A system and method for predicting biodegradable mulch degradation cycle based on machine learning

Abstract

The present invention discloses a biodegradable ground film degradation cycle prediction system and method based on machine learning, which belongs to the field of biodegradation technology, including an information acquisition module for acquiring soil property data and ground film data of preset soil, and building a database based on soil properties, while distinguishing soil properties, and also including obtaining the location of the area where the preset soil is located, an LSTM network data acquisition module; the LSTM network data acquisition module responds to the soil property data acquired by the information acquisition module, and is used to collect characteristic data of soil properties in the area where the preset soil is located, and distinguish the collected soil property characteristic data. The present invention utilizes machine deep learning analysis technology, as well as analysis and division of soil data, to better obtain the degradation changes of ground film under various property data conditions, and then more accurately predict the changes in related data in future time periods, so as to predict the degradation trend of ground film.

Description

Technical Field

[0001] The present invention relates to the field of biodegradation technology, and in particular to a system and method for predicting the degradation cycle of biodegradable ground film based on machine learning. Background Art

[0002] With the continuous reduction of non-renewable energy and the worsening of environmental pollution, biodegradable polyester materials have received widespread attention. Biodegradable polyester materials are a general term for polymers obtained by polycondensation of polyols and polyacids. They are a type of engineering plastics with excellent performance and wide applications. They can also be made into polyester fibers and polyester films. As the world's main plastic consumer products, plastic film products are mostly disposable consumer products. Among them, ground film is the one that comes into direct contact with the soil the most. Therefore, there are a large number of waste ground film materials in China's planting fields. Waste ground film materials are not easily degraded by air or microorganisms in a short period of time, occupying a large amount of space, causing environmental and visual pollution. At the same time, due to ultraviolet radiation and external forces, ground film is prone to breakage, forming film fragments, which not only cause serious pollution to farmland, but also pose a potential threat to the ecosystem and human health. Biodegradable is a material that can be degraded and utilized by microorganisms and eventually converted into microbial substances, carbon dioxide and water. Therefore, biodegradable ground film substitutes Substituting PE mulch is one of the main strategies for preventing plastic pollution in agriculture. However, the degradation cycle of biodegradable mulch in cultivated fields is still unclear. This is mainly because its degradation is a complex process and is closely related to material composition and natural conditions. At present, the research on the degradation cycle of biodegradable mulch is mainly based on the experience and / or experimental data of farmland growers. However, due to the different composition of mulch materials, changes in soil moisture, alkalinity, etc. or other human factors, the analysis results based on experience and / or experimental data in the past are accidental and uncertain, resulting in influencing variables in the prediction and analysis of the biodegradation cycle, which reduces the accuracy of the final analysis and prediction of the biodegradation cycle. Summary of the invention

[0003] In view of the above problems existing in the technical field of existing biodegradation calculation prediction, the present invention is proposed.

[0004] Therefore, one of the objects of the present invention is to provide a biodegradable mulch degradation cycle prediction system and method based on machine learning, which utilizes deep learning analysis technology to divide the preset soil data of the collection site into seasonal change data of the soil ecosystem and soil physical and chemical property change data, and based on the soil itself, further distinguish the collected data according to physical property data, chemical property data and biological property data, so as to obtain the degradation changes of the mulch under various property data, and then more accurately predict the changes of relevant data in future time periods, so as to predict the degradation trend of the mulch.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0006] In one aspect, the present invention provides a biodegradable mulch film degradation cycle prediction system based on machine learning, comprising:

[0007] An information acquisition module is used to acquire soil property data and ground film data of a preset soil, and to construct a database based on the soil properties, while distinguishing the soil properties, and also includes acquiring the location of the area where the preset soil is located;

[0008] Data acquisition module; the data acquisition module responds to the soil property data acquired by the information acquisition module, is used to collect characteristic data of soil properties in the area where the preset soil is located, and distinguish the collected soil property characteristic data;

[0009] A fusion analysis unit, used for analyzing the correlation characteristics between the characteristic data in the soil properties, wherein the correlation characteristics also include the correlation characteristics between the characteristic data and the location of the area, and the fusion analysis unit includes an analysis module, a calculation module and a measurement module;

[0010] The analysis module is used to analyze the changing patterns of the soil properties in different months and distinguish the changing patterns, including distinguishing according to seasonal changes in the soil ecosystem and changes in soil physical and chemical properties, and generating a prediction model;

[0011] The calculation module responds to the change rules of soil properties in different months obtained by the analysis module, and is used to calculate the seasonal change data of the soil ecosystem and the change data of soil physical and chemical properties, and obtain characteristic data therein;

[0012] The measuring module is used to divide the preset soil into regions based on the calculation data of the calculation module, and obtain the change characteristics of soil data in different regions, upload the change characteristics to the prediction model, and update the prediction model;

[0013] A degradation data acquisition module, which is based on the change characteristics of regional soil data, is used to obtain the degradation changes of the mulch film under the corresponding change characteristics, and generate a data set based on the degradation changes, and at the same time divide the data set into a test set, a validation set and a training set of the degradation changes;

[0014] A division module, wherein the division module presets at least two critical values ​​in the soil physical and chemical property changes based on the soil physical and chemical property changes in the analysis module, and the critical values ​​are based on at least ten soil physical and chemical property change data collected for the first time, wherein, among the ten soil physical and chemical property change data, a first critical value is preset based on the median data, and a second critical value is preset based on the maximum data, when the soil physical and chemical property changes of the preset soil exceed the first critical value, the system performs one type of soil area division on the preset soil, and when the soil physical and chemical property changes of the preset soil exceed the second critical value, the system performs another type of soil area division on the preset soil.

[0015] As a preferred solution of the present invention, in the information acquisition module, the soil properties include physical properties, chemical properties and biological properties, wherein:

[0016] The physical properties include preset soil texture, soil structure, soil bulk density and specific gravity, soil moisture and soil air;

[0017] The chemical properties include pH and redox properties;

[0018] The ground film data includes ground film material, thickness and ground film flatness;

[0019] The intensity is divided based on the collected soil acidity and alkalinity, and the division method includes division according to pH value, as shown in the following table:

[0020] pH Soil pH ＜4.5 Very acidic 4.5～5.5 Strong acid 5.5～6.5 Acidic 6.5～7.5 neutral 7.5～8.5 Alkaline 8.5～9.5 Strong alkalinity ＞9.5 Very alkaline

[0021] In the preset soil, the correlation between soil acidity change and mulch degradation is analyzed based on the neutral data in the table, and the degradation characteristics of mulch caused by soil acidity change are calculated, which is calculated according to the following formula:

[0022] f=σ(w[w t +x i ]+b); wherein f represents the preset ground film thickness value;

[0023] In the formula, σ represents the given neutral soil acidity, w represents the soil acidity increase value based on the neutral soil acidity, and w t represents the soil acidity increase value collected in the tth period, x i represents the thickness of mulch degradation collected for the i-th time at the specified soil acidity increase value, and b represents the time required for the soil acidity to increase by one level.

[0024] As a preferred embodiment of the present invention, in the calculation module, the seasonal changes of the soil ecosystem include changes in the microbial community in the soil, and the changes in the physical and chemical properties of the soil include pH value, organic matter content, soil bulk density and porosity. The same pH value is given in the preset soil to analyze the changing patterns of different physical property data on the degradation of the mulch film.

[0025] As a preferred solution of the present invention, at least 3 to 5 soil areas of equal area are divided in the preset soil, and the acidity value of each soil area is given in the divided soil area in a data increment manner, the degradation thickness of the ground film under the same time and different acidity values, and the correlation characteristics between the degradation thickness and the acidity value difference are analyzed, and the correlation characteristics are uploaded to the prediction model, and the Q-learning algorithm is used for reinforcement learning to optimize the prediction model.

[0026] As a preferred solution of the present invention, the thickness of the preset soil is divided into θ1, θ2, ..., θ n , where n represents the nth soil thickness. The data characteristics of soil moisture and soil air under different soil thicknesses are collected, and the correlation effects of different data characteristics on the thickness of mulch film degradation are analyzed.

[0027] As a preferred solution of the present invention, the corresponding soil thickness is given based on the maximum thickness data of the collected mulch film degradation, and the data set of soil moisture and soil air is preset. and in, a represents the ath soil moisture data, Among them, the soil moisture data set and soil air data sets They are all increasing and / or decreasing data sets, analyzing the associated effects of different soil moisture and soil air changes on the thickness of mulch film degradation under given soil thickness.

[0028] As a preferred embodiment of the present invention, the soil moisture data set is set at a given soil thickness. and soil air data sets It is divided into several evaluation indicators, and the soil moisture data and soil air data corresponding to the maximum degradation thickness and / or area of ​​the ground film are counted. The data are normalized to calculate the weight of other data except the soil moisture data and soil air data corresponding to the maximum degradation thickness and / or area of ​​the ground film relative to the ground film degradation thickness and / or area.

[0029] As a preferred embodiment of the present invention, it also includes a data monitoring module, which is used to monitor the changes in preset soil physical properties based on the chemical properties and biological properties of the preset soil collected for the first time in different quarters, analyze the correlation between the changes in the preset soil physical properties and the chemical properties and biological properties, and upload the correlation to the prediction model.

[0030] As a preferred embodiment of the present invention, the sunshine, ambient temperature and sea wave height corresponding to the correlation law between the changes in the preset soil physical properties and the chemical properties and biological properties are collected, and the random gradient descent method is used for training to generate the soil physical properties, chemical properties and biological properties data of the preset soil in the future period at the same sunshine, ambient temperature and sea wave height, and the degradation of the ground film is predicted under the generated data conditions.

[0031] On the other hand, the present invention provides a biodegradable mulch film degradation cycle prediction method based on machine learning, which is applied to a biodegradable mulch film degradation cycle prediction system based on machine learning, comprising the following steps:

[0032] Acquire soil property data of a preset soil, ground film data and the location of the preset soil, and construct a database based on the soil properties, while distinguishing the soil properties;

[0033] Collect characteristic data of soil properties in the area where the preset soil is located, distinguish the collected characteristic data of soil properties, and analyze the correlation characteristics between the characteristic data in the soil properties;

[0034] Analyze the changing patterns of the soil properties in different months and differentiate the changing patterns, including differentiating according to seasonal changes in the soil ecosystem and changes in soil physical and chemical properties, and generate a prediction model;

[0035] Calculating seasonal variation data of the soil ecosystem and variation data of soil physical and chemical properties, and obtaining characteristic data therein;

[0036] Dividing the preset soil into regions, obtaining change characteristics of soil data in different regions, uploading the change characteristics to the prediction model, and updating the prediction model;

[0037] Obtaining degradation changes of the ground film under corresponding change characteristics, and generating a data set based on the degradation changes, and dividing the data set into a test set, a validation set, and a training set of the degradation changes;

[0038] The preset soil is divided into at least 3 to 5 soil areas of equal area, and the acidity value of each soil area is given in the divided soil area in a data increment manner, and the degradation thickness of the ground film under the same time and different acidity values, as well as the correlation characteristics between the degradation thickness and the acidity value difference are analyzed.

[0039] Beneficial effects:

[0040] The present invention utilizes deep learning analysis technology to divide preset soil data at the collection site into seasonal change data of the soil ecosystem and soil physical and chemical property change data, and based on the soil itself, further distinguishes the collected data according to physical property data, chemical property data and biological property data, and at the same time, based on the physical property data, chemical property data and biological property data, gives the same and / or different soil thicknesses to monitor changes in soil physical property data, chemical property data and biological property data, so as to obtain the degradation changes of the ground film under various property data conditions. At the same time, the present invention adopts the stochastic gradient descent method to train and obtain the sunshine, ambient temperature and sea wave height corresponding to the changes in physical property data, chemical property data and biological property data, so as to more accurately predict the changes in relevant data in future time periods, so as to predict the degradation trend of the ground film. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0042] Figure 1 A schematic diagram of a method flow of an embodiment of the present invention;

[0043] Figure 2 This is a schematic diagram of the modular structure of a biodegradable polyester degradation cycle prediction system based on machine learning according to an embodiment of the present invention;

[0044] Figure 3 It is a schematic diagram of the process structure of an embodiment of the present invention;

[0045] Numbers in the figure: 110 - information acquisition module; 120 - LSTM network data acquisition module; 130 - fusion analysis unit; 1301 - analysis module; 1302 - measurement module; 1303 - calculation module; 140 - degradation data acquisition module; 150 - division module; 160 - data monitoring module. DETAILED DESCRIPTION

[0046] In order to make the purpose, technical solution and advantages of the embodiment of the present invention clearer, the technical solution of the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings of the embodiment of the present invention. Obviously, the described embodiment is a part of the embodiment of the present invention, not all of the embodiments. Based on the described embodiment of the present invention, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the present invention.

[0047] Since the current research on the degradation cycle of biodegradable mulch is mainly based on the experience and / or experimental data of farmland growers, and due to the different composition of mulch materials, changes in soil moisture, alkalinity, etc. or other human factors, the analysis results based on experience and / or experimental data in the past are accidental and uncertain, resulting in influencing variables in the predictive analysis of the biodegradation cycle, reducing the accuracy of the final analysis and prediction of the biodegradation cycle.

[0048] Based on this, the present invention proposes a biodegradable mulch degradation cycle prediction system and method based on machine learning, which uses machine deep learning analysis technology to divide the preset soil data of the collection site into seasonal change data of the soil ecosystem and soil physical and chemical property change data, and based on the soil itself, further distinguish the collected data according to physical property data, chemical property data and biological property data, so as to obtain the degradation changes of the mulch under various property data, and then more accurately predict the changes of relevant data in future time periods, so as to predict the degradation trend of the mulch.

[0049] The present invention is further described in detail below through embodiments and in conjunction with the accompanying drawings.

[0050] Reference Figures 1 to 3 , is an embodiment of the present invention, which provides a biodegradable mulch film degradation cycle prediction system based on machine learning, comprising:

[0051] The information acquisition module 110 is used to obtain soil property data and ground film data of the preset soil, and to build a database based on the soil properties, and to distinguish the soil properties, and also includes obtaining the location of the area where the preset soil is located;

[0052] It should be noted that the soil properties include physical properties, chemical properties and biological properties, among which:

[0053] Physical properties include pre-determined soil texture, soil structure, soil bulk density and specific gravity, soil moisture, and soil air;

[0054] Chemical properties include pH and redox properties;

[0055] Ground membrane data include ground membrane material, thickness and ground membrane flatness;

[0056] The intensity is divided based on the collected soil acidity and alkalinity, including division according to pH value, as shown in the following table:

[0057] pH Soil pH ＜4.5 Very acidic 4.5～5.5 Strong acid 5.5～6.5 Acidic 6.5～7.5 neutral 7.5～8.5 Alkaline 8.5～9.5 Strong alkalinity ＞9.5 Very alkaline

[0058] In the preset soil, the correlation between soil acidity change and mulch degradation is analyzed based on the neutral data in the table, and the degradation characteristics of mulch film due to soil acidity change are calculated according to the following formula:

[0059] f=σ(w[w t +x i ]+b); wherein f represents the preset ground film thickness value;

[0060] In the formula, σ represents the given neutral soil acidity, w represents the soil acidity increase based on the neutral soil acidity, and w t represents the soil acidity increase value collected in the tth period, x i represents the thickness of mulch degradation collected for the i-th time at a specified soil acidity increase value, and b represents the time required for soil acidity to increase by one level;

[0061] Data acquisition module 120; the data acquisition module responds to the soil property data acquired by the information acquisition module, and is used to collect characteristic data of soil properties in the area where the preset soil is located, and distinguish the collected soil property characteristic data;

[0062] This embodiment further divides the preset soil into at least 3 to 5 equal-area soil regions, and gives the acidity value of each soil region in the divided soil region in a data increment manner, analyzes the degradation thickness of the mulch film under the same time and different acidity values, and the correlation characteristics between the degradation thickness and the acidity difference, and uploads the correlation characteristics to the prediction model, and uses the Q-learning algorithm for reinforcement learning to optimize the prediction model;

[0063] Based on the above, this embodiment divides the preset soil thickness into θ1, θ2, ..., θ n , where n represents the nth soil thickness. The data characteristics of soil moisture and soil air under different soil thicknesses are collected to analyze the correlation effect of different data characteristics on the thickness of mulch film degradation;

[0064] In this embodiment, the corresponding soil thickness is given based on the maximum thickness data of the collected ground film degradation, and the data set of soil moisture and soil air is preset. and in, a represents the ath soil moisture data, Among them, the soil moisture data set and soil air data sets All are increasing and / or decreasing data sets, analyzing the correlation effects of different soil moisture and soil air changes on the thickness of mulch film degradation under given soil thickness;

[0065] Among them, it also includes the corresponding soil thickness given based on the maximum area data of mulch degradation collected;

[0066] Past planting experience and degradation experience show that the thickness of mulch is not the same. The thickness of mulch can be adjusted according to different needs and application scenarios. According to national regulations, the thickness of mulch is 0.014 mm, micro-film is 0.008 mm, and ultra-micro-film is 0.005 mm. The types of mulch are divided into ordinary mulch and colored mulch;

[0067] Among them, ordinary ground film is colorless and transparent, with good light transmittance. After covering, it can increase the ground temperature by 2 to 4 degrees Celsius and has a wide range of applications.

[0068] Colored mulch films can be made by adding colored substances to polyethylene resin to produce mulch films with different colors. For example, the thickness of black mulch films is 0.01-0.03 mm, and the light transmittance is only 10%. It can effectively inhibit the growth of weeds and can be used in areas with many weeds to save weed control costs.

[0069] The green mulch film is 0.015 mm thick, which can inhibit and kill weeds under the film and is beneficial to the growth of the above-ground parts of crops such as eggplant and sweet pepper.

[0070] Silver-gray mulch, also known as anti-aphid mulch, has a thickness of 0.015-0.02 mm and a high reflectivity to ultraviolet rays. It can repel pests such as aphids, yellow flea beetles, weevils and yellow squash beetles and reduce crop viral diseases. It can also inhibit weed growth and maintain soil moisture. It is suitable for disease-proof and heat-resistant cultivation in spring, summer and autumn.

[0071] Therefore, in real life, the mulch films acting on the planting soil are different. However, no matter what kind of mulch film is degraded in the soil, it is necessary to collect and analyze the degradation thickness and / or area change of the mulch film under different soil thickness conditions. The data obtained is beneficial for analyzing the degradation thickness and / or area change of the mulch film.

[0072] This embodiment also includes analyzing the correlation effect of different soil acidity changes on the thickness and / or area of ​​mulch film degradation under a given soil thickness;

[0073] At the same time, this embodiment collects soil moisture data under a given soil thickness. and soil air data sets It is divided into several evaluation indicators, and the soil moisture data and soil air data corresponding to the maximum degradation thickness and / or area of ​​the ground film are statistically obtained. The data are normalized, and the weight of other data except the soil moisture data and soil air data corresponding to the maximum degradation thickness and / or area of ​​the ground film is calculated to account for the degradation thickness and / or area of ​​the ground film;

[0074] The fusion analysis unit 130 is used to analyze the correlation characteristics between each characteristic data in the soil property, and the correlation characteristics also include the correlation characteristics between each characteristic data and the location of the region. The fusion analysis unit includes an analysis module 1301, a calculation module 1302 and a measurement module 1303;

[0075] The analysis module 1301 is used to analyze the changing patterns of soil properties in different months and distinguish the changing patterns, including distinguishing according to seasonal changes of soil ecosystems and changes in soil physical and chemical properties, and generate a prediction model;

[0076] The calculation module 1302 responds to the change rules of soil properties in different months from the analysis module 1301, and is used to calculate the seasonal change data of the soil ecosystem and the change data of soil physical and chemical properties, and obtain the characteristic data therein;

[0077] Specifically, in this embodiment, seasonal changes in the soil ecosystem include changes in the microbial community in the soil, and changes in the physical and chemical properties of the soil include pH value, organic matter content, soil bulk density and porosity. The same pH value is given in the preset soil, and the change law of different physical property data on the degradation of the mulch film is analyzed;

[0078] Experience with land reclamation and planting shows that beetle larvae in the soil will undergo seasonal vertical movements in search of a better living environment. In addition, the soil undergoes many changes during its formation, so the animals that inhabit the soil will also change accordingly.

[0079] At the same time, land reclamation and planting experience show that soil physical and chemical properties, including soil pH, organic matter content, soil bulk density and porosity, will change with the change of seasons. Among them, the average dissolution in summer soil is 3.87 times that in winter, which shows that there is a good positive correlation between water, heat and seasonal changes in dissolution. Soil CO2 can promote dissolution more in summer with high precipitation. In addition, soil organic matter is negatively correlated with soil pH, and the two have a synergistic effect on dissolution. The influence of soil texture reflected by soil bulk density and soil porosity on dissolution is bidirectional. Clayey soil has good water retention and poor permeability, which can promote dissolution in winter with less precipitation, while sandy soil has good permeability and poor water retention, which is beneficial to dissolution in summer with more precipitation.

[0080] The measurement module 1303 is used to divide the preset soil into regions based on the calculation data of the calculation module 1302, and obtain the change characteristics of the soil data in different regions, upload the change characteristics to the prediction model, and update the prediction model;

[0081] The degradation data acquisition module 140 is used to obtain the degradation changes of the mulch film under the corresponding change characteristics based on the change characteristics of the regional soil data, and to generate a data set based on the degradation changes, and at the same time divide the degradation changes into a test set, a validation set and a training set in the data set;

[0082] The division module 150 presets at least two critical values ​​in the soil physical and chemical property changes based on the soil physical and chemical property changes in the analysis module 1301. The critical values ​​are based on at least ten soil physical and chemical property change data collected for the first time, wherein, among the ten soil physical and chemical property change data, a first critical value is preset based on the median data, and a second critical value is preset based on the maximum data. When the soil physical and chemical property change of the preset soil exceeds the first critical value, the system performs one type of soil area division on the preset soil, and when the soil physical and chemical property change of the preset soil exceeds the second critical value, the system performs another type of soil area division on the preset soil;

[0083] It should be noted that in this embodiment, when the change in the physical and chemical properties of the soil does not exceed the first critical value, the system does not divide the preset soil into regions;

[0084] The present embodiment further includes a data monitoring module 160, which is used to monitor changes in preset soil physical properties based on the chemical properties and biological properties of the preset soil collected for the first time in different quarters, analyze the correlation between the changes in the preset soil physical properties and the chemical properties and biological properties, and upload the correlation to the prediction model;

[0085] On the basis of the above, the present embodiment further collects the sunshine, ambient temperature and sea wave height corresponding to the correlation law between the changes in the preset soil physical properties and the chemical properties and biological properties, and uses the stochastic gradient descent method for training to generate the soil physical properties, chemical properties and biological properties data of the preset soil in the future time period when the soil is at the same sunshine, ambient temperature and sea wave height, and predicts the degradation of the ground film under the generated data conditions.

[0086] Based on the above, this application uses deep learning analysis technology to divide the preset soil data at the collection site into seasonal change data of the soil ecosystem and soil physical and chemical property change data, and based on the soil itself, further distinguish the collected data according to physical property data, chemical property data and biological property data, so as to obtain the degradation changes of the ground film under various property data, and then more accurately predict the changes in relevant data in future time periods, and then predict the degradation trend of the ground film.

[0087] This embodiment combines the above-mentioned biodegradable mulch degradation cycle prediction system based on machine learning and also proposes a working method of the system, as follows:

[0088] S10: obtaining soil property data of the preset soil, ground film data and the location of the preset soil, and building a database based on the soil properties, while distinguishing the soil properties;

[0089] S20: collecting characteristic data of soil properties in the area where the preset soil is located, distinguishing the collected characteristic data of soil properties, and analyzing correlation characteristics between characteristic data in the soil properties;

[0090] S30: Analyze the changing patterns of soil properties in different months and differentiate the changing patterns, including seasonal changes of soil ecosystems and changes in soil physical and chemical properties, and generate a prediction model;

[0091] S40: Calculate the seasonal variation data of soil ecosystem and the variation data of soil physical and chemical properties, and obtain the characteristic data therein;

[0092] S50: Divide the preset soil into regions, obtain change characteristics of soil data in different regions, upload the change characteristics to the prediction model, and update the prediction model;

[0093] S60: Obtaining degradation changes of the mulch film under corresponding change characteristics, generating a data set based on the degradation changes, and dividing the data set into a test set, a validation set, and a training set for the degradation changes;

[0094] S70: Divide the preset soil into at least 3 to 5 equal-sized soil areas, and assign the acidity value of each soil area in the divided soil area in a data increment manner, analyze the degradation thickness of the ground film under the same time and different acidity values, and the correlation characteristics between the degradation thickness and the acidity value difference.

[0095] In summary, the present invention uses deep learning analysis technology to divide the preset soil data of the collection site into seasonal change data of the soil ecosystem and soil physical and chemical property change data, and based on the soil itself, further distinguishes the collected data according to physical property data, chemical property data and biological property data. At the same time, based on the physical property data, chemical property data and biological property data, the same and / or different soil thicknesses are given to monitor the changes in soil physical property data, chemical property data and biological property data, so as to obtain the degradation changes of the ground film under various property data conditions. At the same time, the present invention adopts the stochastic gradient descent method to train and obtain the sunshine, ambient temperature and sea wave height corresponding to the changes in physical property data, chemical property data and biological property data, so as to more accurately predict the changes in related data in future time periods, so as to predict the degradation trend of the ground film.

[0096] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A biodegradable mulch film degradation cycle prediction system based on machine learning, characterized in that: include: An information acquisition module is used to acquire soil property data and ground film data of a preset soil, and to construct a database based on the soil properties, while distinguishing the soil properties, and also includes acquiring the location of the area where the preset soil is located; Data acquisition module; the data acquisition module responds to the soil property data acquired by the information acquisition module, is used to collect characteristic data of soil properties in the area where the preset soil is located, and distinguish the collected soil property characteristic data; A fusion analysis unit, used for analyzing the correlation characteristics between the characteristic data in the soil properties, wherein the correlation characteristics also include the correlation characteristics between the characteristic data and the location of the area, and the fusion analysis unit includes an analysis module, a calculation module and a measurement module; The analysis module is used to analyze the changing patterns of the soil properties in different months and distinguish the changing patterns, including distinguishing according to seasonal changes in the soil ecosystem and changes in soil physical and chemical properties, and generating a prediction model; The calculation module responds to the change rules of soil properties in different months obtained by the analysis module, and is used to calculate the seasonal change data of the soil ecosystem and the change data of soil physical and chemical properties, and obtain characteristic data therein; The measuring module is used to divide the preset soil into regions based on the calculation data of the calculation module, and obtain the change characteristics of soil data in different regions, upload the change characteristics to the prediction model, and update the prediction model; A degradation data acquisition module, which is based on the change characteristics of regional soil data, is used to obtain the degradation changes of the mulch film under the corresponding change characteristics, and generate a data set based on the degradation changes, and at the same time divide the data set into a test set, a validation set and a training set of the degradation changes; A division module, wherein the division module presets at least two critical values ​​in the soil physical and chemical property changes based on the soil physical and chemical property changes in the analysis module, and the critical values ​​are based on at least ten soil physical and chemical property change data collected for the first time, wherein, among the ten soil physical and chemical property change data, a first critical value is preset based on the median data, and a second critical value is preset based on the maximum data, when the soil physical and chemical property changes of the preset soil exceed the first critical value, the system performs one type of soil area division on the preset soil, and when the soil physical and chemical property changes of the preset soil exceed the second critical value, the system performs another type of soil area division on the preset soil.

2. The biodegradable mulch film degradation cycle prediction system based on machine learning as claimed in claim 1, characterized in that: In the information acquisition module, the soil properties include physical properties, chemical properties and biological properties, wherein: The physical properties include preset soil texture, soil structure, soil bulk density and specific gravity, soil moisture and soil air; The chemical properties include pH and redox properties; The ground film data includes ground film material, thickness and ground film flatness; The intensity is divided based on the collected soil acidity and alkalinity, including division according to pH value, as shown in the following table: In the preset soil, the correlation between soil acidity change and mulch degradation is analyzed based on the neutral data in the table, and the degradation characteristics of mulch caused by soil acidity change are calculated, which is calculated according to the following formula: f=σ(w[w t +x i ]+b); wherein f represents the preset ground film thickness value; In the formula, σ represents the given neutral soil acidity, w represents the soil acidity increase value based on the neutral soil acidity, and w t represents the soil acidity increase value collected in the tth period, x i represents the thickness of mulch degradation collected for the i-th time at the specified soil acidity increase value, and b represents the time required for the soil acidity to increase by one level.

3. The biodegradable mulch film degradation cycle prediction system based on machine learning as claimed in claim 1, characterized in that: In the calculation module, the seasonal changes of the soil ecosystem include changes in the microbial community in the soil, and the changes in the physical and chemical properties of the soil include pH value, organic matter content, soil bulk density and porosity. The same pH value is given in the preset soil, and the changing patterns of different physical property data on the degradation of the ground film are analyzed.

4. The biodegradable mulch film degradation cycle prediction system based on machine learning as claimed in claim 1, characterized in that: The preset soil is divided into at least 3 to 5 soil areas of equal area, and the acidity value of each soil area is given in the divided soil area in a data increment manner, the degradation thickness of the ground film under the same time and different acidity values, and the correlation characteristics between the degradation thickness and the acidity value difference are analyzed, and the correlation characteristics are uploaded to the prediction model, and the Q-learning algorithm is used for reinforcement learning to optimize the prediction model.

5. The biodegradable mulch film degradation cycle prediction system based on machine learning as claimed in claim 1, characterized in that: The thickness of the preset soil is divided into θ1, θ2, ..., θ n , where n represents the nth soil thickness. The data characteristics of soil moisture and soil air under different soil thicknesses are collected, and the correlation effects of different data characteristics on the thickness of mulch film degradation are analyzed.

6. The biodegradable mulch film degradation cycle prediction system based on machine learning as claimed in claim 5, characterized in that: The corresponding soil thickness is given based on the maximum thickness data of the collected mulch degradation, and the data set of soil moisture and soil air is preset and in, a represents the ath soil moisture data, Among them, the soil moisture data set and soil air data sets They are all increasing and / or decreasing data sets, analyzing the associated effects of different soil moisture and soil air changes on the thickness of mulch film degradation under given soil thickness.

7. The biodegradable mulch film degradation cycle prediction system based on machine learning as claimed in claim 6, characterized in that: Soil moisture data set at a given soil thickness and soil air data sets It is divided into several evaluation indicators, and the soil moisture data and soil air data corresponding to the maximum degradation thickness and / or area of ​​the ground film are counted. The data are normalized to calculate the weight of other data except the soil moisture data and soil air data corresponding to the maximum degradation thickness and / or area of ​​the ground film relative to the ground film degradation thickness and / or area.

8. The biodegradable mulch film degradation cycle prediction system based on machine learning as claimed in claim 2, characterized in that: It also includes a data monitoring module, which is used to monitor the changes in preset soil physical properties based on the chemical properties and biological properties of the preset soil collected for the first time in different quarters, analyze the correlation between the changes in the preset soil physical properties and the chemical properties and biological properties, and upload the correlation to the prediction model.

9. The biodegradable mulch film degradation cycle prediction system based on machine learning as claimed in claim 8, characterized in that: The sunshine, ambient temperature and sea wave height corresponding to the correlation between the changes in the preset soil physical properties and the chemical properties and biological properties are collected, and the random gradient descent method is used for training to generate the soil physical properties, chemical properties and biological properties data of the preset soil at the same sunshine, ambient temperature and sea wave height in the future period, and the degradation of the ground film is predicted under the generated data.

10. A method for predicting the degradation cycle of biodegradable mulch based on machine learning, applied to the system for predicting the degradation cycle of biodegradable mulch based on machine learning as claimed in claim 1, characterized in that: The following steps are involved: Acquire soil property data of a preset soil, ground film data and the location of the preset soil, and construct a database based on the soil properties, while distinguishing the soil properties; Collect characteristic data of soil properties in the area where the preset soil is located, distinguish the collected characteristic data of soil properties, and analyze the correlation characteristics between the characteristic data in the soil properties; Analyze the changing patterns of the soil properties in different months and differentiate the changing patterns, including differentiating according to seasonal changes in the soil ecosystem and changes in soil physical and chemical properties, and generate a prediction model; Calculating seasonal variation data of the soil ecosystem and variation data of soil physical and chemical properties, and obtaining characteristic data therein; Dividing the preset soil into regions, obtaining change characteristics of soil data in different regions, uploading the change characteristics to the prediction model, and updating the prediction model; Obtaining degradation changes of the ground film under corresponding change characteristics, and generating a data set based on the degradation changes, and dividing the data set into a test set, a validation set, and a training set of the degradation changes; The preset soil is divided into at least 3 to 5 soil areas of equal area, and the acidity value of each soil area is given in the divided soil area in a data increment manner, and the degradation thickness of the ground film under the same time and different acidity values, as well as the correlation characteristics between the degradation thickness and the acidity value difference are analyzed.

Images