A saline-alkali soil fertilization method and system based on intelligent recommendation

Abstract

The application discloses a saline-alkali soil fertilization method and system based on intelligent recommendation. A map model based on saline-alkali soil is constructed, a preset saline-alkali soil is divided into monitoring areas, similarity analysis is performed on the environmental characteristic data of each two monitoring areas, and a feasible prediction area group is determined; in a real-time monitoring period, soil salinity of a sampling area is detected, and the detection data are introduced into an LSTM prediction model to perform salinity content numerical value prediction, so that prediction salinity data of the feasible prediction area group are obtained; through the prediction salinity data and historical salinity content data, a fertilization and irrigation scheme setting is performed on multiple monitoring areas, and intelligent scheme recommendation is performed based on different monitoring areas. Through the application, the salinity content of the saline-alkali soil can be effectively predicted, the consumption of manpower and material resources for detecting soil is reduced, and the saline-alkali soil regulation and control capability is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of saline-alkali data intelligent analysis, and more particularly, to a saline-alkali fertilization method and system based on intelligent recommendation. BACKGROUND

[0002] The saline-alkali soil is the general term of saline soil and alkaline soil, which has the characteristics of low organic matter content, low soil fertility, poor physicochemical properties and the like, and seriously affects the growth of crops. The traditional fertilization method is often based on experience or simple soil test, which is difficult to accurately determine the soil fertility and crop demand, and the detection and analysis efficiency of the saline-alkali soil is low, the analysis accuracy of different environmental regions in the saline-alkali soil is low, and the soil detection and analysis seriously depend on manual detection data. There is no efficient information means to effectively analyze and evaluate different regions, and there is also a lack of corresponding saline-alkali data prediction means, which seriously affects the effective analysis of the fertilization scheme, resulting in low fertilization efficiency, and even aggravating the soil salinization. Therefore, it is of great significance to develop a saline-alkali fertilization method and system based on intelligent recommendation for improving the agricultural production efficiency of saline-alkali soil. SUMMARY

[0003] The present application overcomes the defects of the prior art and provides a saline-alkali fertilization method and system based on intelligent recommendation.

[0004] The present application provides a saline-alkali fertilization method based on intelligent recommendation in the first aspect, comprising:

[0005] Obtaining preset saline-alkali region information, and constructing a map model based on saline-alkali soil through the region information;

[0006] In the map model, the preset monitoring point is used to divide the preset saline-alkali soil into a plurality of monitoring regions, and the soil environment data of the monitoring regions is obtained;

[0007] In the soil environment data, a plurality of dimensional feature vectors corresponding to a plurality of environmental parameters are generated, and the environmental feature data of each monitoring region is formed;

[0008] The environmental feature data of the monitoring region is analyzed for similarity, the regions with expected similarity and continuous geographical position are marked, and each set of continuous regions marked is taken as a feasible prediction region group;

[0009] In a preset historical time period, the saline-alkali content data of the feasible prediction region group is extracted and sequenced, the sequenced data is imported into an LSTM prediction model for training, and one monitoring region is selected as a sampling region in each feasible prediction region group;

[0010] In a real-time monitoring cycle, the soil salinity of the sampling area is detected, and the detection data is imported into the LSTM prediction model to predict the salt content value, and the prediction salt data of the feasible prediction area group is obtained;

[0011] Through the prediction salt data and the historical salt content data, the fertilization and irrigation scheme of the multiple monitoring areas is set, and the intelligent scheme recommendation is carried out based on different monitoring areas.

[0012] In the scheme, the preset salt area information is obtained, and a map model based on the saline soil is constructed based on the area information, specifically:

[0013] The preset salt area information is obtained, and the area information includes area area, area contour, planting range and monitoring point position information;

[0014] Through the area information, a two-dimensional visual map model is constructed.

[0015] In the scheme, in the map model, the preset monitoring point is used to divide the preset saline soil into multiple monitoring areas, and the soil environment data of the monitoring area is obtained, specifically:

[0016] In the map model, the position of the preset monitoring point is obtained;

[0017] Based on the position of the preset monitoring point, the preset saline soil is divided into multiple monitoring areas, and each monitoring area corresponds to a preset monitoring point;

[0018] In a preset historical time period, the soil environment data of each monitoring area is collected.

[0019] In the scheme, in the soil environment data, a plurality of dimensional feature vectors are generated corresponding to a plurality of environmental parameters, and environmental feature data of each monitoring area is formed, specifically:

[0020] In the soil environment data, each environmental parameter is taken as a dimension, and the corresponding parameter value is taken as the value of the corresponding dimension, to generate a multi-dimensional feature vector;

[0021] The multi-dimensional feature vector is taken as the environmental feature data, and the feature extraction of each monitoring area is carried out to form the environmental feature data of each monitoring area.

[0022] In the scheme, the similarity analysis is carried out on the environmental feature data of the monitoring area, the regions with expected similarity and continuous geographical position are marked, and each marked continuous region is taken as a feasible prediction area group, specifically:

[0023] The similarity calculation is performed on the environmental characteristic data of the monitoring areas, and the calculation process is to calculate the environmental similarity between two selected monitoring areas based on the standard Euclidean distance method, and the environmental similarity between the two monitoring areas is obtained through the distance value;

[0024] By the preset similarity threshold, it is judged whether each two monitoring areas have similarity, and all monitoring areas are judged, and the monitoring areas with similarity are marked to form a similar monitoring group, and based on all monitoring areas, a plurality of similar monitoring groups are generated;

[0025] A similar monitoring group includes at least two monitoring areas;

[0026] In the similar monitoring group, each monitoring area is guaranteed to have similarity with at least one monitoring area in the group;

[0027] In a similar monitoring group, the monitoring areas that are continuous in geographical position are screened, and a new group is formed based on the screening result, which is marked as a feasible prediction area group;

[0028] All similar monitoring groups are screened and marked to form a plurality of feasible prediction area groups.

[0029] In the scheme, the salt and alkali content data of the feasible prediction area group in a preset historical time period are extracted and sequenced, and the sequenced data is imported into an LSTM prediction model for training, and one monitoring area is selected as a sampling area in each feasible prediction area group, specifically:

[0030] A feasible prediction area group is taken as an analysis unit;

[0031] In a preset historical time period, the salt and alkali content data of all monitoring areas in the one feasible prediction area group are obtained;

[0032] The salt and alkali content data are sorted based on the geographical continuous order between the monitoring areas in the one feasible prediction area group, and the sorted data are sequenced to form sequenced data;

[0033] An LSTM prediction model is constructed, and parameters and a loss function are initialized;

[0034] The sequenced data is divided into a test set and a training set based on a preset ratio and imported into the LSTM prediction model for cyclic prediction training, and the model parameters are optimized through the loss function during the training, and the training is performed until the prediction accuracy reaches a preset standard;

[0035] The specific model parameters of the LSTM prediction model at this time are recorded, and the model parameters are associated with the one feasible prediction area group;

[0036] In each feasible prediction area group, a monitoring area with a starting position is selected as a sampling area based on geographical continuity.

[0037] In this scheme, the soil salinity of the sampling area is detected in a real-time monitoring period, and the detection data is imported into the LSTM prediction model to predict the salt content value, and the prediction salt data of the feasible prediction area group is obtained, specifically:

[0038] A feasible prediction area group is taken as an analysis unit.

[0039] In a real-time monitoring period, the soil salinity of the sampling area in a feasible prediction area group is detected to obtain detection data.

[0040] The detection data is imported into the LSTM prediction model to predict the salt content value, and the prediction sequence data is obtained.

[0041] The prediction sequence data is parsed, and the data corresponding to each monitoring area in a feasible prediction area group is obtained, and the prediction salt data of each monitoring area in the group is obtained.

[0042] In this scheme, the prediction salt data and historical salt content data are used to set the fertilization and irrigation scheme for multiple monitoring areas, and intelligent scheme recommendation is performed based on different monitoring areas, specifically:

[0043] In a feasible prediction area group, the detection data and prediction salt data of the sampling area are used to evaluate the salt change of the area, and the salt content control index and alkalinity control index of the soil are generated.

[0044] Based on the salt content control index and alkalinity control index of the soil, the fertilization and irrigation plan is analyzed, and the prediction area fertilization scheme is generated.

[0045] According to multiple feasible prediction area groups, multiple prediction area fertilization schemes are generated.

[0046] In the map model, all monitoring areas outside the feasible prediction area group are marked as non-prediction areas, and real-time salt data detection is performed on the non-prediction areas. Combined with historical salt content data, matching selection analysis is performed in the fertilization scheme to screen multiple real-time fertilization schemes.

[0047] Based on multiple prediction area fertilization schemes and multiple real-time fertilization schemes, intelligent scheme recommendation is performed on the preset salt area.

[0048] The second aspect of the present application also provides a saline-alkali soil fertilization system based on intelligent recommendation, which comprises a memory and a processor, the memory comprises a saline-alkali soil fertilization program based on intelligent recommendation, and the saline-alkali soil fertilization program based on intelligent recommendation is executed by the processor to realize the following steps:

[0049] preset saline-alkali region information is acquired, and a map model based on saline-alkali soil is constructed through the region information;

[0050] In the map model, the preset saline-alkali soil is divided into a plurality of monitoring regions through preset monitoring points, and soil environment data of the monitoring regions is acquired;

[0051] In the soil environment data, a plurality of dimensional feature vectors corresponding to a plurality of environmental parameters are generated, and environmental feature data of each monitoring region is formed;

[0052] The environmental feature data of the monitoring regions is subjected to similarity analysis, regions with expected similarity and continuous geographical positions are marked, and each set of marked continuous regions is taken as a feasible prediction region group;

[0053] In a preset historical time period, saline-alkali content data of the feasible prediction region group is extracted and subjected to data serialization, the serialized data is imported into an LSTM prediction model for training, and one monitoring region in each feasible prediction region group is selected as a sampling region;

[0054] In a real-time monitoring period, soil saline-alkali detection is performed on the sampling region, and the detection data is imported into the LSTM prediction model for saline-alkali content numerical prediction to obtain predicted saline-alkali data of the feasible prediction region group;

[0055] Through the predicted saline-alkali data and the historical saline-alkali content data, a fertilization and irrigation scheme setting is performed on a plurality of monitoring regions, and an intelligent scheme recommendation is performed based on different monitoring regions.

[0056] The second aspect of the present application also provides a saline-alkali soil fertilization system based on intelligent recommendation, which comprises a memory and a processor, the memory comprises a saline-alkali soil fertilization program based on intelligent recommendation, and the saline-alkali soil fertilization program based on intelligent recommendation is executed by the processor to realize the following steps:

[0057] The third aspect of the present application also provides a computer readable storage medium, which comprises a saline-alkali soil fertilization program based on intelligent recommendation, and the saline-alkali soil fertilization program based on intelligent recommendation is executed by a processor to realize the steps of the saline-alkali soil fertilization method based on intelligent recommendation according to any one of the above.

[0058] This invention discloses a method and system for fertilizing saline-alkali land based on intelligent recommendation. By constructing a map model of the saline-alkali land, monitoring areas are divided into preset areas. Similarity analysis is performed on the environmental characteristic data of every two monitoring areas to determine feasible prediction area groups. Within a real-time monitoring cycle, soil salinity is tested in the sampling areas, and the test data is imported into an LSTM prediction model to predict salinity content, obtaining predicted salinity data for feasible prediction area groups. Based on the predicted salinity data and historical salinity content data, fertilization and irrigation schemes are set for multiple monitoring areas, and intelligent scheme recommendations are made for different monitoring areas. This invention enables effective prediction of salinity content in saline-alkali land, reduces the manpower and material resources required for soil testing, and improves the ability to regulate saline-alkali land. Attached Figure Description

[0059] Figure 1 A flowchart of a fertilization method for saline-alkali land based on intelligent recommendation according to the present invention is shown;

[0060] Figure 2 A block diagram of a fertilization system for saline-alkali land based on intelligent recommendation according to the present invention is shown. Detailed Implementation

[0061] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0062] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0063] Figure 1 A flowchart of a fertilization method for saline-alkali land based on intelligent recommendation according to the present invention is shown.

[0064] like Figure 1 As shown, the first aspect of the present invention provides a method for fertilizing saline-alkali land based on intelligent recommendation, comprising:

[0065] S102, Obtain preset saline-alkali land area information, and construct a map model based on saline-alkali land using the area information;

[0066] S104. In the map model, the preset saline-alkali land is divided into multiple monitoring areas by preset monitoring points, and soil environmental data of the monitoring areas are obtained.

[0067] S106, in the soil environment data, a plurality of environmental parameters are correspondingly generated into a plurality of dimensional feature vectors, and environmental feature data of each monitoring area is formed;

[0068] S108, similarity analysis is performed on the environmental feature data of the monitoring area, areas with expected similarity and continuous geographical positions are marked, and each set of marked continuous areas is taken as a feasible prediction area group;

[0069] S110, in a preset historical time period, the salt and alkali content data of the feasible prediction area group is extracted and sequenced, the sequenced data is imported into an LSTM prediction model for training, and one monitoring area is selected as a sampling area in each feasible prediction area group;

[0070] S112, in a real-time monitoring period, soil salinity detection is performed on the sampling area, and the detection data is imported into the LSTM prediction model for salt and alkali content numerical prediction to obtain predicted salt and alkali data of the feasible prediction area group;

[0071] S114, by the predicted salt and alkali data and the historical salt and alkali content data, a fertilization and irrigation scheme setting is performed on a plurality of monitoring areas, and intelligent scheme recommendation is performed based on different monitoring areas.

[0072] According to the embodiment of the present application, the preset salt and alkali area information is obtained, and a map model based on the saline-alkali soil is constructed based on the area information, specifically:

[0073] The preset salt and alkali area information is obtained, and the area information includes area area, area contour, planting range and monitoring point position information;

[0074] Based on the area information, a two-dimensional visual map model is constructed.

[0075] It should be noted that the map model is a visual model, which is used for visual display of the current salt and alkali data and corresponding area display, so that the influence distribution of the saline-alkali soil can be conveniently and intuitively understood.

[0076] According to the embodiment of the present application, in the map model, the preset monitoring point is used to divide the preset saline-alkali soil into a plurality of monitoring areas, and the soil environment data of the monitoring area is obtained, specifically:

[0077] In the map model, the position of the preset monitoring point is obtained;

[0078] Based on the position of the preset monitoring point, the preset saline-alkali soil is divided into a plurality of monitoring areas, so that each monitoring area corresponds to a preset monitoring point;

[0079] In a preset historical time period, the soil environment data of each monitoring area is collected.

[0080] It should be noted that the preset monitoring points include multiple. The preset historical time period is generally a user set monitoring period, and sufficient amount of environmental feature data needs to be collected to effectively analyze the environmental data of the region. The soil environmental data includes various environmental parameters such as soil temperature, humidity, evaporation, conductivity, etc.

[0081] According to the embodiment of the present application, the multi-dimensional feature vector corresponding to the various environmental parameters in the soil environmental data is generated, and the environmental feature data of each monitoring region is formed, specifically:

[0082] In the soil environmental data, each environmental parameter is taken as a dimension, and the corresponding parameter value is taken as the value of the corresponding dimension, to generate a multi-dimensional feature vector;

[0083] The multi-dimensional feature vector is taken as the environmental feature data, and the feature extraction is performed on each monitoring region to form the environmental feature data of each monitoring region.

[0084] It should be noted that in the feature vector, the one environmental parameter corresponds to one dimension.

[0085] According to the embodiment of the present application, the similarity analysis is performed on the environmental feature data of the monitoring region, the regions with expected similarity and continuous geographical position are marked, and each set of marked continuous regions is taken as a feasible prediction region group, specifically:

[0086] The similarity calculation is performed on the environmental feature data of the monitoring region, and the calculation process is to calculate the environmental feature data of the selected two monitoring regions based on the standard Euclidean distance method, and the environmental similarity between the two monitoring regions is obtained through the distance value;

[0087] Through the preset similarity threshold, it is judged whether each two monitoring regions have similarity, and all monitoring regions are judged, the monitoring regions with similarity are marked to form a similar monitoring group, and based on all monitoring regions, multiple similar monitoring groups are generated;

[0088] One similar monitoring group includes at least two monitoring regions;

[0089] In the similar monitoring group, it is ensured that each monitoring region has similarity with at least one monitoring region in the group;

[0090] In one similar monitoring group, the monitoring regions with continuous geographical position are screened, and based on the screening result, a new group is formed and marked as a feasible prediction region group;

[0091] All similar monitoring groups are screened and marked to form multiple feasible prediction region groups.

[0092] It should be noted that in the similarity obtained by the distance value, the greater the distance, the smaller the similarity. The feasible prediction area group includes at least two monitoring areas. In the process of forming a plurality of feasible prediction area groups, the areas with association and continuous monitoring data can be classified and grouped to form a plurality of groups, so as to realize the subsequent precise and efficient fertilization scheme analysis and saline-alkali soil regulation.

[0093] In the present application, first, the division of the monitoring area is carried out in the saline-alkali soil, the similarity and the continuity of the geographical position are judged based on the environmental characteristics, and different area groups, i.e. feasible prediction area groups, are selected. In the feasible prediction area group, the areas in the group have high environmental characteristic similarity and detection data association. Based on this, the corresponding saline-alkali content change and detection value have a certain correlation, and therefore have a certain prediction feasibility. Further, the present application trains a prediction model for the historical data of each feasible prediction area group. The prediction model adopts a deep prediction model based on sequence data to simulate the change of the saline-alkali value in the continuous area in the region. Different feasible prediction area groups are associated with different model parameters, so that each feasible prediction area group corresponds to a trained prediction model. In the real-time monitoring period, by combining the sampling area data with the prediction model, the remaining areas in the feasible prediction area group can be precisely predicted, greatly reducing the number of manual sampling and detection frequency, and providing an efficient saline-alkali soil analysis method based on informationization. It is worth mentioning that the current sampling and detection process of land is time-consuming and laborious. For large-scale saline-alkali soil, general area sampling analysis is time-consuming and laborious. Therefore, by using the present application, precise data prediction for multiple areas can be realized based on one-time sampling analysis, greatly improving the monitoring and analysis efficiency of saline-alkali soil, reducing repetitive sampling and analysis process, and achieving cost reduction and efficiency increase.

[0094] According to the embodiment of the present application, in a preset historical time period, the saline-alkali content data of the feasible prediction area group is extracted and sequenced, the sequenced data is imported into the LSTM prediction model for training, and one monitoring area in each feasible prediction area group is selected as a sampling area. Specifically:

[0095] Taking one feasible prediction area group as an analysis unit;

[0096] In a preset historical time period, the saline-alkali content data of all monitoring areas in the one feasible prediction area group is obtained;

[0097] The saline-alkali content data is sorted based on the geographical continuous sequence of the monitoring areas in one feasible prediction area group, and the sorted data is sequenced to form sequenced data;

[0098] Constructing an LSTM prediction model and initializing parameters and setting a loss function;

[0099] The serialized data is divided into a test set and a training set based on a preset ratio and imported into the LSTM prediction model for cyclic prediction training. During the training process, the model parameters are optimized through the loss function, and the training is performed until the prediction accuracy reaches a preset standard.

[0100] The specific model parameters of the LSTM prediction model at this time are recorded, and the model parameters are associated with the one feasible prediction area group.

[0101] In each feasible prediction area group, a monitoring area at a starting position is selected as a sampling area based on the continuity of geographical positions.

[0102] It should be noted that the salt and alkali content data is sorted based on the geographical continuity sequence between the monitoring areas in one feasible prediction area group. Specifically, a region route is formed based on the continuous areas, and the salt and alkali content data of each monitoring area is serialized based on the sequence in the region route. For example, in a preset salt and alkali area, multiple monitoring areas are divided by a grid, and in one feasible prediction area group, multiple continuous monitoring areas are included horizontally. The salt and alkali content data can be serialized based on the sequence from left to right (or from right to left) of the multiple continuous monitoring areas, and each unit data after serialization corresponds to the data of each monitoring area. It is worth mentioning that in each feasible prediction area group, a monitoring area at a starting position is selected as a sampling area based on the continuity of geographical positions. In the above example, the leftmost monitoring area can be selected as the sampling area in the sequence from left to right of the multiple continuous monitoring areas, because it can be used as a starting point to connect the remaining monitoring areas in geographical position. In the following, the salt and alkali data of the sampling area can be used to predict the salt and alkali data of the remaining continuous monitoring areas.

[0103] The specific model parameters of the LSTM prediction model at this time are recorded, and the model parameters are associated with the one feasible prediction area group. Since the environmental characteristics of each feasible prediction area group are different, the corresponding salt and alkali content is also different, and the change rule is different. Therefore, the present application trains the prediction model for different groups to generate model parameters corresponding to the group, and the prediction model parameters of different groups are different.

[0104] The LSTM (Long Short-Term Memory) is a special recurrent neural network (RNN) for data prediction. In the embodiment of the present application, since the environmental characteristics in similar and adjacent continuous monitoring areas are related and continuous, the corresponding saline-alkali content has a certain correlation. The present application can effectively mine the change rule of the saline-alkali content in the continuous area, realize the accurate prediction of the saline-alkali soil, improve the informatization combination ability of the saline-alkali soil, reduce unnecessary sampling monitoring, reduce the consumption of manpower and material resources in the saline-alkali soil monitoring, and realize cost reduction and efficiency increase.

[0105] According to the embodiment of the present application, the soil salinity of the sampling area is detected in a real-time monitoring period, and the detection data is imported into the LSTM prediction model to predict the saline-alkali content value, to obtain the predicted saline-alkali data of the feasible prediction area group, specifically:

[0106] Taking a feasible prediction area group as an analysis unit;

[0107] In a real-time monitoring period, the soil salinity of the sampling area in a feasible prediction area group is detected to obtain detection data;

[0108] The detection data is imported into the LSTM prediction model to predict the saline-alkali content value, to obtain the predicted sequence data;

[0109] The predicted sequence data is parsed, and the data is corresponded to each monitoring area in a feasible prediction area group, to obtain the predicted saline-alkali data of each monitoring area in the group.

[0110] It should be noted that the detection data is specifically the saline-alkali content data. The saline-alkali content data includes the salt content and the alkalinity of the soil. In a feasible prediction area group, the sampling area is the detection data, and the remaining monitoring areas are all predicted saline-alkali data.

[0111] According to the embodiment of the present application, the prediction saline-alkali data and the historical saline-alkali content data are used to set the fertilization and irrigation scheme of the multiple monitoring areas, and the intelligent scheme recommendation is made based on different monitoring areas, specifically:

[0112] In a feasible prediction area group, the detection data and the prediction saline-alkali data of the sampling area are used to evaluate the saline-alkali change of the area, and the salt content control index and the alkalinity control index of the soil are generated;

[0113] The fertilization and irrigation plan analysis is made based on the salt content control index and the alkalinity control index of the soil, and the prediction area fertilization scheme is generated;

[0114] According to a plurality of feasible prediction area groups, a plurality of prediction area fertilization schemes are generated;

[0115] In the map model, the monitoring area outside all the feasible prediction area groups is marked as a non-prediction area, real-time saline-alkali data detection is performed on the non-prediction area, historical saline-alkali content data are combined, matching selection analysis is performed in the prediction fertilization scheme, and a plurality of real-time fertilization schemes are screened out.

[0116] Based on the plurality of prediction area fertilization schemes and the plurality of real-time fertilization schemes, an intelligent scheme recommendation is performed on a preset saline-alkali area.

[0117] It should be noted that in the intelligent scheme recommendation, the prediction area fertilization scheme can be applied to the fertilization scheme recommendation of the feasible prediction area group, and the scheme is specifically a scheme dynamically analyzed based on prediction data; the plurality of real-time fertilization schemes are fixed optimal schemes, and are suitable for the non-prediction area. The prediction fertilization scheme is a fixed recommended scheme, and includes a plurality of.

[0118] In the fertilization and irrigation plan analysis, the quality requirements of the irrigation water source can be determined based on corresponding indexes, including conductivity and pH value, and dynamic regulation and control of irrigation using low-salt water can help to reduce the accumulation of salt in the soil. The fertilization scheme can set the fertilization amount and fertilizer salt proportion based on indexes, and regulate and control the degree of soil salinization.

[0119] According to the embodiment of the application, the method further comprises:

[0120] A feasible prediction area group is marked as a current group;

[0121] In a second monitoring period, second soil environment data of the monitoring area in the current group are acquired;

[0122] The second soil environment data are subjected to feature extraction to obtain second environment feature data;

[0123] Based on the plurality of monitoring areas in the current group, variance calculation is performed on the second environment feature data. In the calculation process, the data difference value is used to calculate the difference value between two second environment feature data in the standard Euclidean distance, to obtain a variance result, and it is judged whether the variance result is within a preset variance range.

[0124] If not, secondary similarity analysis and judgment are performed on the plurality of monitoring areas in the current group, a region with an expected similarity and a continuous geographical position is judged, and based on the judgment result, the monitoring area in the current group is updated to form a new feasible prediction area group.

[0125] It should be noted that with the change of environmental factors, the environmental characteristics of the monitoring areas in the feasible prediction area group will also change accordingly, and when the change exceeds the expected value, the monitoring areas in the feasible prediction area group will no longer have corresponding relevance, for example, a serious sudden pollution or a sudden change in climate environmental factors occurs in a certain area, at this time, the feasible prediction area group needs to be updated in real time to dynamically adapt to different environmental changes, ensure the applicability and prediction accuracy of the prediction model, and at the same time, the fertilization scheme is updated synchronously, and the dynamic planning ability of the scheme in the information analysis is improved.

[0126] Figure 2 A block diagram of a saline-alkali soil fertilization system based on intelligent recommendation is shown.

[0127] The second aspect of the present application also provides a saline-alkali soil fertilization system 2 based on intelligent recommendation, which comprises a memory 21 and a processor 22, the memory comprises a saline-alkali soil fertilization program based on intelligent recommendation, and the saline-alkali soil fertilization program based on intelligent recommendation is executed by the processor to realize the following steps:

[0128] Obtain preset saline-alkali region information, and construct a map model based on saline-alkali soil through the region information;

[0129] In the map model, divide the preset saline-alkali soil into a plurality of monitoring areas through a preset monitoring point, and obtain soil environment data of the monitoring areas;

[0130] In the soil environment data, a plurality of dimensional feature vectors corresponding to a plurality of environmental parameters are generated, and environmental characteristic data of each monitoring area is formed;

[0131] Similarity analysis is performed on the environmental characteristic data of the monitoring areas, regions with expected similarity and continuous geographical positions are marked, and each set of marked continuous regions is taken as a feasible prediction area group;

[0132] In a preset historical time period, the saline-alkali content data of the feasible prediction area group is extracted and sequenced, the sequenced data is imported into an LSTM prediction model for training, and one monitoring area in each feasible prediction area group is selected as a sampling area;

[0133] In a real-time monitoring period, soil salinity of the sampling area is detected, and the detection data is imported into the LSTM prediction model for saline-alkali content numerical prediction to obtain predicted saline-alkali data of the feasible prediction area group;

[0134] Through the predicted saline-alkali data and the historical saline-alkali content data, a fertilization and irrigation scheme is set for a plurality of monitoring areas, and an intelligent scheme is recommended based on different monitoring areas.

[0135] According to the embodiment of the present application, the preset saline area information is acquired, and a map model based on saline land is constructed according to the area information, specifically as follows:

[0136] The preset saline area information is acquired, and the area information includes area area, area contour, planting range and monitoring point position information.

[0137] The map model based on two-dimensional visualization is constructed according to the area information.

[0138] It should be noted that the map model is a kind of visual model, which is used for visual display and corresponding area display of current saline data, so that the influence distribution of saline land can be conveniently and intuitively understood.

[0139] According to the embodiment of the present application, in the map model, the preset saline land is divided into a plurality of monitoring areas by the preset monitoring points, and the soil environment data of the monitoring areas is acquired, specifically as follows:

[0140] In the map model, the position of the preset monitoring point is acquired.

[0141] Based on the position of the preset monitoring point, the preset saline land is divided into a plurality of monitoring areas, and each monitoring area corresponds to a preset monitoring point.

[0142] In a preset historical time period, the soil environment data of each monitoring area is collected.

[0143] It should be noted that the preset monitoring points include a plurality of preset monitoring points. The preset historical time period is generally a monitoring period set by a user, and sufficient amount of environmental characteristic data needs to be collected to effectively analyze the environmental data of the region. The soil environment data includes various environmental parameters, such as soil temperature, humidity, evaporation, conductivity, etc.

[0144] According to the embodiment of the present application, in the soil environment data, a plurality of dimensional feature vectors are generated corresponding to a plurality of environmental parameters, and environmental characteristic data of each monitoring area is formed, specifically as follows:

[0145] In the soil environment data, each environmental parameter is taken as a dimension, and the corresponding parameter value is taken as the value of the corresponding dimension, to generate a multi-dimensional feature vector.

[0146] The multi-dimensional feature vector is taken as environmental characteristic data, and feature extraction is performed on each monitoring area to form environmental characteristic data of each monitoring area.

[0147] It should be noted that in the feature vector, one environmental parameter corresponds to one dimension.

[0148] According to the embodiment of the present application, the similarity analysis is performed on the environmental characteristic data of the monitoring areas, the areas with expected similarity and continuous geographical positions are marked, each set of the marked continuous areas is taken as a feasible prediction area group, and specifically,

[0149] The similarity calculation is performed on the environmental characteristic data of the monitoring areas, the calculation process is that the environmental characteristic data of the selected two monitoring areas is calculated based on the standard Euclidean distance method, and the environmental similarity between the two monitoring areas is obtained through the distance value;

[0150] Whether each two monitoring areas has similarity is judged through the preset similarity threshold, all the monitoring areas are judged, the monitoring areas with similarity are marked to form a similar monitoring group, and based on all the monitoring areas, a plurality of similar monitoring groups are generated;

[0151] One similar monitoring group includes at least two monitoring areas;

[0152] In the similar monitoring group, it is ensured that each monitoring area has similarity with at least one monitoring area in the group;

[0153] In one similar monitoring group, the monitoring areas with continuous geographical positions are screened, a new group is formed based on the screening result, and is marked as a feasible prediction area group;

[0154] All the similar monitoring groups are screened and marked to form a plurality of feasible prediction area groups.

[0155] It should be noted that in the similarity obtained through the distance value, the greater the distance, the smaller the similarity. The feasible prediction area group includes at least two monitoring areas. In the process of forming a plurality of feasible prediction area groups, the areas with association and continuous monitoring data can be classified and grouped to form a plurality of groups, so as to realize the precise and efficient fertilization scheme analysis and saline-alkali soil regulation in the subsequent process.

[0156] In the present application, first, the monitoring area is divided by saline-alkali land, the similarity and geographical continuity are judged based on environmental characteristics, and different regional groups are selected, that is, the feasible prediction regional group, in the feasible prediction regional group, the regions in the group have high environmental characteristic similarity and correlation of detection data, based on this, the corresponding saline-alkali content change and detection value have certain correlation, therefore, it has certain prediction feasibility, further, the present application trains the prediction model for the historical data of each feasible prediction regional group, the prediction model adopts a deep prediction model based on sequence data, to simulate the change of saline-alkali value in the continuous region in the region, different feasible prediction regional groups are associated with different model parameters, so that each feasible prediction regional group corresponds to a trained prediction model, in the real-time monitoring period, through the sampling area data combined with the prediction model, the remaining regions in the feasible prediction regional group can be accurately predicted, which greatly reduces the number of artificial sampling and detection frequency, and provides an efficient information-based saline-alkali land analysis method. It is worth mentioning that the current sampling and detection process of existing land is time-consuming and laborious, and generally uses full-area sampling analysis for large-scale saline-alkali land, which is time-consuming and laborious, therefore, through the present application, accurate data prediction for multiple regions can be realized based on one-time sampling analysis, which greatly improves the monitoring and analysis efficiency of saline-alkali land, reduces the repetitive sampling and analysis process, and realizes cost reduction and efficiency increase.

[0157] According to the embodiment of the present application, in a preset historical time period, the saline-alkali content data of the feasible prediction regional group is extracted and sequenced, the sequenced data is imported into the LSTM prediction model for training, and one monitoring region is selected as a sampling region in each feasible prediction regional group, specifically:

[0158] Taking one feasible prediction regional group as an analysis unit;

[0159] In a preset historical time period, the saline-alkali content data of all monitoring regions in the one feasible prediction regional group is obtained;

[0160] The saline-alkali content data is sorted based on the geographical continuous sequence of the monitoring regions in one feasible prediction regional group, and the sorted data is sequenced to form sequenced data;

[0161] An LSTM prediction model is constructed and the parameters and loss function are initialized;

[0162] The sequenced data is divided into a test set and a training set based on a preset ratio and imported into the LSTM prediction model for cyclic prediction training, the model parameters are optimized by the loss function during the training process, and the training is performed until the prediction accuracy reaches a preset standard;

[0163] record the specific model parameters of the LSTM prediction model at this time, and associate the model parameters with the one feasible prediction area group;

[0164] In each feasible prediction area group, based on the continuity of the geographical position, a monitoring area of a starting position is selected as a sampling area.

[0165] It should be noted that in the sorting of the saline-alkali content data based on the geographical continuity order between the monitoring areas in one feasible prediction area group, a region route is formed based on the continuous areas, and the saline-alkali content data of each monitoring area is serialized based on the order in the region route. For example, in a preset saline-alkali area, a plurality of monitoring areas are divided by a grid, and in one feasible prediction area group, a plurality of continuous monitoring areas are included, and the saline-alkali content data can be serialized based on the order from left to right (or from right to left) of the plurality of continuous monitoring areas, and each unit data after serialization corresponds to the data of each monitoring area. It is worth mentioning that in the selection of a monitoring area of a starting position as a sampling area based on the continuity of the geographical position in each feasible prediction area group, the leftmost monitoring area can be selected as the sampling area in the order from left to right of the plurality of continuous monitoring areas, because it can be used as a starting point to connect the remaining monitoring areas in geographical position. In the follow-up, the saline-alkali data of the sampling area can be used to predict the saline-alkali data of the remaining continuous monitoring areas.

[0166] In the recording of the specific model parameters of the LSTM prediction model at this time, and the association of the model parameters with the one feasible prediction area group, since the environment features of each feasible prediction area group are different, the corresponding saline-alkali content is also different, and the change rule is different, therefore, the present application trains the prediction model for different groups respectively, generates the model parameters of the corresponding group, and the prediction model parameters of different groups are different.

[0167] LSTM, which stands for Long Short-Term Memory, is a special kind of recurrent neural network (RNN) used for data prediction. In the embodiments of the present application, since in similar and adjacent continuous monitoring areas, the environmental features have relevance and continuity, the corresponding saline-alkali content has certain relevance, and the present application can effectively mine the change rule of the saline-alkali content in the continuous area, realize the accurate prediction of the saline-alkali land, improve the informatization combination ability of the saline-alkali land, reduce unnecessary sampling monitoring, reduce the consumption of manpower and material resources for saline-alkali land monitoring, and realize cost reduction and efficiency increase.

[0168] According to the embodiment of the present application, the soil salinity of the sampling area is detected in a real-time monitoring period, and the detection data is introduced into the LSTM prediction model to predict the salt content value, and the prediction salt data of the feasible prediction area group is obtained, specifically:

[0169] Taking a feasible prediction area group as an analysis unit;

[0170] In a real-time monitoring period, the soil salinity of the sampling area in a feasible prediction area group is detected to obtain detection data;

[0171] The detection data is introduced into the LSTM prediction model to predict the salt content value, and the prediction sequence data is obtained;

[0172] The prediction sequence data is parsed, and the data corresponding to each monitoring area in a feasible prediction area group is obtained, and the prediction salt data of each monitoring area in the group is obtained.

[0173] It should be noted that the detection data is specifically salt content data. The salt content data includes the salt content and alkalinity of the soil. In a feasible prediction area group, in addition to the sampling area being the detection data, the remaining monitoring areas correspond to the prediction salt data.

[0174] According to the embodiment of the present application, the prediction salt data and the historical salt content data are used to set the fertilization and irrigation scheme of the multiple monitoring areas, and the intelligent scheme recommendation is made based on different monitoring areas, specifically:

[0175] In a feasible prediction area group, the detection data and the prediction salt data of the sampling area are used to evaluate the salt change of the area, and the salt content control index and the alkalinity control index of the soil are generated;

[0176] Based on the salt content control index and the alkalinity control index of the soil, the fertilization and irrigation plan analysis is performed, and the prediction area fertilization scheme is generated;

[0177] According to multiple feasible prediction area groups, multiple prediction area fertilization schemes are generated;

[0178] In the map model, all monitoring areas outside the feasible prediction area group are marked as non-prediction areas, real-time salt data detection is performed on the non-prediction areas, and the historical salt content data is combined to perform matching selection analysis in the preset fertilization scheme, and multiple real-time fertilization schemes are screened out;

[0179] Based on the multiple prediction area fertilization schemes and the multiple real-time fertilization schemes, the intelligent scheme recommendation is made for the preset salt area.

[0180] It should be noted that in the intelligent scheme recommendation, the predicted regional fertilization scheme can be applied to the fertilization scheme recommendation of the feasible predicted regional group, and the scheme is specifically a scheme dynamically analyzed based on predicted data; the multiple real-time fertilization schemes are fixed optimal schemes, and are suitable for non-predicted regions. The pre-fertilization scheme is a fixed recommended scheme, including multiple pre-fertilization schemes.

[0181] In the analysis of the fertilization and irrigation plan, the quality requirements of the irrigation water source can be determined based on corresponding indexes, including conductivity and pH value, and dynamic regulation and control of irrigation using low-salt water can help to reduce the accumulation of salt in the soil, and the fertilization scheme can set the fertilization amount and fertilizer salt proportion based on the indexes, to regulate the degree of soil salinization.

[0182] The third aspect of the present application also provides a computer readable storage medium, the computer readable storage medium comprises an intelligent recommendation based saline-alkali soil fertilization program, when the intelligent recommendation based saline-alkali soil fertilization program is executed by a processor, the steps of the intelligent recommendation based saline-alkali soil fertilization method are realized.

[0183] The application discloses an intelligent recommendation based saline-alkali soil fertilization method and system. By constructing a map model based on saline-alkali soil, a monitoring area is divided for a preset saline-alkali soil, similarity analysis is performed on environmental characteristic data of each two monitoring areas, and a feasible predicted regional group is determined; in one real-time monitoring period, soil salinity detection is performed on a sampling area, and detection data are imported into an LSTM prediction model for salinity content numerical prediction to obtain predicted salinity data of the feasible predicted regional group; through the predicted salinity data and historical salinity content data, fertilization and irrigation scheme setting is performed on multiple monitoring areas, and intelligent scheme recommendation is performed based on different monitoring areas. Through the application, the salinity content of the saline-alkali soil can be effectively predicted, the human and material resources consumption for soil detection is reduced, and the saline-alkali soil regulation and control capability is improved.

[0184] In several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other manners. The above described device embodiments are only schematic. For example, the division of the units is only a logical function division. There can be another division manner for the actual implementation, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed coupling, direct coupling or communication connection between the components can be indirect coupling or communication connection through some interfaces, devices or units, and can be electrical, mechanical or other forms.

[0185] The units described as separate components above can or can not be physically separate, and the components shown as units can or can not be physical units; they can be located in one place or distributed on multiple network units; and part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0186] In addition, each functional unit in each embodiment of the present application can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be realized in the form of hardware or in the form of hardware plus software functional unit.

[0187] Those skilled in the art can understand that all or part of the steps of the above method embodiments can be completed by program instruction related hardware, and the aforementioned program can be stored in a computer readable storage medium, and the program executes the steps including the above method embodiments when executed; and the aforementioned storage medium includes mobile storage device, read-only memory (ROM), random access memory (RAM), magnetic disc or optical disc, and various storage program codes.

[0188] Alternatively, the integrated unit of the present application, if implemented in the form of a software function module and sold or used as an independent product, can also be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the embodiments of the present application can be embodied in the form of a software product, which is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in the embodiments of the present application. The aforementioned storage medium includes mobile storage devices, ROM, RAM, magnetic discs or optical discs, and various storage program codes.

[0189] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application.

Claims

1. A saline soil fertilization method based on intelligent recommendation, characterized in that, The method comprises the following steps: acquiring preset saline-alkali region information, and constructing a map model based on saline-alkali land through the region information; in the map model, dividing the preset saline-alkali land into multiple monitoring regions through preset monitoring points, and acquiring soil environment data of the monitoring regions; in the soil environment data, corresponding multiple-dimensional feature vectors are generated for multiple environment parameters, and environment feature data of each monitoring region is formed; the environment feature data of the monitoring regions is subjected to similarity analysis, and regions with expected similarity and geographical continuity are marked, and each set of marked continuous regions is taken as a feasible prediction region group, specifically: the environment feature data of the monitoring regions is subjected to similarity calculation, and the calculation process is to calculate the environment feature data of selected two monitoring regions based on a standard Euclidean distance method, and the environment similarity between the two monitoring regions is obtained through the distance value; whether each two monitoring regions has similarity is judged through a preset similarity threshold, all monitoring regions are judged, and monitoring regions with similarity are marked to form a similar monitoring group, and multiple similar monitoring groups are generated based on all monitoring regions; one similar monitoring group comprises at least two monitoring regions; in the similar monitoring group, it is ensured that each monitoring region has similarity with at least one monitoring region in the group; in one similar monitoring group, continuously-geographically monitoring regions are screened, and a new group is formed based on the screening result, and is marked as a feasible prediction region group; all similar monitoring groups are screened and marked to form multiple feasible prediction region groups; in a preset historical time period, the salt content data of the feasible prediction region groups are extracted and sequenced, and the sequenced data is imported into an LSTM prediction model for training, and one monitoring region is selected as a sampling region in each feasible prediction region group, specifically: one feasible prediction region group is taken as an analysis unit; in a preset historical time period, the salt content data of all monitoring regions in the one feasible prediction region group are acquired; the salt content data are sorted based on the geographical continuity order between the monitoring regions in the one feasible prediction region group, and the sorted data are sequenced to form sequenced data; an LSTM prediction model is constructed, and parameters and a loss function are initialized; the sequenced data are divided into a test set and a training set based on a preset proportion, and are imported into the LSTM prediction model for cyclic prediction training, and the model parameters are optimized through the loss function during the training, and the training is performed until the prediction accuracy reaches a preset standard; the specific model parameters of the LSTM prediction model at this time are recorded, and the model parameters are associated with the one feasible prediction region group; in each feasible prediction region group, a monitoring region at a starting position is selected as a sampling region based on the continuity of geographical positions; in a real-time monitoring period, the soil salinity of the sampling region is detected, and the detection data are imported into the LSTM prediction model for salt content numerical prediction to obtain predicted salt data of the feasible prediction region group. The prediction salt and alkali data and historical salt and alkali content data are used to set a fertilization and irrigation scheme for multiple monitoring areas, and intelligent scheme recommendation is performed based on different monitoring areas.

2. The method of claim 1, wherein the method is based on an intelligent recommendation of saline soil fertilization. The preset salt and alkali region information is acquired, and a map model based on the salt and alkali land is constructed based on the region information, specifically as follows. The preset salt and alkali region information is acquired, and the region information includes region area, region contour, planting range and monitoring point position information. A two-dimensional visual map model is constructed based on the region information.

3. The method of claim 1, wherein the method is based on an intelligent recommendation of saline soil fertilization, characterized by, In the map model, the preset monitoring points are used to divide the preset salt and alkali land into multiple monitoring areas, and soil environment data of the monitoring areas is acquired, specifically as follows. In the map model, the preset monitoring point positions are acquired. Based on the preset monitoring point positions, the preset salt and alkali land is divided into regions to form multiple monitoring areas, and it is ensured that each monitoring area corresponds to a preset monitoring point. In a preset historical time period, soil environment data of each monitoring area is collected.

4. The method of claim 3, wherein the method is based on an intelligent recommendation of saline soil fertilization. In the soil environment data, multiple-dimensional feature vectors corresponding to multiple environment parameters are generated, and environment feature data of each monitoring area is formed, specifically as follows. In the soil environment data, each environment parameter is taken as a dimension, and the corresponding parameter value is taken as the value of the corresponding dimension, to generate a multi-dimensional feature vector. The multi-dimensional feature vector is taken as environment feature data, and feature extraction is performed on each monitoring area to form environment feature data of each monitoring area.

5. The method for saline soil fertilization based on intelligent recommendation according to claim 1, characterized in that, In a real-time monitoring period, soil salinity of the sampling areas is detected, and the detection data is input into an LSTM prediction model to predict the salt and alkali content value, to obtain prediction salt and alkali data of a feasible prediction region group, specifically as follows. A feasible prediction region group is taken as an analysis unit. In a real-time monitoring period, soil salinity of the sampling areas in a feasible prediction region group is detected to obtain detection data. The detection data is input into an LSTM prediction model to predict the salt and alkali content value, to obtain prediction sequence data. The prediction sequence data is analyzed, and the data is corresponded to each monitoring area in a feasible prediction region group, to obtain prediction salt and alkali data of each monitoring area in the group.

6. The method of claim 5, wherein the method is based on an intelligent recommendation of saline soil fertilization. The prediction salt and alkali data and historical salt and alkali content data are used to set a fertilization and irrigation scheme for multiple monitoring areas, and intelligent scheme recommendation is performed based on different monitoring areas, specifically as follows. In a feasible prediction region group, the detection data and prediction salt and alkali data of the sampling areas are used to evaluate the salt and alkali change of the region, and a salt content control index and an alkalization degree control index of the soil are generated. Based on the salt content control index and the alkalization degree control index of the soil, fertilization and irrigation plan analysis is performed, and a prediction region fertilization scheme is generated. According to multiple feasible prediction region groups, multiple prediction region fertilization schemes are generated. In the map model, all monitoring areas except the feasible prediction region groups are marked as non-prediction areas, real-time salt and alkali data detection is performed on the non-prediction areas, and matching selection analysis is performed in the fertilization scheme in combination with historical salt and alkali content data, to screen multiple real-time fertilization schemes. Based on multiple prediction area fertilization schemes and multiple real-time fertilization schemes, an intelligent scheme recommendation is performed for a preset saline-alkali area.

7. A saline soil fertilization system based on intelligent recommendation, characterized by, The system comprises a memory and a processor, the memory comprising an intelligent recommendation-based saline-alkali field fertilization program, and the processor implementing the steps of the intelligent recommendation-based saline-alkali field fertilization method as claimed in claim 1 when the program is executed.

8. A computer-readable storage medium, characterized in that, The computer readable storage medium comprises an intelligent recommendation-based saline-alkali field fertilization program, and the processor implements the steps of the intelligent recommendation-based saline-alkali field fertilization method as claimed in any one of claims 1 to 6 when the program is executed.

Images