Irrigation prediction method, device, equipment and medium

Abstract

The invention discloses an irrigation prediction method and device, equipment and a medium. The method comprises the following steps: acquiring an initial noise action sequence; when the obtained current iteration step length number is smaller than or equal to the total number of model iteration step lengths, determining a current noise action sequence corresponding to the current iteration step length number; determining current supplementary data corresponding to the current iteration step length; inputting the current supplementary data and the current noise action sequence into an irrigation prediction model, and determining a predicted noise action sequence corresponding to the current iteration step length; when the current iteration step length number is greater than the total number of model iteration step lengths, obtaining a predicted action sequence corresponding to the current iteration step length number; and in response to the fact that the current iteration step length number is smaller than or equal to the total number of model iteration step lengths, taking the predicted action sequence corresponding to the current iteration step length number as a current noise action sequence corresponding to the next iteration step length number, and accumulating 1 on the current iteration step length number. According to the embodiment of the invention, the irrigation prediction accuracy can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular, to an irrigation prediction method, device, equipment and medium. Background Art

[0002] With the rapid development of agriculture, the types of crops have gradually increased. To ensure the normal growth of crops, growers need to irrigate various crops in a timely manner.

[0003] Currently, irrigation equipment is manually operated to irrigate various crops at regular intervals and in fixed quantities.

[0004] However, when manually operating irrigation equipment to irrigate crops, the irrigation water volume or irrigation time may not accurately meet the growth needs of the crops, and the quality of the crops cannot be guaranteed. Summary of the Invention

[0005] The present invention provides an irrigation prediction method, device, equipment and medium to improve the accuracy of irrigation prediction.

[0006] In a first aspect, an embodiment of the present invention provides an irrigation prediction method, which includes:

[0007] Obtain an initial noise action sequence;

[0008] When the currently obtained iteration step number is less than or equal to the total number of model iteration step numbers, and based on the initial noise action sequence, determine the current noise action sequence corresponding to the currently obtained iteration step number;

[0009] Determine the current supplementary data corresponding to the currently obtained iteration step number according to the parity of the currently obtained iteration step number;

[0010] Input the current supplementary data and the current noise action sequence into the irrigation prediction model to determine the predicted noise action sequence corresponding to the currently obtained iteration step number;

[0011] When the currently obtained iteration step number is greater than the total number of model iteration step numbers, obtain the predicted action sequence corresponding to the currently obtained iteration step number and determine the predicted irrigation data;

[0012] In response to the currently obtained iteration step number being less than or equal to the total number of model iteration step numbers, use the predicted action sequence corresponding to the currently obtained iteration step number as the current noise action sequence corresponding to the next iteration step number, and increment the currently obtained iteration step number by 1.

[0013] In a second aspect, an embodiment of the present invention further provides an irrigation prediction device, which includes:

[0014] A sequence acquisition module, configured to obtain an initial noise action sequence;

[0015] A current sequence acquisition module, configured to determine a current noise action sequence corresponding to the current iteration step number according to the initial noise action sequence when the acquired current iteration step number is less than or equal to the total number of model iteration steps;

[0016] A data acquisition module, configured to determine current supplementary data corresponding to the current iteration step number according to the parity of the current iteration step number;

[0017] A data input module, configured to input the current supplementary data and the current noise action sequence into an irrigation prediction model to determine a predicted noise action sequence corresponding to the current iteration step number;

[0018] A predicted data acquisition module, configured to acquire a predicted action sequence corresponding to the current iteration step number and determine predicted irrigation data when the current iteration step number is greater than the total number of model iteration steps;

[0019] A step accumulation module, configured to, in response to the current iteration step number being less than or equal to the total number of model iteration steps, use the predicted action sequence corresponding to the current iteration step number as the current noise action sequence corresponding to the next iteration step number, and add 1 to the current iteration step number.

[0020] In a third aspect, an embodiment of the present invention further provides an irrigation prediction device, where the irrigation prediction device includes:

[0021] At least one processor; and

[0022] A memory communicatively connected to the at least one processor; wherein,

[0023] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the irrigation prediction method of any embodiment of the present invention.

[0024] According to another aspect of the present invention, there is provided a computer-readable storage medium storing computer instructions for causing a processor to implement the irrigation prediction method of any embodiment of the present invention when executed.

[0025] The technical solution of the embodiment of the present invention includes: obtaining an initial noise action sequence; when the obtained current iteration step number is less than or equal to the total number of model iteration steps, determining the current noise action sequence corresponding to the current iteration step number according to the initial noise action sequence; determining the current supplementary data corresponding to the current iteration step number according to the parity of the current iteration step number; inputting the current supplementary data and the current noise action sequence into an irrigation prediction model to determine the predicted noise action sequence corresponding to the current iteration step number; when the current iteration step number is greater than the total number of model iteration steps, obtaining the predicted action sequence corresponding to the current iteration step number and determining the predicted irrigation data; in response to the current iteration step number being less than or equal to the total number of model iteration steps, using the predicted action sequence corresponding to the current iteration step number as the current noise action sequence corresponding to the next iteration step number, and incrementing the current iteration step number by 1. By judging the parity of the current iteration step number, different data can be alternately used as conditions to denoise the initial noise action sequence, avoiding the mutual influence of different conditions corresponding to different iteration step numbers during the denoising process of the initial noise action sequence, and improving the accuracy of denoising the initial noise action sequence.

[0026] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present invention, nor is it used to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0028] Figure 1 is a flowchart of an irrigation prediction method provided according to an embodiment of the present invention;

[0029] Figure 2 is a flowchart of an irrigation prediction method provided according to an embodiment of the present invention;

[0030] Figure 3 is a flowchart of an irrigation prediction method provided according to an embodiment of the present invention;

[0031] Figure 4 is a structural diagram of an irrigation prediction device provided according to an embodiment of the present invention;

[0032] Figure 5 is a structural schematic diagram of an irrigation prediction device provided according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0033] In order to enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0034] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily need to be used to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments of the present invention described here can be implemented in an order other than those illustrated or described here. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0035] In the technical solution of the embodiment of the present invention, the acquisition, storage, application, etc. of the initial noise action sequence and the like all comply with the provisions of relevant laws and regulations and do not violate public order and good customs.

[0036] Embodiment 1

[0037] Figure 1 It is a flowchart of an irrigation prediction method provided for Embodiment 1 of the present invention. The embodiment of the present invention is applicable to the situation of irrigation prediction. This method can be executed by an irrigation prediction device, and the irrigation prediction device can be implemented in the form of hardware and / or software.

[0038] See Figure 1 The irrigation prediction method shown includes:

[0039] S101. Obtain an initial noise action sequence.

[0040] Among them, the initial noise action sequence can be used to describe a data set representing a preset irrigation action instruction.

[0041] Specifically, crop roots are distributed at different soil depths. Shallow roots absorb available water, while deep roots absorb water that gradually seeps into the deep soil. If only surface irrigation is relied on, it is impossible to accurately understand whether the moisture in the deep soil meets the crop's requirements, which may lead to restricted crop growth or reduced stress resistance. Therefore, fine and precise monitoring and irrigation adjustment of soil humidity at different depths are carried out to meet the water requirements of crops at each growth stage. The installation depth position of irrigation nozzles in the soil can be preset. An irrigation device may include at least one nozzle, and a temperature sensor and a humidity sensor are installed at each nozzle position. Each nozzle can perform a water spraying operation.

[0042] Irrigation data required for performing an irrigation operation on crops can be obtained. The irrigation data includes at least one parameter item related to the irrigation operation. By way of example, the parameter items related to the irrigation operation can be nozzle identifier, current time, the burial depth of the nozzle corresponding to the nozzle identifier in the soil, the humidity to be achieved in the soil, the temperature to be achieved in the soil, the start time of water spraying, and the water spraying duration, etc. The parameter items in the irrigation data can be obtained through the experience of industry experts. According to at least one parameter item in the irrigation data, parameter values corresponding to the parameter items can be randomly generated, and based on the parameter values corresponding to the parameter items, an initial noise action sequence is determined. The initial noise sequence can be input into the model, and the initial noise sequence is modulated by different feature data sets as conditional tokens of the model to generate a desired action sequence. The model can be a DiffusionTransformer (DiT) model.

[0043] S102. When the obtained current iteration step number is less than or equal to the total number of model iteration steps, and based on the initial noise action sequence, determine the current noise action sequence corresponding to the current iteration step number.

[0044] Among them, the current iteration step number can be used to describe the number of times of the current iteration to be performed on the initial noise action sequence at the current time. The total number of model iteration steps can be used to describe the total number of iterations to be performed on the initial noise action sequence currently.

[0045] Specifically, when the obtained current iteration step number is less than or equal to the total number of model iteration steps, it indicates that the iteration end condition has not been reached and the denoising process needs to be continued. Therefore, based on the initial noise action sequence, determine the noise action sequence after denoising the initial noise action sequence in the previous iteration round. Determining the noise action sequence after denoising in the previous iteration round as the current noise action sequence corresponding to the current iteration step number can facilitate subsequent denoising operations for the current noise sequence in the current iteration step number corresponding to this round.

[0046] S103. Determine the current supplementary data corresponding to the current iteration step number according to the parity of the current iteration step number.

[0047] Among them, the current supplementary data can be used to describe the modulation data for denoising the initial noise action sequence.

[0048] Specifically, determine the current supplementary data corresponding to the current iteration step number according to the parity of the current iteration step number. The current supplementary data when the current iteration step number is odd is different from the current supplementary data when the current iteration step number is even. Judge the parity of the current iteration step number in ascending order of the current iteration step number, and determine the current supplementary data corresponding to the current iteration step number according to the judgment result. After the iterative operation corresponding to the current iteration step number is completed, the next round of iterative operation is carried out.

[0049] S104. Input the current supplementary data and the current noise action sequence into the irrigation prediction model to determine the predicted noise action sequence corresponding to the current iteration step number.

[0050] Among them, the irrigation prediction model can be used to describe the model for predicting the noise action sequence.

[0051] Specifically, input the current supplementary data and the current noise action sequence into the irrigation prediction model, use the current supplementary data as the tuning condition to modulate the current noise action sequence, and determine the predicted noise action sequence corresponding to the current iteration step number. The predicted noise action sequence corresponding to the current iteration step number can be used as the current noise action sequence to be iteratively processed in the next round.

[0052] S105. When the current iteration step number is greater than the total value of the model iteration step numbers, obtain the predicted action sequence corresponding to the current iteration step number and determine the predicted irrigation data.

[0053] Among them, the predicted irrigation data can be used to describe the operations that the irrigation equipment needs to perform.

[0054] Specifically, when the current iteration step number is greater than the total value of the model iteration step numbers, there is no need to continue denoising the initial noise action sequence, and the predicted action sequence corresponding to the current iteration step number can be obtained, that is, the predicted action sequence obtained after the last iterative denoising operation. Identify the data features of the predicted action sequence, and determine the predicted irrigation data according to the data features of the predicted action sequence.

[0055] S106. In response to the current iteration step number being less than or equal to the total value of the model iteration step numbers, use the predicted action sequence corresponding to the current iteration step number as the current noise action sequence corresponding to the next iteration step number, and add 1 to the current iteration step number.

[0056] Specifically, in response to the current iteration step number being less than or equal to the total number of model iteration steps, it indicates that the end condition for ending the iteration has not been reached, and the prediction action sequence corresponding to the current iteration step number is still required for denoising. Then, the prediction action sequence corresponding to the current iteration step number is used as the current noise action sequence corresponding to the next iteration step number. Since the iteration operations are completed one by one in sequence, therefore, 1 is added to the current iteration step number, and the parity of the value of the current iteration step number after accumulation is judged, so as to determine the supplementary data corresponding to the current iteration step number after accumulation. Using the supplementary data as the modulation condition data, the prediction action sequence corresponding to the current iteration step number before accumulation is denoised until the current iteration step number is less than or equal to the total number of model iteration steps, and the condition for ending the iteration is reached.

[0057] The technical solution of the embodiment of the present invention is as follows: by obtaining an initial noise action sequence; when the obtained current iteration step number is less than or equal to the total number of model iteration steps, and according to the initial noise action sequence, determining the current noise action sequence corresponding to the current iteration step number; according to the parity of the current iteration step number, determining the current supplementary data corresponding to the current iteration step number; inputting the current supplementary data and the current noise action sequence into the irrigation prediction model to determine the predicted noise action sequence corresponding to the current iteration step number; when the current iteration step number is greater than the total number of model iteration steps, obtaining the prediction action sequence corresponding to the current iteration step number and determining the predicted irrigation data; in response to the current iteration step number being less than or equal to the total number of model iteration steps, using the prediction action sequence corresponding to the current iteration step number as the current noise action sequence corresponding to the next iteration step number, and adding 1 to the current iteration step number. By judging the parity of the current iteration step number, different data can be alternately used as conditions to denoise the initial noise action sequence, avoiding the mutual influence of different conditions corresponding to different iteration step numbers during the denoising process of the initial noise action sequence, and improving the accuracy of denoising the initial noise action sequence.

[0058] Embodiment 2

[0059] Figure 2 It is a flowchart of an irrigation prediction method provided by Embodiment 2 of the present invention. On the basis of the above embodiment, the irrigation prediction operation is optimized and improved in this embodiment of the present invention.

[0060] Further, refining "determining the current supplementary data corresponding to the current iteration step number according to the parity of the current iteration step number" to "obtaining the historical irrigation action sequence; obtaining the updated spatial feature distribution information and the updated temporal feature distribution information; if the current iteration step number is odd, determining the updated spatial feature distribution information and the updated temporal feature distribution information as the current supplementary data corresponding to the current iteration step number; if the current iteration step number is even, determining the historical irrigation action sequence as the current supplementary data corresponding to the current iteration step number" to improve the operation of irrigation prediction.

[0061] It should be noted that for parts not detailed in the embodiments of the present invention, reference may be made to the descriptions of other embodiments.

[0062] See Figure 2 The irrigation prediction method shown includes:

[0063] S201. Obtain the initial noise action sequence.

[0064] S202. When the obtained current iteration step number is less than or equal to the total value of the model iteration step number, and based on the initial noise action sequence, determine the current noise action sequence corresponding to the current iteration step number.

[0065] S203. Obtain the historical irrigation action sequence.

[0066] Among them, the historical irrigation action sequence can be used to describe the data corresponding to the irrigation operations of the irrigation device in the past time period.

[0067] Specifically, obtain the preset historical time period, and collect the irrigation data corresponding to the irrigation device within the historical time period according to the historical time period. The irrigation data may be the sprinkler identifier, the current time, the burial depth of the sprinkler corresponding to the sprinkler identifier in the soil, the humidity that the soil needs to reach, the temperature that the soil needs to reach, the start time of water spraying, and the water spraying duration. Through conversion of the irrigation data within the historical time period, the historical irrigation action sequence is obtained. The obtaining method includes but is not limited to: reading the records of the irrigation management system, collecting through sensors, or obtaining device logs or monitoring records, etc. The embodiments of the present invention do not limit this.

[0068] S204. Obtain the updated spatial feature distribution information and the updated temporal feature distribution information.

[0069] Among them, the updated spatial feature distribution information can be used to describe the set of spatial feature data obtained at the current time. The updated temporal feature distribution information can be used to describe the set of temporal feature data obtained at the current time.

[0070] Specifically, updating the spatial feature distribution information (spatial global token) is a special token that globally represents the model input or features in the spatial dimension. Updating the temporal feature distribution information (temporal global token) is a special token that globally represents the model input or features in the temporal dimension. The updated spatial feature distribution information can be obtained by updating the spatial feature distribution information. The updated temporal feature distribution information can be obtained by updating the temporal feature distribution information. The spatial feature distribution information or the temporal feature distribution information can be subjected to cross-attention calculation. Through the learned cross-attention mechanism, key information in the spatial distribution and temporal distribution can be selectively extracted and condensed, focusing on important regions and features therein to improve the accuracy of decision-making. Therefore, update cross-attention calculation is performed on the initial spatial feature distribution information or temporal feature distribution information to obtain the updated spatial feature distribution information and the updated temporal feature distribution information.

[0071] S205. If the current iteration step number is odd, determine the updated spatial feature distribution information and the updated temporal feature distribution information as the current supplementary data corresponding to the current iteration step number.

[0072] Specifically, if the current iteration step number is odd, determine the updated spatial feature distribution information and the updated temporal feature distribution information as the current supplementary data corresponding to the current iteration step number. Through different conditional tokens, the model can incorporate various external information, thereby better capturing complex relationships in the data, reasonably adjusting the generation strategy, controlling the progress of the generation process, and using the updated spatial feature distribution information and the updated temporal feature distribution information as conditions to input into the model. By way of example, the model can be a DiT model.

[0073] S206. If the current iteration step number is even, determine the historical irrigation action sequence as the current supplementary data corresponding to the current iteration step number.

[0074] Specifically, if the current iteration step number is even, determine the historical irrigation action sequence as the current supplementary data corresponding to the current iteration step number. Use the historical irrigation action sequence as conditional features to input into the model and modulate the current noise action sequence corresponding to the current iteration step number. By way of example, the model can be a DiT model.

[0075] S207. Input the current supplementary data and the current noise action sequence into the irrigation prediction model to determine the predicted noise action sequence corresponding to the current iteration step number.

[0076] S208. When the current iteration step number is greater than the total number of model iteration steps, obtain the predicted action sequence corresponding to the current iteration step number and determine the predicted irrigation data;

[0077] S209. In response to the current iteration step number being less than or equal to the total number of model iteration steps, use the predicted action sequence corresponding to the current iteration step number as the current noise action sequence corresponding to the next iteration step number, and increment the current iteration step number by 1.

[0078] In the embodiment of the present invention, by obtaining the historical irrigation action sequence; obtaining the updated spatial feature distribution information and the updated temporal feature distribution information; if the current iteration step number is odd, determining the updated spatial feature distribution information and the updated temporal feature distribution information as the current supplementary data corresponding to the current iteration step number; if the current iteration step number is even, determining the historical irrigation action sequence as the current supplementary data corresponding to the current iteration step number, and by alternately using the updated spatial feature distribution information, the updated temporal feature distribution information, and the historical irrigation action sequence as conditions in different steps, it avoids their mutual interference during the learning process, improves the learning effect of the model on spatio-temporal features while ensuring the model's acquisition of features of previous actions, and improves the accuracy of irrigation prediction.

[0079] Optionally, obtaining the updated spatial feature distribution information and the updated temporal feature distribution information includes: obtaining the spatial feature distribution information and the temporal feature distribution information; obtaining the update matrix corresponding to the spatial feature distribution information and the temporal feature distribution information; performing cross-attention calculation on the spatial feature distribution information and the update matrix to update the spatial feature distribution information to obtain the updated spatial feature distribution information; performing cross-attention calculation on the temporal feature distribution information and the update matrix to update the temporal feature distribution information to obtain the updated temporal feature distribution information.

[0080] Among them, the spatial feature distribution information can be used to describe the data set of the initial spatial feature information of the model when no iterative operation is performed. The temporal feature distribution information can be used to describe the data set of the initial temporal feature information of the model when no iterative operation is performed.

[0081] Specifically, obtain the spatial feature distribution information and temporal feature distribution information of the model. For example, the model can be a DiT model, and the spatial feature distribution information and temporal feature distribution information can be generated through an initialization method, where the initialization method includes but is not limited to random initialization or preset in advance, and the embodiments of the present invention do not limit this. The spatial feature distribution information can be processed by expanding it in the time dimension to obtain an update matrix corresponding to the spatial feature distribution information. The temporal feature distribution information can be processed by expanding it in the spatial dimension to obtain an update matrix corresponding to the temporal feature distribution information. After obtaining the update matrices corresponding to the spatial feature distribution information and the temporal feature distribution information, perform cross-attention calculation on the spatial feature distribution information and the update matrix to obtain a spatial cross result, replace the spatial feature distribution information with the spatial cross result, update the spatial feature distribution information, and obtain updated spatial feature distribution information, and determine spatial feature data. At this time, the data of the updated spatial feature distribution information is the same as that of the spatial cross result. Perform cross-attention calculation on the temporal feature distribution information and the update matrix to obtain a temporal cross result, replace the temporal feature distribution information with the temporal cross result, update the temporal feature distribution information, and obtain updated temporal feature distribution information, and determine temporal feature data. At this time, the data of the updated temporal feature distribution information is the same as that of the temporal cross result.

[0082] By obtaining the spatial feature distribution information and the temporal feature distribution information; obtaining the update matrices corresponding to the spatial feature distribution information and the temporal feature distribution information; performing cross-attention calculation on the spatial feature distribution information and the update matrix to update the spatial feature distribution information and obtain updated spatial feature distribution information; performing cross-attention calculation on the temporal feature distribution information and the update matrix to update the temporal feature distribution information and obtain updated temporal feature distribution information, the spatial feature extraction enables the model to capture the spatial distribution characteristics of soil moisture at different depths. The temporal feature extraction enables the model to capture the dynamic characteristics of the change of soil moisture over time, facilitating the model to identify irrigation decisions based on spatio-temporal features to obtain more accurate irrigation decisions.

[0083] Optionally, obtaining the update matrices corresponding to the spatial feature distribution information and the temporal feature distribution information includes: obtaining historical collection data, where the historical collection data includes: collection time period, collection depth corresponding to the collection time period, collection temperature, collection humidity, and historical irrigation action sequence; obtaining the feature matrix corresponding to the historical depth and historical temperature in the historical collection data; processing the feature matrix corresponding to the historical depth and historical temperature in the historical collection data by expanding it in the time dimension to obtain an update matrix corresponding to the spatial feature distribution information; processing the feature matrix corresponding to the historical depth and historical temperature in the historical collection data by expanding it in the spatial dimension to obtain an update matrix corresponding to the temporal feature distribution information.

[0084] Among them, the feature matrix can be used to describe the feature matrix of the acquisition information distribution determined according to the historical acquisition data.

[0085] Specifically, a plurality of humidity sensors and temperature sensors are arranged at different depths of the soil. With the accumulation of sensors over time, historical acquisition data is obtained. The historical acquisition data includes: acquisition time period, acquisition depth corresponding to the acquisition time period, acquisition temperature, acquisition humidity, and historical irrigation action sequence, forming a data matrix. The real-time monitoring of soil humidity at different depths is realized, comprehensive soil moisture information is obtained, and high-quality data support is provided for subsequent precise irrigation decision-making. Obtain the feature matrix corresponding to the historical depth and historical temperature in the historical acquisition data; expand and process the feature matrix corresponding to the historical depth and historical temperature in the historical acquisition data in the time dimension to obtain an update matrix corresponding to the spatial feature distribution information; expand and process the feature matrix corresponding to the historical depth and historical temperature in the historical acquisition data in the spatial dimension to obtain an update matrix corresponding to the time feature distribution information, which is convenient for obtaining the time feature information and spatial feature information of the historical acquisition data.

[0086] By obtaining historical acquisition data, the historical acquisition data includes: acquisition time period, acquisition depth corresponding to the acquisition time period, acquisition temperature, acquisition humidity, and historical irrigation action sequence; obtaining the feature matrix corresponding to the historical depth and historical temperature in the historical acquisition data; expanding and processing the feature matrix corresponding to the historical depth and historical temperature in the historical acquisition data in the time dimension to obtain an update matrix corresponding to the spatial feature distribution information; expanding and processing the feature matrix corresponding to the historical depth and historical temperature in the historical acquisition data in the spatial dimension to obtain an update matrix corresponding to the time feature distribution information, the global features of the data in time and space can be extracted according to the historical acquisition data, and the accuracy of the irrigation decision-making can be improved.

[0087] Optionally, obtaining the feature matrix corresponding to the historical depth and historical temperature in the historical acquisition data includes: performing high-dimensional mapping processing on the historical acquisition data to obtain a mapping matrix corresponding to the historical depth and historical temperature in the historical acquisition data; obtaining spatial position encoding and time position encoding; fusing the spatial position encoding and the time position encoding with the mapping matrix to obtain the feature matrix corresponding to the historical depth and historical temperature.

[0088] Specifically, perform high-dimensional mapping processing on the historical acquisition data, map the data matrix to a high-dimensional feature space through the embedding layer of the model to obtain a mapping matrix corresponding to the historical depth and historical temperature in the historical acquisition data; obtain spatial position encoding and time position encoding, where the spatial position encoding represents the depth position information of each sensor on the irrigation device, and the time position encoding represents the time step information; add the spatial position encoding, the time position encoding and the mapping matrix to perform data fusion to obtain the feature matrix corresponding to the historical depth and historical temperature.

[0089] By performing high-dimensional mapping processing on historical acquisition data, a mapping matrix corresponding to the historical depth and historical temperature in the historical acquisition data is obtained; a spatial position encoding and a temporal position encoding are acquired; the spatial position encoding and the temporal position encoding are fused with the mapping matrix to obtain a feature matrix corresponding to the historical depth and historical temperature, which can accurately determine the depth, temperature, humidity, and temporal features of different sensors. Through data analysis using multi-dimensional data, the accuracy of irrigation prediction is improved.

[0090] Optionally, obtaining a predicted action sequence corresponding to the current iteration step number and determining predicted irrigation data includes: obtaining a target decoder trained on the initial decoder according to historical acquisition data; determining the correspondence between the predicted action sequence and irrigation operations according to the target decoder; and determining the predicted irrigation data according to the correspondence between the predicted action sequence and irrigation operations.

[0091] Among them, the target decoder can be used to describe the network layer for predicting results from feature data.

[0092] Specifically, training the initial decoder according to historical acquisition data to generate the correspondence between different feature data and irrigation operation times, obtaining a target decoder trained on the initial decoder. Obtaining the target decoder, according to the target decoder, the correspondence between each feature data in the predicted action sequence and irrigation operations can be determined, and according to the correspondence between the predicted action sequence and irrigation operations, the predicted irrigation data is determined. The predicted irrigation data can be used to describe the set of data of the operations to be performed by the irrigation device.

[0093] By obtaining a target decoder trained on the initial decoder according to historical acquisition data; determining the correspondence between the predicted action sequence and irrigation operations according to the target decoder; and determining the predicted irrigation data according to the correspondence between the predicted action sequence and irrigation operations, the corresponding feature change law between the predicted action sequence and irrigation operations can be determined through historical acquisition data, which is convenient for verifying and improving the model and improves the prediction accuracy of the model.

[0094] Optionally, after obtaining the target decoder trained on the initial decoder according to historical acquisition data, it further includes: obtaining real-time acquisition data and a random noise action sequence. The real-time acquisition data includes: the acquisition time period, the acquisition depth, acquisition temperature, acquisition humidity corresponding to the acquisition time period, and the historical irrigation action sequence; determining a target action sequence according to the random noise action sequence and the real-time acquisition data; inputting the target action sequence into the target decoder to obtain real-time irrigation data; and sending the real-time irrigation data to the irrigation device so that the irrigation device performs irrigation operations according to the real-time irrigation data.

[0095] Specifically, real-time collected data is obtained to collect the environment related to irrigation in real time. These data can reflect the current moisture condition of the soil and provide a basic basis for subsequent irrigation decisions. A random noise action sequence is obtained. The random noise action sequence can randomly generate parameter values corresponding to each parameter item according to each parameter item in the initial noise action sequence, and determine the random noise action sequence according to each parameter value. The random noise action sequence can simulate some uncertain factors or increase the robustness of the model. For example, in actual irrigation, it may be affected by some unpredictable factors, such as slight wind direction changes resulting in uneven water evaporation, small differences in soil texture, etc. The random noise action sequence can reflect these uncertainties to a certain extent. According to the random noise action sequence and the real-time collected data, the time and space feature data extracted from the real-time collected data and the historical irrigation action sequence are used as modulation conditions to denoise the random noise action sequence until the iteration end condition is reached, then the modulation is ended, the target action sequence is determined, the target action sequence is input into the target decoder, according to the correspondence between each feature information in the target action sequence and the irrigation action, the real-time irrigation data is obtained, and the real-time irrigation data is sent to the irrigation device so that the irrigation device performs irrigation operations according to the real-time irrigation data. The irrigation device can control each sprinkler to irrigate the crops. The irrigation prediction method can be as Figure 3 shown.

[0096] By obtaining the real-time collected data and the random noise action sequence, the real-time collected data includes: the collection time period, the collection depth corresponding to the collection time period, the collection temperature, the collection humidity and the historical irrigation action sequence; according to the random noise action sequence and the real-time collected data, the target action sequence is determined; the target action sequence is input into the target decoder to obtain the real-time irrigation data; the real-time irrigation data is sent to the irrigation device so that the irrigation device performs irrigation operations according to the real-time irrigation data. The irrigation data can be predicted by analyzing the real-time monitored data, so that the irrigation device irrigates the crops according to the irrigation data, automatically irrigates the crops, and can perform refined analysis for different depths, different temperatures and different humidities, ensuring the quality of the crops.

[0097] Embodiment III

[0098] Figure 4 FIG. is a schematic structural diagram of an irrigation prediction device provided in Embodiment III of the present invention. The embodiment of the present invention is applicable to the situation of irrigation prediction. The device can execute the irrigation prediction method, and the device can be implemented in the form of hardware and / or software.

[0099] See Figure 4The irrigation prediction device shown includes: a sequence acquisition module 401, a current sequence acquisition module 402, a data acquisition module 403, a data input module 404, a predicted data acquisition module 405, and a step size accumulation module 406, where,

[0100] The sequence acquisition module 401 is configured to acquire an initial noise action sequence;

[0101] The current sequence acquisition module 402 is configured to, when the number of current iteration steps obtained is less than or equal to the total number of model iteration steps, and based on the initial noise action sequence, determine the current noise action sequence corresponding to the number of current iteration steps;

[0102] The data acquisition module 403 is configured to determine the current supplementary data corresponding to the number of current iteration steps according to the parity of the number of current iteration steps;

[0103] The data input module 404 is configured to input the current supplementary data and the current noise action sequence into the irrigation prediction model to determine the predicted noise action sequence corresponding to the number of current iteration steps;

[0104] The predicted data acquisition module 405 is configured to, when the number of current iteration steps is greater than the total number of model iteration steps, acquire the predicted action sequence corresponding to the number of current iteration steps and determine the predicted irrigation data;

[0105] The step size accumulation module 406 is configured to, in response to the number of current iteration steps being less than or equal to the total number of model iteration steps, use the predicted action sequence corresponding to the number of current iteration steps as the current noise action sequence corresponding to the next iteration step number, and accumulate 1 on the number of current iteration steps.

[0106] In the technical solution of the embodiment of the present invention, an initial noise action sequence is obtained; when the currently obtained iteration step number is less than or equal to the total number of model iteration step numbers, and according to the initial noise action sequence, a current noise action sequence corresponding to the currently obtained iteration step number is determined; according to the parity of the currently obtained iteration step number, a current supplementary data corresponding to the currently obtained iteration step number is determined; the current supplementary data and the current noise action sequence are input into the irrigation prediction model to determine a predicted noise action sequence corresponding to the currently obtained iteration step number; when the currently obtained iteration step number is greater than the total number of model iteration step numbers, a predicted action sequence corresponding to the currently obtained iteration step number is obtained, and predicted irrigation data is determined; in response to the currently obtained iteration step number being less than or equal to the total number of model iteration step numbers, the predicted action sequence corresponding to the currently obtained iteration step number is used as the current noise action sequence corresponding to the next iteration step number, and 1 is added to the currently obtained iteration step number. By judging the parity of the currently obtained iteration step number, different data can be alternately used as conditions to denoise the initial noise action sequence, avoiding the mutual influence of different conditions corresponding to different iteration step numbers during the denoising process of the initial noise action sequence, and improving the accuracy of denoising the initial noise action sequence.

[0107] Optionally, the data acquisition module 403 includes:

[0108] A historical sequence acquisition sub-module for acquiring a historical irrigation action sequence;

[0109] A multi-dimensional information acquisition sub-module for acquiring updated spatial feature distribution information and updated temporal feature distribution information;

[0110] An odd supplementary value acquisition sub-module for, if the currently obtained iteration step number is odd, determining the updated spatial feature distribution information and the updated temporal feature distribution information as the current supplementary data corresponding to the currently obtained iteration step number;

[0111] An even supplementary value acquisition sub-module for, if the currently obtained iteration step number is even, determining the historical irrigation action sequence as the current supplementary data corresponding to the currently obtained iteration step number.

[0112] Optionally, the multi-dimensional information acquisition sub-module includes:

[0113] A spatio-temporal information acquisition unit for acquiring spatial feature distribution information and temporal feature distribution information;

[0114] A matrix acquisition unit for acquiring an updated matrix corresponding to the spatial feature distribution information and the temporal feature distribution information;

[0115] A spatial information update unit for performing cross-attention calculation on the spatial feature distribution information and the updated matrix to update the spatial feature distribution information and obtain updated spatial feature distribution information;

[0116] A time information update unit for performing cross-attention calculation on the time feature distribution information and an update matrix to update the time feature distribution information and obtain updated time feature distribution information.

[0117] Optionally, the matrix acquisition unit includes:

[0118] A historical data acquisition sub-unit for acquiring historical acquisition data, where the historical acquisition data includes: an acquisition time period, an acquisition depth corresponding to the acquisition time period, an acquisition temperature, an acquisition humidity, and a historical irrigation action sequence;

[0119] A feature matrix acquisition sub-unit for acquiring a feature matrix corresponding to the historical depth and historical temperature in the historical acquisition data;

[0120] A time dimension expansion sub-unit for expanding and processing the feature matrix corresponding to the historical depth and historical temperature in the historical acquisition data in the time dimension to obtain an update matrix corresponding to the spatial feature distribution information;

[0121] A spatial dimension expansion sub-unit for expanding and processing the feature matrix corresponding to the historical depth and historical temperature in the historical acquisition data in the spatial dimension to obtain an update matrix corresponding to the time feature distribution information.

[0122] Optionally, the feature matrix acquisition sub-unit is specifically used for:

[0123] Performing high-dimensional mapping processing on the historical acquisition data to obtain a mapping matrix corresponding to the historical depth and historical temperature in the historical acquisition data;

[0124] Obtaining a spatial position encoding and a time position encoding;

[0125] Fusing the spatial position encoding and the time position encoding with the mapping matrix to obtain a feature matrix corresponding to the historical depth and historical temperature.

[0126] Optionally, the predicted data acquisition module 405 includes:

[0127] A decoder acquisition sub-module for acquiring a target decoder trained on an initial decoder according to the historical acquisition data;

[0128] A relationship determination sub-module for determining the correspondence between the predicted action sequence and the irrigation operation according to the target decoder;

[0129] An irrigation data acquisition sub-module for determining predicted irrigation data according to the correspondence between the predicted action sequence and the irrigation operation.

[0130] Optionally, the predicted data acquisition module 405 further includes:

[0131] A real-time data acquisition sub-module, configured to acquire real-time acquisition data and a random noise action sequence. The real-time acquisition data includes: an acquisition time period, an acquisition depth corresponding to the acquisition time period, an acquisition temperature, an acquisition humidity, and a historical irrigation action sequence;

[0132] A target sequence acquisition sub-module, configured to determine a target action sequence according to the random noise action sequence and the real-time acquisition data;

[0133] An information acquisition sub-module, configured to input the target action sequence into a target decoder to obtain real-time irrigation data;

[0134] An information sending sub-module, configured to send the real-time irrigation data to an irrigation device, so that the irrigation device performs an irrigation operation according to the real-time irrigation data.

[0135] The irrigation prediction device provided by the embodiment of the present invention can execute the irrigation prediction method provided by any embodiment of the present invention, and has corresponding functional modules and beneficial effects for executing the irrigation prediction method.

[0136] Embodiment Four

[0137] Figure 5 FIG. shows a schematic structural diagram of an irrigation prediction device 500 that can be used to implement the embodiments of the present invention.

[0138] As Figure 5 shown, the irrigation prediction device 500 includes at least one processor 501 and a memory communicatively connected to the at least one processor 501, such as a read-only memory (ROM) 502, a random access memory (RAM) 503, etc. Among them, the memory stores a computer program executable by the at least one processor. The processor 501 can execute various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 502 or the computer program loaded from the storage unit 508 into the random access memory (RAM) 503. In the RAM 503, various programs and data required for the operation of the irrigation prediction device 500 can also be stored. The processor 501, the ROM 502, and the RAM 503 are connected to each other through a bus 504. The input / output (I / O) interface 505 is also connected to the bus 504.

[0139] Multiple components in the irrigation prediction device 500 are connected to the I / O interface 505, including: an input unit 506, such as a keyboard, a mouse, etc.; an output unit 507, such as various types of displays, speakers, etc.; a storage unit 508, such as a disk, an optical disc, etc.; and a communication unit 509, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 509 allows the irrigation prediction device 500 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0140] The processor 501 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the processor 501 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The processor 501 executes the various methods and processes described above, such as the irrigation prediction method.

[0141] In some embodiments, the irrigation prediction method can be implemented as a computer program tangibly embodied in a computer-readable storage medium, such as the storage unit 508. In some embodiments, part or all of the computer program can be loaded and / or installed onto the irrigation prediction device 500 via the ROM 502 and / or the communication unit 509. When the computer program is loaded into the RAM 503 and executed by the processor 501, one or more steps of the irrigation prediction method described above can be executed. Alternatively, in other embodiments, the processor 501 can be configured to execute the irrigation prediction method by any other suitable means (e.g., by means of firmware).

[0142] The various embodiments of the systems and techniques described above in this document can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGA), application-specific integrated circuits (ASIC), application-specific standard products (ASSP), system-on-a-chip systems (SOC), complex programmable logic devices (CPLD), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a dedicated or general-purpose programmable processor, and can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit the data and instructions to the storage system, the at least one input device, and the at least one output device.

[0143] The computer program for implementing the method of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to the processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the computer program is executed by the processor, the functions / operations specified in the flowchart and / or block diagram are implemented. The computer program can be executed entirely on the machine, partially on the machine, as an independent software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0144] In the context of the present invention, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. The computer-readable storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, the computer-readable storage medium can be a machine-readable signal medium. More specific examples of the machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0145] To provide for interaction with a user, the systems and techniques described herein can be implemented on an irrigation prediction device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the irrigation prediction device. Other kinds of devices can also be used to provide for interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0146] The systems and techniques described herein can be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer having a graphical user interface or a web browser through which the user can interact with an implementation of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0147] A computing system may include a client and a server. The client and the server are generally far from each other and usually interact via a communication network. The relationship between the client and the server is created by computer programs that run on respective computers and have a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or a cloud host, which is a host product in the cloud computing service system, solving the defects of difficult management and weak business scalability existing in traditional physical hosts and VPS (Virtual Private Server) services.

[0148] It should be understood that various forms of processes shown above can be used, steps can be reordered, added or deleted. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and no limitation is made herein.

[0149] The above specific embodiments do not constitute a limitation on the protection scope of the present invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. An irrigation prediction method, characterized in that: The method comprises: Obtain an initial noise action sequence; When the current iteration step number obtained is less than or equal to the total value of the model iteration step number, and based on the initial noise action sequence, determining the current noise action sequence corresponding to the current iteration step number; Determining current supplementary data corresponding to the current iteration step number according to the parity of the current iteration step number; Inputting the current supplementary data and the current noise action sequence into an irrigation prediction model to determine a predicted noise action sequence corresponding to the current number of iteration steps; When the current iteration step number is greater than the total value of the model iteration step number, a prediction action sequence corresponding to the current iteration step number is obtained, and predicted irrigation data is determined; In response to the current iteration step number being less than or equal to the total model iteration step number, the predicted action sequence corresponding to the current iteration step number is used as the current noise action sequence corresponding to the next iteration step number, and 1 is accumulated to the current iteration step number.

2. The method according to claim 1, characterized in that The determining, according to the parity of the current iteration step number, current supplementary data corresponding to the current iteration step number comprises: Get historical irrigation action sequence; Obtain updated spatial feature distribution information and updated temporal feature distribution information; If the current iteration step number is an odd number, the updated spatial feature distribution information and the updated temporal feature distribution information are determined as current supplementary data corresponding to the current iteration step number; If the current iteration step number is an even number, the historical irrigation action sequence is determined as the current supplementary data corresponding to the current iteration step number.

3. The method according to claim 2, characterized in that The obtaining and updating of spatial feature distribution information and temporal feature distribution information includes: Obtaining spatial feature distribution information and temporal feature distribution information; Obtaining an update matrix corresponding to the spatial feature distribution information and the temporal feature distribution information; Performing cross-attention calculation on the spatial feature distribution information and the update matrix, updating the spatial feature distribution information, and obtaining updated spatial feature distribution information; The time feature distribution information and the update matrix are cross-attention calculated, and the time feature distribution information is updated to obtain updated time feature distribution information.

4. The method according to claim 3, characterized in that The obtaining of the update matrix corresponding to the spatial feature distribution information and the temporal feature distribution information includes: Acquire historical collection data, the historical collection data including: collection time period, collection depth corresponding to the collection time period, collection temperature, collection humidity and historical irrigation action sequence; Obtain a feature matrix corresponding to the historical depth and historical temperature in the historical collection data; Expand and process the feature matrix corresponding to the historical depth and the historical temperature in the historical collection data in the time dimension to obtain an update matrix corresponding to the spatial feature distribution information; The characteristic matrix corresponding to the historical depth and the historical temperature in the historical collection data is expanded and processed in the spatial dimension to obtain an update matrix corresponding to the time characteristic distribution information.

5. The method according to claim 4, characterized in that The step of obtaining a characteristic matrix corresponding to the historical depth and the historical temperature in the historical acquisition data includes: Performing high-dimensional mapping processing on the historically collected data to obtain a mapping matrix corresponding to the historical depth and the historical temperature in the historically collected data; Obtaining spatial position coding and temporal position coding; The spatial position code and the temporal position code are fused with the mapping matrix to obtain a feature matrix corresponding to the historical depth and historical temperature.

6. The method according to claim 1, characterized in that The step of obtaining the prediction action sequence corresponding to the current iteration step number and determining the prediction irrigation data includes: Obtaining a target decoder trained on an initial decoder based on the historically collected data; determining, according to the target decoder, a correspondence between a predicted action sequence and an irrigation operation; According to the corresponding relationship between the predicted action sequence and the irrigation operation, the predicted irrigation data is determined.

7. The method according to claim 6, characterized in that After acquiring the target decoder trained on the initial decoder according to the historical acquisition data, the method further includes: Acquire real-time collection data and random noise action sequence, wherein the real-time collection data includes: collection time period, collection depth corresponding to the collection time period, collection temperature, collection humidity and historical irrigation action sequence; Determining a target action sequence according to the random noise action sequence and the real-time collected data; Inputting the target action sequence into the target decoder to obtain real-time irrigation data; The real-time irrigation data is sent to an irrigation device, so that the irrigation device performs an irrigation operation according to the real-time irrigation data.

8. An irrigation prediction device, characterized in that: The device comprises: A sequence acquisition module, used to acquire an initial noise action sequence; A current sequence acquisition module, used for determining a current noise action sequence corresponding to the current iteration step number when the acquired current iteration step number is less than or equal to the total value of the model iteration step number and according to the initial noise action sequence; A data acquisition module, used to determine the current supplementary data corresponding to the current iteration step number according to the parity of the current iteration step number; A data input module, used to input the current supplementary data and the current noise action sequence into the irrigation prediction model, and determine the predicted noise action sequence corresponding to the current iteration step number; A prediction data acquisition module, used for acquiring a prediction action sequence corresponding to the current iteration step number and determining the prediction irrigation data when the current iteration step number is greater than the total value of the model iteration step number; A step accumulation module is used to respond to the current iteration step number being less than or equal to the total value of the model iteration step number, use the predicted action sequence corresponding to the current iteration step number as the current noise action sequence corresponding to the next iteration step number, and accumulate 1 on the current iteration step number.

9. An irrigation prediction device, characterized in that: The irrigation prediction device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can perform the irrigation prediction method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the irrigation prediction method according to any one of claims 1 to 7 when executed.

Images