syrup water level value in the box to obtain a time series implicit correlation feature vector of the honey syrup water consumption in the box, including: S121, calculating the difference between the honey syrup water level values ​​in the box at each two adjacent predetermined time points in the time queue of the honey syrup water level value in the box to obtain a time queue representing the consumption of the honey syrup water in the box; S122, inputting the time queue representing the consumption of the honey syrup water in the box into a sequence encoder based on a forward LSTM model to obtain a time series implicit correlation feature vector of the consumption of the honey syrup water in the box.

[0052] In the above method for batch-rearing adult larvae of the lycopodidae to kill stored tobacco powder borers, the S121 is to calculate the difference between the liquid level values ​​of the lycopod water in the box at each adjacent two predetermined time points in the time queue of the liquid level values ​​of the lycopod water in the box to obtain a time queue representing the consumption of the lycopod water in the box. It should be understood that by calculating the difference between the liquid level values ​​of the lycopod water in the box at each adjacent two predetermined time points in the time queue of the liquid level values ​​of the lycopod water in the box, the consumption of the lycopod water is quantitatively represented, and the time queue representing the consumption of the lycopod water in the box is obtained, thereby providing more intuitive data support for the analysis of the time series consumption trend of the lycopod water.

[0053] In the above-mentioned method of batch-rearing adult bees of the meal moth braconid to kill stored tobacco powder borers, the S122 inputs the time queue represented by the consumption of honey syrup water in the box into a sequence encoder based on a forward LSTM model to obtain the implicit correlation feature vector of the time series of the consumption of honey syrup water in the box. It should be understood that in order to accurately capture the dynamic change characteristics of the consumption of honey syrup water, the present application further uses a forward LSTM model with excellent performance in the time series data analysis task as a sequence encoder to perform time series encoding on the time queue represented by the consumption of honey syrup water in the box, so as to use the long-term memory capacity of the LSTM model to capture the long-term dependency in the time queue, thereby extracting the time series change pattern of the consumption of honey syrup water, so as to obtain the implicit correlation feature vector of the time series of the consumption of honey syrup water in the box.

[0054] In the above-mentioned method of batch rearing adult bees of the wheat moth braconid to kill the stored tobacco powder borer, the S13, the time queue of the temperature in the box and the time queue of the humidity in the box are subjected to a temporal correlation analysis to obtain a temporal correlation implicit feature vector of the temperature-humidity in the box. It should be understood that the temperature and humidity in the rearing box have a significant joint effect on the growth and development of the wheat moth braconid and its demand for honey sugar water. For example, in an environment of high temperature and low humidity, the water demand of the wheat moth braconid and the host larvae increases, and the evaporation rate of water in the honey sugar water is also accelerated. This dual factor may lead to a significant increase in the consumption of honey sugar water. Therefore, in order to further reveal the influence of the temporal correlation interaction between temperature and humidity on the consumption of honey sugar water, the present application first performs a temporal correlation interaction analysis on the time queue of the temperature in the box and the time queue of the humidity in the box to obtain a comprehensive representation of the environmental state in the box.

[0055] Optionally, in one embodiment of the present application, Figure 5 In the method of batch rearing adult lycopodidae to kill stored tobacco moths, a schematic flow chart of performing temporal correlation analysis on the time sequence of the temperature in the box and the time sequence of the humidity in the box to obtain the implicit feature vector of the temporal correlation between the temperature and humidity in the box is provided. Figure 5 As shown, a time series correlation analysis is performed on the time queue of the temperature inside the box and the time queue of the humidity inside the box to obtain an implicit feature vector of the time series correlation between the temperature inside the box and the humidity inside the box, including: S131, after arranging the time queue of the temperature inside the box and the time queue of the humidity inside the box into a time series input vector of the temperature inside the box and a time series input vector of the humidity inside the box according to the time dimension, calculating the correlation matrix between the time series input vector of the temperature inside the box and the time series input vector of the humidity inside the box to obtain an implicit feature vector of the temperature inside the box and the humidity inside the box; S132, inputting the time series correlation matrix of the temperature inside the box into an in-box environment time series pattern feature extractor based on a void convolutional neural network model to obtain an implicit feature vector of the time series correlation between the temperature inside the box and the humidity inside the box.

[0056] In the above-mentioned method of batch rearing adult bees of the meal moth braconid to kill the stored tobacco powder borer, the S131, after arranging the time queue of the temperature in the box and the time queue of the humidity in the box into the time series input vector of the temperature in the box and the time series input vector of the humidity in the box according to the time dimension, calculates the association matrix between the time series input vector of the temperature in the box and the time series input vector of the humidity in the box to obtain the time series association matrix of the temperature in the box-humidity in the box. It should be understood that the time queue of the temperature in the box and the time queue of the humidity in the box are arranged according to the time dimension to maintain the time sequence relationship and time series correlation of the data, forming a structured time series input vector of the temperature in the box and the time series input vector of the humidity in the box. Then, by calculating the association matrix between the time series input vector of the temperature in the box and the time series input vector of the humidity in the box, the dynamic correlation between temperature and humidity is revealed. In a specific example of the present application, the Hadamard product between the transposed vector of the temperature time series input vector inside the box and the humidity time series input vector inside the box is calculated to reveal the degree of correlation between temperature and humidity in the time dimension, thereby obtaining the temperature-humidity time series correlation matrix inside the box.

[0057] In the above-mentioned method of batch-rearing adult bees of the meal moth cocoon wasp to kill the stored tobacco powder borer, the S132 inputs the temperature-humidity temporal association matrix in the box into the temporal pattern feature extractor of the box environment based on the hollow convolutional neural network model to obtain the implicit feature vector of the temporal association of the temperature-humidity in the box. It should be understood that in order to further abstract the deep-level temporal association characteristics of the temperature-humidity temporal association matrix in the box, the present application uses the temporal pattern feature extractor of the box environment based on the hollow convolutional neural network model to extract the features of the temporal association matrix in the box, so as to obtain the implicit feature vector of the temporal association of the temperature-humidity in the box. It should be known to those of ordinary skill in the art that the hollow convolutional neural network model can expand the receptive field without increasing the number of parameters by introducing the hollow convolution operation, and capture a wider range of temporal change association patterns of temperature and humidity, so as to more comprehensively understand the associated changes of environmental conditions.

[0058] In the above-mentioned method of batch-rearing adult bees of the meal moth cocoon wasp to kill the stored tobacco powder borer, the S14 is to interactively fuse the implicitly associated characteristic vector of the honey water consumption time series in the box and the implicitly associated characteristic vector of the temperature-humidity time series in the box based on feature mapping deviation compensation to obtain the characteristic vector of the honey water consumption time series pattern in the box under the modulation of the box environment. It should be understood that in order to realize the understanding of the temporal dynamic association between the environmental conditions in the box and the honey water consumption, the present application further interactively analyzes the implicitly associated characteristic vector of the honey water consumption time series in the box and the implicitly associated characteristic vector of the temperature-humidity time series in the box. In particular, in order to improve the feature interaction effect between multi-source data, the present application proposes an optimized feature interaction method, which realizes feature space alignment by co-spatial mapping the implicitly associated characteristic vector of the honey water consumption time series in the box and the implicitly associated characteristic vector of the temperature-humidity time series in the box, and enhances the correlation and information transmission efficiency between the two through co-spatial mapping deviation compensation, thereby improving the accuracy of feature interaction.

[0059] Optionally, in one embodiment of the present application, the implicit correlation feature vector of the honey water consumption time series in the box and the implicit correlation feature vector of the temperature-humidity time series in the box are interactively fused based on feature mapping deviation compensation to obtain the time series pattern feature vector of the honey water consumption in the box under the modulation of the environment in the box, including: performing co-space mapping modulation on the implicit correlation feature vector of the honey water consumption time series in the box and the implicit correlation feature vector of the temperature-humidity time series in the box to obtain the implicit correlation feature vector of the honey water consumption time series in the box after modulation and the time series pattern feature vector of the temperature-humidity in the box after modulation. Based on the feature mapping deviation of the implicit correlation feature vector of the time series of syrup water consumption in the box after modulation and the implicit correlation feature vector of the temperature-humidity time series in the box after modulation relative to the implicit correlation feature vector of the time series of syrup water consumption in the box and the implicit correlation feature vector of the temperature-humidity time series in the box, the implicit correlation feature vector of the time series of syrup water consumption in the box after modulation and the implicit correlation feature vector of the temperature-humidity time series in the box after modulation are interactively fused in a compensatory manner to obtain the time series pattern feature vector of syrup water consumption in the box under the modulation of the environment in the box.

[0060] Specifically, since the implicit feature vector of the temporal association of the honey water consumption in the box and the implicit feature vector of the temporal association of the temperature-humidity in the box are in different feature spaces, respectively, cross-space semantic interaction cannot be achieved. This application first uses a fully connected layer to construct a common space mapper, and uses the powerful nonlinear mapping capability of the fully connected layer to perform feature modulation on the two, and map them to the same feature space, ensuring that the two have the same feature scale to facilitate subsequent interactive processing.

[0061] Furthermore, considering that part of the original feature information may be lost during the feature space mapping process, the intrinsic connection or covariation relationship between the environmental features and the honey water consumption features is destroyed, thereby increasing the difficulty of subsequent association pattern mining. Therefore, in the technical solution of the present application, after the co-space mapping is performed, the implicit correlation feature vector of the honey water consumption time series in the box and the implicit feature vector of the temperature-humidity time series in the box after the co-space mapping modulation are further compensated for the feature mapping deviation. Specifically, first, the original implicit correlation feature vector of the honey water consumption time series in the box and the implicit feature vector of the temperature-humidity time series in the box are interpolated, so that the features after the interpolation processing have the same feature scale as the features after the co-space modulation, and while ensuring the alignment of the scales, more important information of the original features is retained, and the model's perception of detail information is improved. Then, the feature mapping deviation measurement network is used to compare and analyze the environmental features and honey water consumption features after the interpolation processing with the environmental features and honey water consumption features after the co-space modulation, so as to identify the deviations generated in the co-space mapping process, and accordingly calculate the corresponding compensation factor to quantify the degree of semantic deviation introduced in the transformation process. Then, based on the calculated compensation factor, the implicit correlation feature vector of the honey water consumption time series in the box and the implicit correlation feature vector of the temperature-humidity time series in the box after the above-mentioned co-space modulation are compensated and corrected to eliminate the information distortion caused by the feature space transformation and improve the model's sensitivity to detail information. Finally, a multi-dimensional interactive analysis is performed on the compensated implicit correlation feature vector of the honey water consumption time series in the box and the implicit correlation feature vector of the temperature-humidity time series in the box. Through multi-level feature interactive fusion, the correlation pattern of environmental features and honey water consumption features at different levels is learned, and the correlation response law between the two is fully excavated, thereby generating the characteristic vector of the honey water consumption time series pattern in the box under the modulation of the box environment.

[0062] Optionally, in one embodiment of the present application, the implicit correlation feature vector of the time series of syrup water consumption in the box and the implicit correlation feature vector of the time series of temperature-humidity in the box are co-space mapped and modulated to obtain the implicit correlation feature vector of the time series of syrup water consumption in the box after modulation and the implicit correlation feature vector of the time series of temperature-humidity in the box after modulation, including: inputting the implicit correlation feature vector of the time series of syrup water consumption in the box and the implicit correlation feature vector of the time series of temperature-humidity in the box into a co-space mapper based on a fully connected layer to obtain the implicit correlation feature vector of the time series of syrup water consumption in the box after modulation and the implicit correlation feature vector of the time series of temperature-humidity in the box after modulation, wherein the implicit correlation feature vector of the time series of syrup water consumption in the box after modulation and the implicit correlation feature vector of the time series of temperature-humidity in the box after modulation have the same characteristic scale.

[0063] Optionally, in one embodiment of the present application, based on the feature mapping deviation of the implicit correlation feature vector of the time series of honey syrup water consumption in the box after modulation and the implicit correlation feature vector of the time series of temperature-humidity in the box after modulation relative to the implicit correlation feature vector of the time series of honey syrup water consumption in the box and the implicit correlation feature vector of the time series of temperature-humidity in the box, the implicit correlation feature vector of the time series of honey syrup water consumption in the box after modulation and the implicit correlation feature vector of the time series of temperature-humidity in the box after modulation are compensated interactively fused to obtain the time series pattern feature vector of honey syrup water consumption in the box under the modulation of the box environment. , including: interpolating the implicit correlation feature vector of the honey water consumption time series in the box and the implicit correlation feature vector of the temperature-humidity time series in the box to obtain the implicit correlation feature vector of the honey water consumption time series in the box after interpolation and the implicit correlation feature vector of the temperature-humidity time series in the box after interpolation, wherein the implicit correlation feature vector of the honey water consumption time series in the box after interpolation, the implicit correlation feature vector of the temperature-humidity time series in the box after interpolation, the implicit correlation feature vector of the honey water consumption time series in the box after modulation and the implicit correlation feature vector of the temperature-humidity time series in the box after modulation are obtained. The quantities have the same characteristic scale; the implicit correlation feature vector of the honey water consumption time series in the box after interpolation processing and the implicit correlation feature vector of the honey water consumption time series in the box after modulation, the implicit correlation feature vector of the temperature-humidity time series in the box after interpolation processing and the implicit correlation feature vector of the temperature-humidity time series in the box after modulation are respectively input into the feature mapping deviation measurement network to obtain the honey water consumption time series feature mapping deviation compensation factor and the box temperature-humidity time series correlation feature mapping deviation compensation factor; based on the honey water consumption time series feature mapping deviation compensation factor and the box temperature-humidity time series correlation feature mapping deviation compensation factor, the honey water consumption time series feature mapping deviation compensation factor is calculated. The time series correlation feature mapping deviation compensation factor compensates the implicit correlation feature vector of the time series of syrup water consumption in the box after modulation and the implicit correlation feature vector of the temperature-humidity time series in the box after modulation to obtain the implicit correlation feature vector of the time series of syrup water consumption in the box after compensation and the implicit correlation feature vector of the temperature-humidity time series in the box after compensation; the implicit correlation feature vector of the time series of syrup water consumption in the box after compensation and the implicit correlation feature vector of the temperature-humidity time series in the box after compensation are input into the multi-dimensional significant interaction network between features to obtain the time series pattern feature vector of syrup water consumption in the box under the modulation of the box environment.

[0064] Optionally, in one embodiment of the present application, the interpolation-processed implicitly associated feature vector of the time series of syrup water consumption in the box and the modulation-processed implicitly associated feature vector are input into a feature mapping deviation measurement network to obtain a syrup water consumption time series feature mapping deviation compensation factor, including: calculating a positional dot division vector between the modulation-processed implicitly associated feature vector and the interpolation-processed implicitly associated feature vector, and calculating a base-two logarithmic function value of the absolute value of each eigenvalue in the dot division vector to obtain a syrup water consumption time series feature mapping deviation representation vector; calculating a dot product vector between the syrup water consumption time series feature mapping deviation representation vector and the modulation-processed implicitly associated feature vector, and calculating an exponential function value with a natural constant as the base and the sum of the eigenvalues ​​of the dot product vector as the exponent to obtain the syrup water consumption time series feature mapping deviation compensation factor.

[0065] Optionally, in one embodiment of the present application, the implicit correlation feature vector of the time series of honey water consumption in the box after compensation and the implicit correlation feature vector of the time series of temperature-humidity in the box after compensation are input into the multi-dimensional significant interaction network between features to obtain the time series pattern feature vector of honey water consumption in the box under the modulation of the environment in the box, including: respectively calculating the implicit correlation feature vector of the time series of honey water consumption in the box after compensation and the implicit correlation feature vector of the time series of temperature-humidity in the box after compensation, by position point subtraction, by position point multiplication, by position point addition and by position point division to obtain the first box environment-honey water consumption time series interaction representation vector the first in-box environment - honey sugar water consumption timing interaction representation vector, the second in-box environment - honey sugar water consumption timing interaction representation vector, the third in-box environment - honey sugar water consumption timing interaction representation vector and the fourth in-box environment - honey sugar water consumption timing interaction representation vector; the first in-box environment - honey sugar water consumption timing interaction representation vector, the second in-box environment - honey sugar water consumption timing interaction representation vector, the third in-box environment - honey sugar water consumption timing interaction representation vector and the fourth in-box environment - honey sugar water consumption timing interaction representation vector are cascaded and fused, and then input into a one-dimensional convolutional layer including a maximum pooling layer to obtain the in-box honey sugar water consumption timing pattern feature vector under the modulation of the in-box environment.

[0066] Optionally, in one embodiment of the present application, the syrup water consumption timing implicitly associated feature vector in the box after modulation is compensated based on the syrup water consumption timing characteristic mapping deviation compensation factor to obtain the compensated syrup water consumption timing implicitly associated feature vector in the box, which is used to: calculate the square of the syrup water consumption timing characteristic mapping deviation compensation factor and the sum of the square of the syrup water consumption timing implicitly associated feature vector as a correction factor, and divide each eigenvalue of the syrup water consumption timing implicitly associated feature vector in the box after modulation by the correction factor to obtain the compensated syrup water consumption timing implicitly associated feature vector in the box.

[0067] In summary, in the embodiment of the present application, the implicitly associated characteristic vector of the honey water consumption time series in the box and the implicitly associated characteristic vector of the temperature-humidity time series in the box are processed by the following characteristic mapping deviation compensation interaction formula to obtain the characteristic vector of the honey water consumption time series pattern in the box under the modulation of the box environment, wherein the characteristic mapping deviation compensation interaction formula is:

[0068] v t1 =Relu(W cm v1+b cm )

[0069] v t2 =Relu(W cm v1+b cm )

[0070] v c1 =L(v1)

[0071] v c2 =L(v2)

[0072]

[0073]

[0074]

[0075]

[0076]

[0077] Wherein, v1 represents the implicit correlation feature vector of the honey water consumption time series in the box, v2 represents the implicit correlation feature vector of the temperature-humidity time series in the box, and W cm and b cm Represent the weight matrix and bias term respectively, Relu represents the activation function, v t1 and v t2 They represent the implicit correlation feature vector of the honey water consumption time series in the box after modulation and the implicit correlation feature vector of the temperature-humidity time series in the box after modulation, L(·) represents the interpolation processing, v c1 and v c2 They represent the implicit correlation feature vector of the honey water consumption time series in the box after interpolation processing and the implicit correlation feature vector of the temperature-humidity time series in the box after interpolation processing, and represents the implicit correlation feature vector of the honey water consumption time series in the box after modulation, the implicit correlation feature vector of the honey water consumption time series in the box after interpolation, the implicit correlation feature vector of the temperature-humidity time series in the box after modulation, and the eigenvalue of the ith position in the implicit correlation feature vector of the temperature-humidity time series in the box after interpolation, log(·) represents the logarithmic function with base 2, exp(·) represents the natural exponential function, L represents the characteristic scale of the implicit correlation feature vector of the honey water consumption time series in the box after modulation, s1 and s2 represent the deviation compensation factor of the honey water consumption time series feature mapping and the deviation compensation factor of the temperature-humidity time series correlation feature mapping in the box, respectively, v b1 and v b2 They represent the implicit correlation feature vector of the honey water consumption time series in the box after compensation and the implicit correlation feature vector of the temperature-humidity time series in the box after compensation, ⊙ represents the dot product, Indicates point addition, Indicates point reduction, represents dot division, conv1D represents a one-dimensional convolution operation, MaxPool represents a maximum pooling operation, and V represents a temporal pattern feature vector of honey water consumption in the box under the modulation of the in-box environment.

[0078] In the above-mentioned method of batch-rearing adult bees of the lycopodidae to kill stored tobacco powder borers, the S15 generates an optimization instruction for the flow rate of lycopod water based on the characteristic vector of the time series pattern of lycopod water consumption in the box under the modulation of the box environment. Optionally, in one embodiment of the present application, based on the characteristic vector of the time series pattern of lycopod water consumption in the box under the modulation of the box environment, the optimization instruction for the flow rate of lycopod water is generated, including: inputting the characteristic vector of the time series pattern of lycopod water consumption in the box under the modulation of the box environment into a lycopod water flow dynamic optimizer based on a decoder to obtain the optimization instruction, and the optimization instruction includes a recommended flow value of lycopod water. That is, the characteristic vector of the time series pattern of lycopod water consumption in the box under the modulation of the box environment is input into a lycopod water flow dynamic optimizer based on a decoder for decoding and prediction, and according to the dynamic correlation between lycopod water consumption and the box environment conditions, the consumption of lycopod water under the current box environment conditions is analyzed, thereby realizing intelligent recommendation and optimization control of the lycopod water flow value.

[0079] Optionally, in one embodiment of the present application, the characteristic vector of the time series pattern of the consumption of syrup water in the box under the modulation of the box environment is input into a syrup water flow rate dynamic optimizer based on a decoder to obtain the optimization instruction, and the optimization instruction includes the recommended flow rate value of the syrup water, including: using the decoder to decode and predict the characteristic vector of the time series pattern of the consumption of syrup water in the box under the modulation of the box environment by the following formula to obtain the recommended flow rate value of the syrup water; wherein the formula is: Wherein, V is the characteristic vector of the time series pattern of honey water consumption in the box under the modulation of the box environment, Y is the recommended flow value of the honey water, W is the weight matrix of the decoder, Represents matrix multiplication.

[0080] Preferably, inputting the characteristic vector of the time series pattern of the honey water consumption in the box under the modulation of the box environment into the honey water flow dynamic optimizer based on the decoder to obtain the optimization instruction comprises:

[0081] Determine the number of zero eigenvalues ​​in the characteristic vector of the time series pattern of honey syrup water consumption in the box under the modulation of the box environment, and respectively calculate the reciprocal of the logarithm of the number of zero eigenvalues ​​with base two and the exponential value of the reciprocal of the number of zero eigenvalues ​​with base natural constants to obtain the characteristic representation value of the time series pattern of honey syrup water consumption in the box under the modulation of the first box environment and the characteristic representation value of the time series pattern of honey syrup water consumption in the box under the modulation of the second box environment;

[0082] Calculate the power function of each eigenvalue of the characteristic vector of the time series pattern of honey syrup water consumption in the box under the modulation of the box environment with the inverse of the number of zero eigenvalues ​​as the exponent to obtain the characteristic structure state vector of the time series pattern of honey syrup water consumption in the box under the modulation of the box environment;

[0083] Performing a dot multiplication of the characteristic structure state vector of the time series pattern of honey syrup water consumption in the box under the modulation of the box environment and the characteristic representation value of the time series pattern of honey syrup water consumption in the box under the modulation of the first box environment to obtain a characteristic information representation vector of the time series pattern of honey syrup water consumption in the box under the modulation of the box environment;

[0084] The autocorrelation matrix of the characteristic structure state vector of the time series pattern of honey syrup water consumption in the box under the modulation of the in-box environment is respectively multiplied by the characteristic representation value and weight hyperparameter of the time series pattern of honey syrup water consumption in the box under the modulation of the in-box environment to obtain the characteristic regression understanding matrix of the time series pattern of honey syrup water consumption in the box under the modulation of the in-box environment;

[0085] Performing matrix multiplication of the characteristic information representation vector of the time series pattern of honey syrup water consumption in the box under the modulation of the box environment and the characteristic regression understanding matrix of the time series pattern of honey syrup water consumption in the box under the modulation of the box environment to obtain the optimized characteristic vector of the time series pattern of honey syrup water consumption in the box under the modulation of the box environment; and

[0086] The characteristic vector of the timing pattern of the consumption of syrup water in the box under the optimized in-box environment modulation is input into the decoder-based syrup water flow dynamic optimizer to obtain the optimization instruction.

[0087] That is, the characteristic representation value of the time series mode of honey water consumption in the box under the first box environment modulation and the characteristic representation value of the time series mode of honey water consumption in the box under the second box environment modulation are expressed as:

[0088] α=[log2(n)] -1

[0089] β=exp[(n) -1 ]

[0090] Wherein, n is the number of zero eigenvalues ​​in the characteristic vector of the time series pattern of syrup water consumption in the box under the modulation of the environment in the box, α is the characteristic representation value of the time series pattern of syrup water consumption in the box under the modulation of the first environment in the box, β is the characteristic representation value of the time series pattern of syrup water consumption in the box under the modulation of the second environment in the box, and exp(·) is an exponential function with a natural constant as the base.

[0091] And the characteristic vector of the time series pattern of honey water consumption in the box under the optimized box environment modulation is expressed as:

[0092]

[0093]

[0094]

[0095] Among them, V is the characteristic vector of the time series pattern of honey water consumption in the box under the modulation of the in-box environment, w is the weight hyperparameter, M is the characteristic regression understanding matrix of the time series pattern of honey water consumption in the box under the modulation of the in-box environment, V1 is the characteristic information representation vector of the time series pattern of honey water consumption in the box under the modulation of the in-box environment, n is the number of zero eigenvalues ​​in the characteristic vector of the time series pattern of honey water consumption in the box under the modulation of the in-box environment, α is the characteristic representation value of the time series pattern of honey water consumption in the box under the modulation of the first in-box environment, β is the characteristic representation value of the time series pattern of honey water consumption in the box under the modulation of the second in-box environment, represents a power function that calculates each eigenvalue of the characteristic vector of the time series pattern of honey water consumption in the box under the modulation of the box environment with the inverse of the number of zero eigenvalues ​​as the exponent, T represents the transpose of the vector, ⊙ represents the point multiplication by position, is a matrix multiplication, and V′ is the characteristic vector of the time series pattern of honey water consumption in the box under the modulation of the optimized in-box environment.

[0096] Here, when the implicit correlation feature vector of the time series of syrup water consumption in the box and the implicit correlation feature vector of the time series of temperature-humidity in the box respectively represent the local time domain correlation features of the syrup water consumption in the box and the local time domain collaborative implicit correlation features of temperature-humidity, when performing feature interaction based on feature mapping deviation compensation, the difference in the distribution of source time domain data will cause the feature mapping compensation deviation to be biased, so that the time series pattern feature vector of the syrup water consumption in the box under the modulation of the in-box environment obtained by the interaction has an imbalance in the remote decoding regression understanding representation, thereby deviating from the decoding comprehension of the source time series semantic feature representation, affecting the accuracy of the decoding result.

[0097] Based on this, the present application realizes effective long-range modeling of the characteristic vector of the timing pattern of syrup water consumption in the box under the modulation of the in-box environment under the linear complexity of the field by modeling the vector field zero dimension of the characteristic vector of the timing pattern of syrup water consumption in the box under the modulation of the in-box environment as the global field redundant dependency. In this way, in the feature regression process of the characteristic vector of the timing pattern of syrup water consumption in the box under the modulation of the in-box environment, the structured correlation self-distillation between information representation and class understanding is performed in the selective state space based on the feature distribution, so as to maintain the overall information complexity of the feature distribution of the characteristic vector of the timing pattern of syrup water consumption in the box under the modulation of the in-box environment, promote the long-range dynamic decoding and understanding balance of its feature convergence, so as to improve the decoding and understanding consistency between the decoding regression and the feature extraction process, and improve the accuracy of the optimization instructions obtained by inputting the characteristic vector of the timing pattern of syrup water consumption in the box under the modulation of the in-box environment into the syrup water flow dynamic optimizer based on the decoder.

[0098] In summary, the method of killing stored tobacco powder borer by batch-rearing adult bees of wheat moth braconids according to the embodiment of the present application is explained, which adds honey sugar water to the feeding box of the host larvae of wheat moth braconids for feeding, and uses the data processing technology based on deep learning to perform time series analysis on the temperature, humidity and honey sugar water level in the feeding box, so as to extract the time series consumption trend of the honey sugar water level and the time series correlation mode of the temperature and humidity in the box, and then, through the interactive analysis of the time series correlation characteristics of the temperature and humidity in the box and the time series consumption characteristics of honey sugar water, to reveal the dynamic correlation relationship between honey sugar water consumption and the environmental conditions in the box, and thereby realize the intelligent recommendation of honey sugar water flow under the current environmental conditions. In this way, the optimized control of honey sugar water supply can be achieved, the feeding conditions of adult bees of wheat moth braconids can be improved, thereby improving the survival rate and adult quality of wheat moth braconids, and enhancing the effect of biological control.

[0099] The basic principle of the present invention is described above in conjunction with specific embodiments. However, it should be pointed out that the advantages, strengths, effects, etc. mentioned in the present invention are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. must be possessed by each embodiment of the present invention. In addition, the specific details of the above embodiments are only for the purpose of illustration and facilitation of understanding, rather than limitation, and the above details do not limit the present invention to being implemented by adopting the above specific details.

[0100] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, please refer to the relevant description of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the module division is only a logical function division, and there may be other division methods in actual implementation. The modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0101] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0102] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference to a figure in a claim should not be considered as limiting the claim to which it relates.

[0103] In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units stated in the device claim can also be implemented by one unit through software or hardware.

[0104] Finally, it should be noted that the above description has been given for the purpose of illustration and description. In addition, the above embodiments are only used to illustrate the technical solution of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A method for batch rearing adult bees of Psoralea spp. to kill stored tobacco borer, comprising: The host larvae of the meal moth braconid wasp are placed in a first rearing box, and pairs of first-generation meal moth braconid wasp adult bees are introduced into the first rearing box, and honey water with a concentration of 10-70% is added to the first rearing box every day, and the first rearing box is placed in an environment with a temperature of 27-29°C, a relative humidity of 70-80% and a photoperiod of 16L, and the second generation of meal moth braconid wasp can be obtained after the first generation of meal moth braconid wasp parasitizes and lays eggs on the host larvae; after the second generation of meal moth braconid wasp eggs finally emerge into adult bees through eggs, larvae, and pupae, the second rearing box is transferred to The method comprises the steps of: heating the adult bees to a temperature of 2-6°C until the adult bees are anesthetized by low temperature and cannot move, and transferring the adult bees of the lycopodiella from the first feeding box to the transport box for loading; transporting the transport box to a warehouse where tobacco leaves are stored, and transferring the adult bees of the lycopodiella in the transport box to a bee release device, and when tobacco moth borer larvae appear on the tobacco bales, the bee release device can be opened to release the bees, and the adult bees of the lycopodiella parasitize and kill the tobacco moth borer on the stored tobacco leaves, characterized in that honey water with a concentration of 10-70% is added to the first feeding box every day, comprising: Obtaining the time queue of the temperature in the box and the time queue of the humidity in the box collected by the temperature sensor and the humidity sensor, and obtaining the time queue of the honey water level value in the box collected by the liquid level sensor; Performing a time series consumption analysis on the time queue of the honey syrup water level value in the box to obtain an implicit correlation feature vector of the time series of the honey syrup water consumption in the box; Performing time series correlation analysis on the time series of the temperature inside the box and the time series of the humidity inside the box to obtain an implicit feature vector of the time series correlation between the temperature inside the box and the humidity inside the box; Interactively fusing the implicit correlation feature vector of the honey water consumption time series in the box and the implicit correlation feature vector of the temperature-humidity time series in the box based on feature mapping deviation compensation to obtain the feature vector of the honey water consumption time series pattern under the modulation of the box environment; Based on the characteristic vector of the time series pattern of honey syrup water consumption in the box under the modulation of the environment in the box, an optimization instruction for the flow rate of honey syrup water is generated.

2. The method for killing stored tobacco borer by raising adult bees of Lycopodiella spp. in batches according to claim 1, characterized in that: Performing a time series consumption analysis on the time queue of the honey syrup water level value in the box to obtain an implicit correlation feature vector of the honey syrup water consumption time series in the box, including: Calculate the difference between the values ​​of the honey syrup water level in the box at each two adjacent predetermined time points in the time queue of the honey syrup water level in the box to obtain a time queue representing the consumption of the honey syrup water in the box; The time sequence representing the consumption of honey water in the box is input into a sequence encoder based on a forward LSTM model to obtain a temporal implicit correlation feature vector of the consumption of honey water in the box.

3. The method for killing stored tobacco borer by raising adult bees of Lycopodiella spp. in batches according to claim 2, characterized in that: Performing a time series correlation analysis on the time series of the temperature inside the box and the time series of the humidity inside the box to obtain an implicit feature vector of the time series correlation of the temperature and humidity inside the box, including: After arranging the time queue of the temperature inside the box and the time queue of the humidity inside the box into a time series input vector of the temperature inside the box and a time series input vector of the humidity inside the box according to the time dimension, calculating the association matrix between the time series input vector of the temperature inside the box and the time series input vector of the humidity inside the box to obtain a time series association matrix of the temperature-humidity inside the box; The in-box temperature-humidity temporal association matrix is ​​input into the in-box environment temporal pattern feature extractor based on the hole convolutional neural network model to obtain the in-box temperature-humidity temporal association implicit feature vector.

4. The method for killing stored tobacco borer by raising adult bees of Lycopodiella spp. in batches according to claim 3, characterized in that: The implicit correlation feature vector of the honey water consumption time series in the box and the implicit correlation feature vector of the temperature-humidity time series in the box are interactively fused based on feature mapping deviation compensation to obtain the time series pattern feature vector of the honey water consumption in the box under the modulation of the box environment, including: Performing co-space mapping modulation on the honey water consumption time series implicit correlation feature vector in the box and the temperature-humidity time series implicit correlation feature vector in the box to obtain the modulated honey water consumption time series implicit correlation feature vector in the box and the modulated temperature-humidity time series implicit correlation feature vector in the box; Based on the feature mapping deviation of the implicit correlation feature vector of the time series of syrup water consumption in the box after modulation and the implicit correlation feature vector of the time series of temperature-humidity in the box after modulation relative to the implicit correlation feature vector of the time series of syrup water consumption in the box and the implicit correlation feature vector of the time series of temperature-humidity in the box, the implicit correlation feature vector of the time series of syrup water consumption in the box after modulation and the implicit correlation feature vector of the time series of temperature-humidity in the box after modulation are interactively fused in a compensatory manner to obtain the time series pattern feature vector of syrup water consumption in the box under the modulation of the in-box environment.

5. The method for killing stored tobacco borer by raising adult bees of Lycopodiella spp. in batches according to claim 4, characterized in that: The honey water consumption time series implicit correlation feature vector in the box and the temperature-humidity time series implicit correlation feature vector in the box are subjected to co-space mapping modulation to obtain the honey water consumption time series implicit correlation feature vector in the box and the temperature-humidity time series implicit correlation feature vector in the box after modulation, including: The implicit correlation feature vector of the time series of syrup water consumption in the box and the implicit correlation feature vector of the time series of temperature-humidity in the box are input into a co-space mapper based on a fully connected layer to obtain the implicit correlation feature vector of the time series of syrup water consumption in the box after modulation and the implicit correlation feature vector of the time series of temperature-humidity in the box after modulation, wherein the implicit correlation feature vector of the time series of syrup water consumption in the box after modulation and the implicit correlation feature vector of the time series of temperature-humidity in the box after modulation have the same characteristic scale.

6. The method for killing stored tobacco borer by raising adult bees of Lycopodiella spp. in batches according to claim 5, characterized in that: Based on the feature mapping deviation of the implicit correlation feature vector of the time series of syrup water consumption in the box after modulation and the implicit correlation feature vector of the time series of temperature-humidity in the box after modulation relative to the implicit correlation feature vector of the time series of syrup water consumption in the box and the implicit correlation feature vector of the time series of temperature-humidity in the box, the implicit correlation feature vector of the time series of syrup water consumption in the box after modulation and the implicit correlation feature vector of the time series of temperature-humidity in the box after modulation are compensated interactively fused to obtain the time series pattern feature vector of syrup water consumption in the box under the modulation of the box environment, including: Performing interpolation processing on the implicit correlation feature vector of the honey water consumption time series in the box and the implicit correlation feature vector of the temperature-humidity time series in the box to obtain the implicit correlation feature vector of the honey water consumption time series in the box after interpolation processing and the implicit correlation feature vector of the temperature-humidity time series in the box after interpolation processing, wherein the implicit correlation feature vector of the honey water consumption time series in the box after interpolation processing, the implicit correlation feature vector of the temperature-humidity time series in the box after interpolation processing, the implicit correlation feature vector of the honey water consumption time series in the box after modulation and the implicit correlation feature vector of the temperature-humidity time series in the box after modulation have the same characteristic scale; Input the implicit correlation feature vector of the honey water consumption time series after the interpolation processing and the implicit correlation feature vector of the honey water consumption time series after the modulation, the implicit correlation feature vector of the temperature-humidity time series after the interpolation processing and the implicit correlation feature vector of the temperature-humidity time series after the modulation into the feature mapping deviation measurement network respectively to obtain the honey water consumption time series feature mapping deviation compensation factor and the temperature-humidity time series correlation feature mapping deviation compensation factor; Based on the honey water consumption time series feature mapping deviation compensation factor and the box temperature-humidity time series correlation feature mapping deviation compensation factor, the modulation box honey water consumption time series implicit correlation feature vector and the modulation box temperature-humidity time series implicit correlation feature vector are compensated to obtain the compensated box honey water consumption time series implicit correlation feature vector and the compensated box temperature-humidity time series implicit correlation feature vector; The implicit correlation feature vector of the time series of honey water consumption in the box after compensation and the implicit correlation feature vector of the time series of temperature-humidity in the box after compensation are input into the multi-dimensional significant interaction network between features to obtain the time series pattern feature vector of honey water consumption in the box under the modulation of the box environment.

7. The method for killing stored tobacco borer by raising adult bees of Lycopodiella spp. in batches according to claim 6, characterized in that: Inputting the interpolated implicit correlation feature vector of the honey syrup water consumption time series in the box and the modulated implicit correlation feature vector of the honey syrup water consumption time series in the box into a feature mapping deviation measurement network to obtain a honey syrup water consumption time series feature mapping deviation compensation factor, including: Calculate the positional dotted division vector between the time series implicitly associated feature vector of the honey syrup water consumption in the box after modulation and the time series implicitly associated feature vector of the honey syrup water consumption in the box after interpolation processing, and calculate the base-two logarithmic function value of the absolute value of each feature value in the dotted division vector to obtain a honey syrup water consumption time series feature mapping deviation representation vector; Calculate the dot product vector between the honey water consumption timing feature map deviation representation vector and the honey water consumption timing implicit correlation feature vector in the modulated box, and calculate the exponential function value with the natural constant as the base and the sum of the eigenvalues ​​of the dot product vector as the exponent to obtain the honey water consumption timing feature map deviation compensation factor.

8. The method for killing stored tobacco borer by raising adult bees of Lycopodiella spp. in batches according to claim 7, characterized in that: Inputting the compensated honey water consumption time series implicit correlation feature vector and the compensated temperature-humidity time series implicit correlation feature vector into the multi-dimensional significant interaction network between features to obtain the honey water consumption time series pattern feature vector under the modulation of the box environment, including: Respectively calculating the position point subtraction, position point multiplication, position point addition and position point division between the compensated in-box syrup water consumption time series implicit correlation feature vector and the compensated in-box temperature-humidity time series implicit correlation feature vector to obtain a first in-box environment-syrup water consumption time series interaction representation vector, a second in-box environment-syrup water consumption time series interaction representation vector, a third in-box environment-syrup water consumption time series interaction representation vector and a fourth in-box environment-syrup water consumption time series interaction representation vector; The first in-box environment-honey sugar water consumption timing interaction representation vector, the second in-box environment-honey sugar water consumption timing interaction representation vector, the third in-box environment-honey sugar water consumption timing interaction representation vector and the fourth in-box environment-honey sugar water consumption timing interaction representation vector are cascaded and fused, and then input into a one-dimensional convolutional layer including a maximum pooling layer to obtain a characteristic vector of the in-box honey sugar water consumption timing pattern under the modulation of the in-box environment.

9. The method for killing stored tobacco borer by raising adult bees of Lycopodiella spp. in batches according to claim 8, characterized in that: Based on the characteristic vector of the time series pattern of honey syrup water consumption in the box under the modulation of the box environment, an optimization instruction for the flow rate of honey syrup water is generated, including: The characteristic vector of the time series pattern of the syrup water consumption in the box under the modulation of the box environment is input into the syrup water flow dynamic optimizer based on the decoder to obtain the optimization instruction, and the optimization instruction includes the recommended flow value of the syrup water.

Images