Methods and systems for executing machine learning algorithms
By designing a machine learning model storage, execution and management platform for agriculture, the problem of difficulty in providing customized recommendations and model management in existing systems is solved, and the specific recommendations and model management of seed selection and planting are simplified and scalable. .
Patent Information
- Application Number
- CN201980076419.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Priority Date
- 2018-09-21
- Filing Date
- 2019-09-18
- Publication Date
- 2024-11-05
- Estimated Expiration
- 2039-09-18
AI Technical Summary
Existing agricultural model systems are difficult to provide customized seed selection and planting recommendations, and they face version tracking and cataloging difficulties when managing and maintaining customized models.
Designed a computer-implemented platform, including a model repository, model execution infrastructure and model history log, to store, catalog, execute and manage machine learning models, support model version management and history, and provide a graphical user interface for Browse, select and execute the model.
It realizes specific seed selection and planting recommendations for crop growers, simplifies model management and maintenance, and improves the scalability and maintainability of the system.
Smart Images

Figure CN113168577B_ABST
Abstract
Description
[0001] Copyright Notice
[0002] A portion of the disclosure of this patent document contains material that is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction by anyone of the patent document or the patent disclosure, as it appears in the Patent and Trademark Office patent file or records, but otherwise reserves all copyright rights whatsoever. Climate Corporation. Technical Field
[0003] One technical area of the present disclosure is to provide a computer-assisted platform for executing machine learning algorithms. Another technical area is to classify, catalog, execute and monitor machine learning models and provide model scaling capabilities for executing models. Another technical area is a computer-implemented technology for browsing, selecting and executing machine learning models and using outputs generated by the models to control agricultural equipment. Background Art
[0004] The methods described in this section are methods that may be protected, but are not necessarily methods that have been previously conceived or protected. Therefore, unless otherwise indicated, it should not be assumed that any method described in this section qualifies as prior art merely by virtue of its inclusion in this section.
[0005] In the context of agriculture, crop growers often seek recommendations for seed selection and seed planting for their fields. These recommendations can be customized based on, for example, the location, climate and soil type of the field, as well as the grower's goals.
[0006] Typically, seed and planting recommendations are generated by complex computer-based systems that execute digital models on large amounts of agricultural data and require significant computing resources. However, these systems rarely offer customization capabilities. For example, some systems may not be able to configure and scale the models to generate sufficiently specific recommendations.
[0007] Furthermore, even if customization of agricultural models is achieved, customization of the models may result in obtaining such a large number of customized models that management and maintenance of the models may be problematic. For example, it may be difficult to track and catalog the various versions of the models and make them easily accessible to farmers. Summary of the invention
[0008] The following claims may serve as a summary of the disclosure. BRIEF DESCRIPTION OF THE DRAWINGS
[0009] In the attached picture:
[0010] Figure 1An example computer system configured to perform the functions described herein is shown in a field environment along with other devices with which the system can interact.
[0011] Figure 2 Two views of an example logical organization of a set of instructions in main memory when an example mobile application is loaded for execution are shown.
[0012] Figure 3 A programming process is shown by which an agricultural intelligence computer system generates one or more preconfigured agronomic models using agronomic data provided by one or more data sources.
[0013] Figure 4 is a block diagram illustrating a computer system upon which embodiments of the invention may be implemented.
[0014] Figure 5 An example embodiment of a timeline view for data entry is depicted.
[0015] Figure 6 An example embodiment of a spreadsheet view for data entry is depicted.
[0016] Fig. 7A An exemplary computer-based platform for executing a machine learning algorithm is depicted.
[0017] Figure 7B An exemplary computer-based model repository is depicted.
[0018] Figure 7C Depicted are example data structures maintained by a computer-based model repository.
[0019] Fig.7D Depicted is an example flow chart for building, storing, and cataloging machine-based models.
[0020] Fig. 8A An exemplary computer-based model execution infrastructure for executing machine learning algorithms is depicted.
[0021] Figure 8B Depicted is an exemplary flowchart for selecting and executing an example machine learning model.
[0022] Fig. 9 A diagram depicting an example user interface for executing a machine learning algorithm. DETAILED DESCRIPTION
[0023] In the following description, for the purpose of explanation, many specific details are set forth in order to provide a thorough understanding of the present disclosure. However, it will be apparent that embodiments can be practiced without these specific details. In other instances, well-known structures and devices are shown in block diagram form to avoid unnecessarily obscuring the present disclosure. According to the following summary, embodiments are disclosed in various sections:
[0024] 1. Overview
[0025] 2. Example Agricultural Intelligent Computer System
[0026] 2.1. Structural Overview
[0027] 2.2. Application Overview
[0028] 2.3. Data Ingestion into Computer Systems
[0029] 2.4. Process Overview - Agricultural Model Training
[0030] 2.5. Implementation Example - Hardware Overview
[0031] 3. Systems for executing machine learning algorithms
[0032] 3.1. Example Model Repository
[0033] 3.2. Example Model Execution Infrastructure
[0034] 3.3. Example Model History Log
[0035] 4. Scalability considerations
[0036] 5. Example Implementation
[0037] 6. Benefits of Certain Embodiments
[0038] 7. Extensions and Replacements
[0039] ***
[0040] 1. General Overview
[0041] In one embodiment, computer-implemented methods, computer systems, and computer program products for executing machine learning algorithms are described. They may include a computer-based platform for storing, cataloging, and executing machine learning models. The platform is programmed to catalog common definitions and metadata of models, store models in a repository, browse and select models from the repository, and abstract execution components to facilitate execution of models in a repeatable manner. Examples of machine-based models include operational research models, machine learning models, deep learning models, and physical models.
[0042] In one embodiment, a method for executing machine learning algorithms provides functionality for customizing agricultural models, cataloging, maintaining, and managing a large number of customized models, and facilitating the execution of models to provide specific recommendations to crop growers. The method for storing, executing, and cataloging machine-based models allows for the deployment of different versions of models, adding new models and new versions of models to a model repository, and provides a hardware / software runtime infrastructure for executing models.
[0043] In one embodiment, a platform for storing, executing, and cataloging machine-based models includes a model repository, a model execution infrastructure platform, and a model history log.
[0044] A model repository is a computer-based digital data repository configured to store digital data models. These models may have different configurations and different execution requirements. For each model, the repository may store, for example, a model definition and model metadata that captures the components and characteristics of the model. The model definition and metadata associated with a model may include information about the model's name, inputs, outputs, infrastructure requirements, version, status, etc. The model repository is searchable, modifiable, and browsable.
[0045] The model execution infrastructure platform includes hardware and / or software components that collectively provide a platform for executing machine-based models. The components of the model execution infrastructure are referred to herein as common infrastructure components. These components are the building blocks for executing models stored in the model repository.
[0046] The model history log is a computer-based digital log that contains information about the execution of a machine-based model. The log of a model can include information about each execution instance performed for the model. This can include information about versions, inputs, outputs, execution requirements, modifications and updates, execution dates, statistics, and other information related to the execution of the model.
[0047] In one embodiment, a method for executing a machine learning algorithm is implemented for agricultural applications. The algorithm can be configured, for example, to solve agricultural related problems and provide answers to queries requesting agricultural recommendations. For example, the method can allow a crop grower to browse models stored in a model repository, select a model from the repository, provide input for the selected model, initiate execution of the selected model on a model infrastructure platform to generate recommendations for the grower, and receive recommendations from the platform.
[0048] The machine learning algorithm may include an implementation of a neural network that is configured to receive input, for example, from a crop grower, process the received input, and generate recommendations for the grower. The recommendations may include, for example, recommendations about seed types and seed planting times that are beneficial to the grower based on input provided by the grower. The input may include information about soil, climate, and the location of the grower's field. The recommendations may be transmitted from the neural network to one or more computing devices available to the crop grower. For example, the recommendations may be transmitted from the neural network to a mobile device of the crop grower that is used to communicate with the network.
[0049] According to another example, a machine learning algorithm may include an implementation of a neural network configured to receive a seed planting request from a crop grower. The neural network may process the request and generate sowing instructions for sowing seeds using agricultural equipment such as a seed drill and a planter. The planting instructions may be transmitted to the grower's mobile device. The planting instructions may also be transmitted to a computer or controller mounted on the agricultural machinery. The controller may use the instructions to control the behavior of the machine as it performs seed planting in the field.
[0050] In one embodiment, a method for executing a machine learning algorithm implements a graphical user interface. The graphical user interface can be configured to access a model repository, a model execution infrastructure platform, and a model history log. The graphical user interface can be used, for example, to issue queries to the repository, facilitate browsing the contents of the repository, display information about models stored in the repository, receive a selection of a model from the models stored in the repository, facilitate providing input to the selected model, facilitate execution of the selected model, and display output generated by the executed model.
[0051] 2. Example Agricultural Intelligent Computer System
[0052] 2.1. Structural Overview
[0053] Figure 1 An example computer system configured to perform the functions described herein is shown in a field environment along with other devices with which the system can interoperate. In one embodiment, a user 102 owns, operates, or possesses a field manager computing device 104 in or associated with a field location, such as a field for agricultural activities or a management location for one or more agricultural fields. The field manager computer device 104 is programmed or configured to provide field data 106 to an agricultural intelligence computer system 130 via one or more networks 109.
[0054] Examples of field data 106 include: (a) identification data (e.g., area, field name, field identifier, geographic identifier, boundary identifier, crop identifier, and any other suitable data that can be used to identify agricultural fields, such as common land units (CLUs), lot and block numbers, parcel numbers, geographic coordinates and boundaries, farm serial numbers (FSNs), farm numbers, zone numbers, field numbers, regions, townships, and / or ranges); (b) harvest data (e.g., crop type, crop variety, crop rotation, whether the crop is grown organically, harvest date, actual production history (APH), expected yield, yield, crop price, crop income, grain moisture, tillage practices, and previous growing season information); (c) soil data (e.g., type, composition, pH, organic matter (OM), cation exchange capacity (CEC)); (d) planting data (e.g., planting date, type of one or more seeds, relative maturity (RM) of one or more seeds planted, seed population); (e) fertilizer data (e.g., nutrient type (nitrogen, phosphorus, potassium) , type of application, date of application, amount, source, method); (f) chemical application data (e.g., pesticides, herbicides, fungicides, other substances or mixtures of substances intended for use as plant regulators, defoliants or desiccants, date of application, amount, source, method); (g) irrigation data (e.g., date of application, amount, source, method); (h) weather data (e.g., precipitation, rainfall rate, predicted rainfall, water runoff rate area, temperature, wind, forecast, pressure, visibility, clouds, heat index, dew point, humidity, snow depth, air quality , sunrise, sunset); (i) imagery data (e.g., images and spectral information from agricultural device sensors, cameras, computers, smart phones, tablets, unmanned aerial vehicles, aircraft, or satellites); (j) reconnaissance observations (photos, videos, free-form notes, voice recordings, voice transcriptions, weather conditions (temperature, precipitation (current and over time), soil moisture, crop growth stage, wind speed, relative humidity, dew point, dark layer)); and (k) soil, seed, crop phenology, pest and disease reports, and forecast sources and databases.
[0055] The data server computer 108 is communicatively coupled to the agricultural intelligent computer system 130 and is programmed or configured to send external data 110 to the agricultural intelligent computer system 130 via one or more networks 109. The external data server computer 108 may be owned or operated by the same legal person or entity as the agricultural intelligent computer system 130, or owned or operated by a different person or entity (such as a government agency, a non-governmental organization (NGO) and / or a private data service provider). Examples of external data include weather data, image data, soil data, or statistics related to crop yields, etc. The external data 110 may consist of the same type of information as the field data 106. In some embodiments, the external data 110 is provided by an external data server 108 owned by the same entity that owns and / or operates the agricultural intelligent computer system 130. For example, the agricultural intelligent computer system 130 may include a data server that is specifically focused on a type of data (such as weather data) that may otherwise be obtained from a third-party source. In some embodiments, the external data server 108 may actually be incorporated into the system 130.
[0056] The agricultural device 111 may have one or more remote sensors 112 fixed thereon, which are directly or indirectly communicatively coupled to the agricultural intelligent computer system 130 via the agricultural device 111 and are programmed or configured to send sensor data to the agricultural intelligent computer system 130. Examples of agricultural devices 111 include: tractors, combines, harvesters, planters, trucks, fertilizer equipment, aircraft including unmanned aerial vehicles, and any other physical machinery or hardware items that are generally mobile machinery and can be used for tasks associated with agriculture. In some embodiments, a single unit of the device 111 may include multiple sensors 112 locally coupled in a network on the device; a controller area network (CAN) is an example of such a network that can be installed in combines, harvesters, sprayers, and cultivators. The application controller 114 is communicatively coupled to the agricultural intelligent computer system 130 via one or more networks 109 and is programmed or configured to receive one or more scripts for controlling the operating parameters or implementation of the agricultural vehicle from the agricultural intelligent computer system 130. For example, a controller area network (CAN) bus interface may be used to enable communication from the agricultural intelligence computer system 130 to the agricultural device 111, such as how the CLIMATE FIELDVIEW DRIVE available from the Climate Corporation of San Francisco, California may be used. The sensor data may consist of the same type of information as the field data 106. In some embodiments, the remote sensor 112 may not be fixed to the agricultural device 111, but may be remotely located in the field and may communicate with the network 109.
[0057] The device 111 may include a cab computer 115 programmed with a cab application, which may include a version or variation of a mobile application for the device 104 further described elsewhere herein. In one embodiment, the cab computer 115 includes a compact computer, typically a tablet-sized computer or smart phone, with a graphical screen display, such as a color display, mounted in the cab of the operator of the device 111. The cab computer 115 may implement some or all of the operations and functions further described herein for the mobile computer device 104.
[0058] One or more networks 109 broadly represent any combination of one or more data communication networks including a local area network, a wide area network, the Internet, or the Internet using any of wired or wireless links (including terrestrial or satellite links). The one or more networks may be provided by Figure 1 The data exchange between the various elements can be achieved by any medium or mechanism. Figure 1 The various elements of the system may also have direct (wired or wireless) communication links. The sensors 112, controller 114, external data server computer 108, and other elements of the system each include an interface compatible with one or more networks 109, and are programmed or configured to use standardized protocols for communication across networks, such as TCP / IP, Bluetooth, CAN protocols, and higher layer protocols (such as HTTP, TLS, etc.).
[0059] The agricultural intelligence computer system 130 is programmed or configured to receive field data 106 from the field manager computing device 104, external data 110 from the external data server computer 108, and sensor data from the remote sensor 112. The agricultural intelligence computer system 130 may be further configured to host, use or execute one or more computer programs, other software elements, digitally programmed logic (such as an FPGA or ASIC), or any combination thereof, to perform conversion and storage of data values, construction of digital models of one or more crops on one or more fields, generation of recommendations and notifications, and generation and transmission of scripts to the application controller 114 in a manner further described in other parts of this disclosure.
[0060] In one embodiment, the agricultural intelligence computer system 130 is programmed with or includes a communication layer 132, a presentation layer 134, a data management layer 140, a hardware / virtualization layer 150, and a model and field data repository 160, and code instructions 180. "Layers" in this context refer to any combination of electronic digital interface circuits, microcontrollers, firmware (such as drivers), and / or computer programs or other software elements.
[0061] The communication layer 132 may be programmed or configured to perform input / output interface functions, including sending requests for field data, external data, and sensor data, respectively, to the field manager computing device 104, the external data server computer 108, and the remote sensors 112. The communication layer 132 may be programmed or configured to send received data to the model and field data repository 160 for storage as the field data 106.
[0062] In one embodiment, the code instructions 180 include model repository instructions 136, model execution instructions 137, model history instructions 138, and model query instructions 139. Other code instructions may also be included. The model repository instructions 136 can be used to generate and maintain a model repository for creating, storing, cataloging, and browsing machine-based models. The model execution instructions 137 can be used to determine the execution requirements of machine-based models, manage the execution resources available in the model execution infrastructure platform, and manage the execution of models in the model execution infrastructure platform. The model history instructions 138 can be used to generate a historical log of the model executed in the model execution infrastructure platform, aggregate the log, and provide a log browsing function. The model query instructions 139 can be used to receive, process, and parse queries issued to the model learning algorithm platform.
[0063] The presentation layer 134 may be programmed or configured to generate a graphical user interface (GUI) to be displayed on the field manager computing device 104, the cab computer 115, or other computer coupled to the system 130 via the network 109. The GUI may include controls for inputting data to be sent to the agricultural intelligence computer system 130, generating requests for models and / or recommendations, and / or displaying recommendations, notifications, models, and other field data.
[0064] The data management layer 140 may be programmed or configured to manage read operations and write operations involving the repository 160 and other functional elements of the system, including queries and result sets transmitted between the functional elements of the system and the repository. Examples of the data management layer 140 include JDBC, SQL server interface code, and / or HADOOP interface code, among others. The repository 160 may include a database. As used herein, the term "database" may refer to a body of data, a relational database management system (RDBMS), or both. As used herein, a database may include any collection of data (including hierarchical databases, relational databases, flat file databases, object-relational databases, object-oriented databases, distributed databases), and any other structured collection of records or data stored in a computer system. Examples of RDBMS include, but are not limited to MYSQL, DB2, SQL SERVER, and POSTGRESQL databases. However, any database implementing the systems and methods described herein may be used.
[0065] When field data 106 is provided directly to the agricultural intelligent computer system without one or more agricultural machinery or agricultural machinery equipment interacting with the agricultural intelligent computer system, the user can be prompted to enter such information via one or more user interfaces on the user device (served by the agricultural intelligent computer system). In an example embodiment, the user can specify identification data by accessing a map on the user device (served by the agricultural intelligent computer system) and selecting a specific CLU that has been graphically shown on the map. In an alternative embodiment, the user 102 can specify identification data by accessing a map on the user device (served by the agricultural intelligent computer system 130) and drawing the boundaries of the field on the map. Such CLU selection or map drawing represents a geographic identifier. In an alternative embodiment, the user can specify identification data by accessing field identification data from the Farm Service Agency of the United States Department of Agriculture or other sources (provided as a shape file or in a similar format) via a user device and providing such field identification data to the agricultural intelligent computer system.
[0066] In an example embodiment, the agricultural intelligence computer system 130 is programmed to generate a graphical user interface including a data manager for data input and cause the graphical user interface to be displayed. After one or more fields have been identified using the above method, the data manager may provide one or more graphical user interface widgets that, when selected, may identify changes in fields, soils, crops, tillage or nutrient practices. The data manager may include a timeline view, a spreadsheet view, and / or one or more editable programs.
[0067] Figure 5 An example embodiment of a timeline view of data input is depicted. Figure 5In the display described in , the user computer can input a selection of a specific field and a specific date to add an event. The events described at the top of the timeline may include nitrogen, planting, practices, and soil. To add a nitrogen application event, the user computer may provide input to select a nitrogen label. The user computer may then select a location on the timeline for the specific field to indicate the application of nitrogen on the selected field. In response to receiving a selection of a location on the timeline for a specific field, the data manager may display a data entry overlay, allowing the user computer to enter data related to nitrogen application, planting procedures, soil management, tillage procedures, irrigation practices, or other information related to the specific field. For example, if the user computer selects a portion of the timeline and indicates the application of nitrogen, the data entry overlay may include fields for entering the amount of nitrogen applied, the date of application, the type of fertilizer used, and any other information related to the application of nitrogen.
[0068] In one embodiment, the data manager provides an interface for creating one or more programs. A "program" in this context refers to a set of data relating to nitrogen applications, planting procedures, soil applications, tillage procedures, irrigation practices, or other information that may be associated with one or more fields and that can be stored in a digital data store for reuse as a set in other operations. After a program has been created, it can be conceptually applied to one or more fields, and a reference to the program can be stored in the digital storage in association with the data identifying those fields. Thus, instead of manually entering the same data relating to the same nitrogen application for multiple different fields, a user computer can create a program indicating a particular application of nitrogen, and then apply the program to multiple different fields. For example, in Figure 5 In the timeline view of FIG. 1 , the first two timelines have selected the "Spring Application" program, which includes applying 150 lbs N / ac in early April. The data manager may provide an interface for editing programs. In one embodiment, when a particular program is edited, each field that has selected that particular program is edited. For example, in Figure 5 If the "Spring Application" program is edited to reduce the nitrogen application to 130 lbs N / ac, the first two fields can be updated with the reduced nitrogen application based on the edited program.
[0069] In one embodiment, in response to receiving an edit to a field for which a program was selected, the data manager removes the association of the field with the selected program. For example, if nitrogen application is added to Figure 5 If the top field in the , the interface can be updated to indicate that the "Spring Application" program is no longer applied to the top field. Although the early April nitrogen application may remain, the update to the "Spring Application" program will not change the April application of nitrogen.
[0070] Figure 6 Depicted is an example embodiment of a spreadsheet view for data entry. Figure 6 The display depicted in FIG. 10 allows a user to create and edit information for one or more fields. The data manager may include information such as Figure 6 The spreadsheet depicted in FIG. 1 is used to enter information about nitrogen, planting, practices, and soils. To edit a specific entry, the user computer can select the specific entry in the spreadsheet and update the value. For example, Figure 6 An ongoing update of the target yield value for the second field is depicted. In addition, the user computer can select one or more fields for application of one or more programs. In response to receiving a program selection for a particular field, the data manager can automatically complete the entry for the particular field based on the selected program. Figure 1 In the same way, the data manager can update the entry of each field associated with a particular program in response to receiving an update to the program. In addition, the data manager can delete the corresponding relationship between the selected program and the field in response to receiving an edit to one of the entries of the field.
[0071] In one embodiment, the model and field data are stored in the model and field data repository 160. The model data includes a data model created for one or more fields. For example, a crop model may include a digitally constructed model of the development of crops on one or more fields. "Model" in this context refers to an electronic digital storage collection of executable instructions and data values associated with each other, which are capable of receiving a programmatic or other digital call, call or parsing request and responding to the programmatic or other digital call, call or parsing request based on a specified input value to produce one or more stored or calculated output values, which can be used as a basis for computer-implemented advice, output data display or machine control, etc. Those skilled in the art find it convenient to express the model using mathematical equations, but this form of expression does not limit the models disclosed herein to abstract concepts; on the contrary, each model herein has practical application in a computer in the form of stored executable instructions and data, which use a computer to implement the model. The model may include a model of past events on one or more fields, a model of the current state of one or more fields, and / or a model of predicted events on one or more fields. The model and field data may be stored in a data structure in memory, in rows in a database table, in a flat file or spreadsheet, or in other forms of stored digital data.
[0072] The hardware / virtualization layer 150 includes one or more central processing units (CPUs), memory controllers, and other devices, components, or elements of a computer system, such as volatile or non-volatile memory, non-volatile storage (such as disks), and, for example, a combination of Figure 4Layer 150 may also include programming instructions configured to support virtualization, containerization, or other technologies.
[0073] For the purpose of illustrating the example clearly, Figure 1 A limited number of instances of certain functional elements are shown. However, in other embodiments, there may be any number of such elements. For example, an embodiment may use thousands or millions of different mobile computing devices 104 associated with different users. In addition, the system 130 and / or the external data server computer 108 may be implemented using two or more processors, cores, clusters, or instances of physical or virtual machines that are configured in discrete locations or co-located with other elements in a data center, shared computing facility, or cloud computing facility.
[0074] 2.2. Application Overview
[0075] In one embodiment, the functions described herein are implemented using one or more computer programs or other software elements that are loaded into and executed using one or more general-purpose computers, which will cause these general-purpose computers to be configured as specific machines or computers configured specifically for performing the functions described herein. In addition, each flowchart further described herein can be used alone or in combination with the description of the procedures and functions in the prose form herein as an algorithm, plan or guide that can be used to program a computer or logic to implement the described functions. In other words, all prose and all attached text herein are intended to be used in conjunction with the above. Figure 1 The present invention is intended to provide a disclosure of algorithms, plans or instructions sufficient to allow a skilled person, taking into account the skill level applicable to the type of invention and disclosure combined with the skills and knowledge of such persons, to program a computer to perform the functions described herein.
[0076] In one embodiment, a user 102 interacts with an agricultural intelligence computer system 130 using a field manager computing device 104 configured with an operating system and one or more applications or applications; the field manager computing device 104 may also independently and automatically interoperate with the agricultural intelligence computer system under program control or logic control, and direct user interaction is not always required. The field manager computing device 104 broadly represents one or more of the following: a smart phone, a PDA, a tablet computing device, a laptop computer, a desktop computer, a workstation, or any other computing device capable of sending and receiving information and performing the functions described herein. The field manager computing device 104 can communicate via a network using a mobile application stored on the field manager computing device 104, and in some embodiments, the device can be coupled to a sensor 112 and / or a controller 114 using a cable 113 or connector. A particular user 102 may own, operate, or own and use more than one field manager computing device 104 in conjunction with the system 130 at a time.
[0077] The mobile application may provide client-side functionality to one or more mobile computing devices via a network. In an example embodiment, the field manager computing device 104 may access the mobile application via a web browser or a local client application or application. The field manager computing device 104 may send data to and receive data from one or more front-end servers using a network-based protocol or format (such as HTTP, XML and / or JSON) or an application-specific protocol. In an example embodiment, the data may take the form of a request to the mobile computing device and user information input (such as field data). In some embodiments, the mobile application interacts with location tracking hardware and software on the field manager computing device 104, which uses standard tracking technology (such as multi-lateration of radio signals, global positioning system (GPS), WiFi positioning system, or other mobile positioning methods) to determine the location of the field manager computing device 104. In some cases, location data or other data associated with the device 104, the user 102, and / or one or more user accounts may be obtained by querying the operating system of the device or by requesting an application on the device to obtain data from the operating system.
[0078] In one embodiment, the field manager computing device 104 sends field data 106 to the agricultural intelligence computer system 130, the field data 106 including or including, but not limited to, data values representing one or more of the following: geographic location of one or more fields, farming information for one or more fields, crops grown in one or more fields, and soil data extracted from one or more fields. The field manager computing device 104 may send the field data 106 in response to user input from the user 102 specifying data values for one or more fields. In addition, the field manager computing device 104 may automatically send the field data 106 when one or more of the data values become available to the field manager computing device 104. For example, the field manager computing device 104 may be communicatively coupled to the remote sensor 112 and / or the application controller 114, including an irrigation sensor and / or an irrigation controller. In response to receiving data indicating that the application controller 114 released water onto one or more fields, the field manager computing device 104 may send field data 106 to the agricultural intelligence computer system 130 indicating that water was released onto the one or more fields. The field data 106 identified in the present disclosure may be input and transmitted using electronic digital data transmitted between computing devices using parameterized URLs over HTTP or another suitable communication or messaging protocol.
[0079] A commercial example of a mobile application is CLIMATE FIELDVIEW, available from The Climate Corporation of San Francisco, California. The CLIMATE FIELDVIEW application or other applications may be modified, expanded, or adapted to include features, functionality, and programming that have not been disclosed prior to the filing date of this disclosure. In one embodiment, the mobile application includes an integrated software platform that allows growers to make fact-based decisions for their operations because it combines historical data related to the grower's fields with any other data that the grower wishes to compare. The combination and comparison can be done in real time and is based on scientific models that provide possible scenarios to allow growers to make better, more informed decisions.
[0080] Figure 2 Two views of an example logical organization of a set of instructions in main memory when an example mobile application is loaded for execution are shown. Figure 2, each named element represents a region of one or more pages of RAM or other main memory, or a region of one or more blocks of disk storage or other non-volatile storage, and programming instructions within those regions. In one embodiment, in view (a), the mobile computer application 200 includes account-field-data ingest-share instructions 202, overview and alert instructions 204, digital map manual instructions 206, seed and planting instructions 208, nitrogen instructions 210, weather instructions 212, field health instructions 214, and execution instructions 216.
[0081] In one embodiment, the mobile computer application 200 includes account, field, data ingestion, sharing instructions 202, which are programmed to receive, convert and ingest field data from third-party systems via manual upload or API. Data types may include field boundaries, yield maps, planting maps, soil test results, application maps, and / or management zones, etc. Data formats may include shape files, third-party native data formats, and / or farm management information system (FMIS) exports, etc. Receiving data may occur via: manual upload, email with attachment, external API that pushes data to the mobile application, or instructions that call the API of an external system to pull data into the mobile application. In one embodiment, the mobile computer application 200 includes a data inbox. In response to receiving a selection of the data inbox, the mobile computer application 200 may display a graphical user interface for manually uploading data files and importing the uploaded files into the data manager.
[0082] In one embodiment, the digital map book instructions 206 include a field map data layer stored in the device memory and are programmed with data visualization tools and geospatial field annotations. This provides growers with convenient information at their fingertips for reference, record keeping, and visual insight into field performance. In one embodiment, the overview and alert instructions 204 are programmed to provide an operational range view of what is important to growers, as well as timely suggestions to take action or focus on specific issues. This allows growers to focus their time on what needs to be focused on to save time and maintain yields throughout the season. In one embodiment, the seed and planting instructions 208 are programmed to provide tools for seed selection, mix placement, and script creation (including variable rate (VR) script creation) based on scientific models and empirical data. This enables growers to maximize yield or return on investment through optimized seed purchases, placement, and populations.
[0083] In one embodiment, the script generation instructions 205 are programmed to provide an interface for generating scripts, including variable rate (VR) fertility scripts. The interface enables growers to create scripts for field tools, such as nutrient application, planting, and irrigation. For example, the planting script interface may include a tool for identifying the type of seed to be planted. After receiving a selection of the seed type, the mobile computer application 200 may display one or more fields divided into management areas, such as a field map data layer created as part of the digital map book instructions 206. In one embodiment, the management area includes soil areas and a panel that identifies each soil area and the soil name, texture, drainage, or other field data for each area. The mobile computer application 200 may also display tools for editing or creating such on a map of one or more fields, such as a graphical tool for drawing management areas, such as soil areas. The planting process may be applied to all management areas, or different planting processes may be applied to different subsets of management areas. When a script is created, the mobile computer application 200 may make the script available for download in a format readable by the application controller, such as an archived or compressed format. Additionally, and / or alternatively, the script may be sent directly from the mobile computer application 200 to the cab computer 115 and / or uploaded to one or more data servers and stored for further use.
[0084] In one embodiment, the nitrogen instructions 210 are programmed to provide tools to inform nitrogen decisions by visualizing the availability of nitrogen to crops. This enables growers to maximize yields or return on investment through optimized nitrogen applications during the season. Example programming functions include: displaying images (such as SSURGO images) to enable fertilizer application areas to be mapped at high spatial resolution (depending on the proximity and resolution of the sensor, down to millimeters or less) and / or images generated based on sub-field soil data (such as data obtained from sensors); uploading existing grower-defined areas; providing maps of plant nutrient availability and / or maps that enable adjustment of one or more nitrogen applications across multiple areas; outputting scripts to drive machinery; tools for bulk data entry and adjustment; and / or maps for data visualization, etc. "Bulk data entry" in this context may mean entering data once and then applying the same data to multiple fields and / or areas that have been defined in the system; example data may include nitrogen application data that is the same for many fields and / or areas of the same grower, but such bulk data entry is applicable to entering any type of field data into the mobile computer application 200. For example, the nitrogen instructions 210 may be programmed to accept definitions of nitrogen application and practice programs and to accept user input specifying the application of those programs across multiple fields. A "nitrogen application program" in this context refers to a stored named data set associated with: a name, color code or other identifier, one or more application dates, the type of material or product and amount for each date, the method of application or incorporation (such as injection or broadcasting), and / or the application amount or application rate for each date, the crop or hybrid to which the application is applied, etc. A "nitrogen practice program" in this context refers to a stored named data set associated with: the name of the practice; the previous crop; the tillage system; the date of primary tillage; one or more previous tillage systems that were used; one or more indicators of the type of application (such as fertilization) that was used. The nitrogen instructions 210 may also be programmed to generate and cause the display of a nitrogen map that indicates the forecast for plant use of a specified nitrogen and whether a surplus or deficit is forecast; in some embodiments, different color indicators may represent the magnitude of the surplus or the magnitude of the deficit. In one embodiment, a nitrogen map includes: a graphical display in a computer display device including multiple rows, each row being associated with and identifying a field; data specifying what crops are planted in the field, the field size, the field location, and a graphical representation of the field perimeter; in each row, a monthly timeline with a graphical indicator specifying each nitrogen application and amount at a point associated with the month name; and a numerical and / or colored surplus indicator or deficit indicator, wherein the color indicates the magnitude.
[0085] In one embodiment, the nitrogen map may include one or more user input features such as a dial or a slider for dynamically changing the nitrogen planting and practice program so that the user can optimize their nitrogen map. The user can then use their optimized nitrogen map and the associated nitrogen planting and practice program to implement one or more scripts, including a variable rate (VR) fertility script. The nitrogen instruction 210 can also be programmed to generate and cause a nitrogen map to be displayed, which indicates the forecast of plant use of a specified nitrogen and whether a surplus or a shortage is predicted; in some embodiments, different color indicators can represent the magnitude of the surplus or the magnitude of the shortage. The nitrogen map can use numbers and / or colored surplus indicators or shortage indicators to display the forecast of plant use of a specified nitrogen and whether a surplus or a shortage is predicted for different times in the past and future (such as daily, weekly, monthly, or annually), where the color indicates the magnitude. In one embodiment, the nitrogen map may include one or more user input features such as a dial or a slider for dynamically changing the nitrogen planting and practice program so that the user can optimize their nitrogen map, for example, to obtain a preferred amount of surplus to shortage. The user can then use its optimized nitrogen map and associated nitrogen planting and practice programs to implement one or more scripts, including a variable rate (VR) fertility script. In other embodiments, instructions similar to nitrogen instructions 210 can be used for the application of other nutrients (such as phosphorus and potassium), the application of pesticides, and irrigation programs.
[0086] In one embodiment, weather instructions 212 are programmed to provide field-specific recent weather data and forecast weather information. This enables growers to save time and have an efficient integrated display regarding daily operational decisions.
[0087] In one embodiment, field health instructions 214 are programmed to provide timely remote sensing images that highlight in-season crop changes and potential concerns. Example programmed functions include: cloud checks to identify possible clouds or cloud shadows; determination of nitrogen index based on field images; graphical visualization of scouting layers, including those related to field health, and viewing and / or sharing of scouting notes; and / or downloading satellite imagery from multiple sources, with priority given to growers, etc.
[0088] In one embodiment, the performance instructions 216 are programmed to provide reporting, analysis, and insight tools that use farm data for evaluation, insight, and decision making. This enables growers to seek improved results for the next year through fact-based conclusions about why return on investment is at previous levels and insights into yield limiting factors. The performance instructions 216 can be programmed to be transmitted to a back-end analysis program executed at the agricultural intelligence computer system 130 and / or the external data server computer 108 via one or more networks 109, and are configured to analyze metrics such as yield, yield differences, hybrids, populations, SSURGO zones, soil test attributes, or altitude. Programmed reports and analyses may include: yield variability analysis, treatment effect estimates, benchmarking of yield and other metrics against other growers based on anonymous data collected from many growers or seed and planted data, and the like.
[0089] An application with instructions configured in this manner can be implemented for different computing device platforms while maintaining the same general user interface appearance. For example, a mobile application can be programmed to execute on a tablet, smartphone, or server computer accessed using a browser at a client computer. In addition, a mobile application configured for a tablet computer or smartphone can provide a complete application experience or a cockpit application experience that is suitable for the display and processing capabilities of the cockpit computer 115. For example, referring now to Figure 2In view (b), in one embodiment, the cab computer application 220 may include map-cab instructions 222, remote view instructions 224, data collection and transmission instructions 226, machine alarm instructions 228, script transmission instructions 230, and reconnaissance-cab instructions 232. The code base of the instructions of view (b) can be the same as that of view (a), and the executable files implementing the code can be programmed to detect the type of platform on which they are executed and to expose only those functions applicable to the cab platform or the full platform through the graphical user interface. This method enables the system to recognize the distinct user experiences applicable to the different technical environments of the in-cab environment and the cab. The map-cab instructions 222 can be programmed to provide map views of the field, farm, or area that are useful in guiding machine operations. The remote view instructions 224 can be programmed to open, manage, and provide real-time or near real-time views of machine activity to other computing devices connected to the system 130 via a wireless network, a wired connector or adapter, etc. Data collection and transmission instructions 226 may be programmed to open, manage, and provide for transmission of data collected at sensors and controllers to system 130 via a wireless network, wired connector or adapter, etc. Machine alert instructions 228 may be programmed to detect operational issues with a machine or implement associated with a cab and generate operator alerts. Script transmission instructions 230 may be configured to transmit in the form of instruction scripts configured to direct the collection of machine operations or data. Scouting-cab instructions 232 may be programmed to display location-based alerts and information received from system 130 based on the location of field manager computing device 104, agricultural implement 111, or sensor 112 in the field, and to ingest, manage, and provide for transmission of location-based scouting observations to system 130 based on the location of agricultural implement 111 or sensor 112 in the field.
[0090] 2.3. Data Ingestion into Computer Systems
[0091] In one embodiment, the external data server computer 108 stores external data 110, which includes soil data representing the soil composition of one or more fields and weather data representing the temperature and precipitation on one or more fields. The weather data may include past and current weather data and predictions for future weather data. In one embodiment, the external data server computer 108 includes multiple servers hosted by different entities. For example, a first server may contain soil composition data, while a second server may include weather data. In addition, the soil composition data may be stored in multiple servers. For example, one server may store data representing the percentage of sand, silt, and clay in the soil, while a second server may store data representing the percentage of organic matter (OM) in the soil.
[0092] In one embodiment, the remote sensor 112 includes one or more sensors that are programmed or configured to generate one or more observations. The remote sensor 112 can be an aerial sensor (such as a satellite), a vehicle sensor, a planting equipment sensor, a tillage sensor, a fertilizer or pesticide application sensor, a harvester sensor, and any other appliance capable of receiving data from one or more fields. In one embodiment, the application controller 114 is programmed or configured to receive instructions from the agricultural intelligent computer system 130. The application controller 114 can also be programmed or configured to control the operating parameters of an agricultural vehicle or implement. For example, the application controller can be programmed or configured to control the operating parameters of a vehicle (such as a tractor), a planting equipment, a tillage equipment, a fertilizer or pesticide equipment, a harvester equipment, or other farm implements (such as a water valve). Other embodiments can use any combination of sensors and controllers, of which the following are only selected examples.
[0093] The system 130 may obtain or ingest data on a large scale from a large number of growers who have contributed data to a shared database system under the control of the user 102. When one or more user-controlled computer operations are requested or triggered to obtain data for use by the system 130, this form of acquiring data may be referred to as "manual data ingestion." For example, the CLIMATE FIELDVIEW application available from The Climate Corporation of San Francisco, California, may be operated to export data to the system 130 for storage in the repository 160.
[0094] For example, the seed monitor system can both control planter assembly components and obtain planting data, including signals from seed sensors via a signal harness that includes a CAN backbone and point-to-point connections for registration and / or diagnostics. The seed monitor system can be programmed or configured to display seed spacing, population, and other information to a user via the cab computer 115 or other device within the system 130. Examples are disclosed in U.S. Patent No. 8,738,243 and U.S. Patent Publication 20150094916, and the present disclosure assumes knowledge of those other patent disclosures.
[0095] Likewise, the yield monitor system may include yield sensors for harvester equipment that send yield measurement data to the cab computer 115 or other equipment within the system 130. The yield monitor system may utilize one or more remote sensors 112 to obtain grain moisture measurements in a combine or other harvester and send these measurements to a user via the cab computer 115 or other equipment within the system 130.
[0096] In one embodiment, examples of sensors 112 that can be used with any mobile vehicle or device of the type described elsewhere herein include kinematic sensors and position sensors. Kinematic sensors may include any speed sensor, such as a radar or wheel speed sensor, an accelerometer, or a gyroscope. Position sensors may include a GPS receiver or transceiver, or a WiFi-based location or mapping application, etc., which are programmed to determine location based on nearby WiFi hotspots.
[0097] In one embodiment, examples of sensors 112 that can be used with a tractor or other moving vehicle include: an engine speed sensor, a fuel consumption sensor, an area counter or distance counter that interacts with a GPS or radar signal, a PTO (power take-off) speed sensor, a tractor hydraulic sensor configured to detect a hydraulic parameter (such as pressure or flow) and / or a hydraulic pump speed, a wheel speed sensor or a wheel slip sensor. In one embodiment, examples of controllers 114 that can be used with a tractor include: a hydraulic directional controller, a pressure controller, and / or a flow controller; a hydraulic pump speed controller; a speed controller or governor; a suspension position controller; or a wheel position controller to provide automatic steering.
[0098] In one embodiment, examples of sensors 112 that may be used with seed planting equipment such as a planter, seed drill, or air seeder include: a seed sensor, which may be an optical, electromagnetic, or collision sensor; a downforce sensor, such as a weigh pin, a load cell, a pressure sensor; a soil property sensor, such as a reflectivity sensor, a moisture sensor, a conductivity sensor, an optical residue sensor, or a temperature sensor; a component operation standard sensor, such as a planting depth sensor, a downforce cylinder pressure sensor, a seed tray speed sensor, a seed drive motor encoder, a seed conveyor system speed sensor, or a vacuum sensor; or a pesticide application sensor, such as an optical or other electromagnetic sensor or a collision sensor. In one embodiment, examples of controllers 114 that may be used with such seed planting equipment include: a toolbar fold controller, such as a controller for a valve associated with a hydraulic cylinder; a down force controller, such as a controller for a valve associated with a pneumatic cylinder, an airbag, or a hydraulic cylinder, and programmed to apply down force to a single row unit or an entire planter frame; a planting depth controller, such as a linear actuator; a metering controller, such as an electric seeder drive motor, a hydraulic seeder drive motor, or a swath control clutch; a mix selection controller, such as a seeder drive motor, or other actuator programmed to selectively allow or prevent seed or an air-seed mixture from delivering seed to or from a seeder or a central bulk hopper; a metering controller, such as an electric seeder drive motor or a hydraulic seeder drive motor; a seed conveyor system controller, such as a controller for a belt seed delivery conveyor motor; a marking controller, such as a controller for a pneumatic or hydraulic actuator; or a pesticide application rate controller, such as a metering drive controller, an orifice size, or a position controller.
[0099] In one embodiment, examples of sensors 112 that may be used with the tillage equipment include: a position sensor for a tool such as a handle or a disk; a tool position sensor for such a tool configured to detect depth, coaxial angle, or lateral spacing; a downforce sensor; or a draft force sensor. In one embodiment, examples of controllers 114 that may be used with the tillage equipment include a downforce controller or a tool position controller, such as a controller configured to control tool depth, coaxial angle, or lateral spacing.
[0100] In one embodiment, examples of sensors 112 that may be used in connection with an apparatus for applying fertilizer, pesticides, fungicides, etc., such as a priming fertilizer system on a planter, a subsoil fertilizer applicator, or a fertilizer sprayer, include: fluid system standard sensors, such as flow sensors or pressure sensors; sensors indicating which sprinkler valves or fluid line valves are open; sensors associated with a tank, such as a level sensor; supply line sensors for sections or the entire system, or row-specific supply line sensors; or kinematic sensors, such as accelerometers disposed on a sprayer boom. In one embodiment, examples of controllers 114 that may be used with such an apparatus include: pump speed controllers; valve controllers that are programmed to control pressure, flow, direction, PWM, etc.; or position actuators such as for boom height, deep tillage blade depth, or boom position.
[0101] In one embodiment, examples of sensors 112 that may be used with a harvester include: a yield monitor, such as a strike plate strain gauge or position sensor, a capacitive flow sensor, a load sensor, a weight sensor, or a torque sensor associated with an elevator or auger, or an optical or other electromagnetic grain height sensor; a grain moisture sensor, such as a capacitive sensor; a grain loss sensor, including a collision sensor, an optical sensor, or a capacitive sensor; a header operation standard sensor, such as a header height sensor, a header type sensor, a deck clearance sensor, a feeder speed sensor, and a reel speed sensor; a separator operation standard sensor, such as a concave clearance, a rotor speed, a brake shoe clearance, or a screen clearance sensor; an auger sensor for position, operation, or speed; or an engine speed sensor. In one embodiment, examples of controllers 114 that may be used with a harvester include: a header operation standard controller for elements such as header height, header type, deck clearance, feeder speed, or reel speed; a separator operation standard controller for features such as concave clearance, a rotor speed, a brake shoe clearance, or a screen clearance; or a controller for auger position, operation, or speed.
[0102] In one embodiment, examples of sensors 112 that can be used with a grain trailer include weight sensors or sensors for auger position, operation or speed. In one embodiment, examples of controllers 114 that can be used with a grain trailer include controllers for auger position, operation or speed.
[0103] In one embodiment, examples of sensors 112 and controllers 114 may be installed in an unmanned aerial vehicle (UAV) device or "drone." Such sensors may include cameras with detectors effective for any range of the electromagnetic spectrum, including visible light, infrared, ultraviolet, near infrared (NIR), etc.; accelerometers; altimeters; temperature sensors; humidity sensors; pitot tube sensors or other airspeed or wind speed sensors; battery life sensors; or radar transmitters and reflected radar energy detection devices; other electromagnetic radiation transmitters and reflected electromagnetic radiation detection devices. Such controllers may include guidance or motor control equipment, control surface controllers, camera controllers, or controllers programmed to turn on any of the foregoing sensors, operate any of the foregoing sensors, obtain data from any of the foregoing sensors, manage and configure any of the foregoing sensors. Examples are disclosed in U.S. Patent Application No. 14 / 831,165, and the present disclosure assumes knowledge of that other patent disclosure.
[0104] In one embodiment, the sensor 112 and controller 114 may be secured to a soil sampling and measuring device that is configured or programmed to sample soil and perform soil chemical tests, soil moisture tests, and other soil-related tests. For example, the devices disclosed in U.S. Pat. No. 8,767,194 and U.S. Pat. No. 8,712,148 may be used, and the present disclosure assumes knowledge of those patent disclosures.
[0105] In one embodiment, the sensor 112 and the controller 114 may include a weather device for monitoring weather conditions of the field. For example, the devices disclosed in U.S. Provisional Application No. 62 / 154,207 filed on April 29, 2015, U.S. Provisional Application No. 62 / 175,160 filed on June 12, 2015, U.S. Provisional Application No. 62 / 198,060 filed on July 28, 2015, and U.S. Provisional Application No. 62 / 220,852 filed on September 18, 2015 may be used, and the present disclosure assumes knowledge of those patent disclosures.
[0106] 2.4. Treatment Overview - Agronomic Model Training
[0107] In one embodiment, the agricultural intelligent computer system 130 is programmed or configured to create an agronomic model. In this context, the agronomic model is a data structure in the memory of the agricultural intelligent computer system 130, which includes field data 106, such as identification data and harvest data of one or more fields. The agronomic model may also include calculated agronomic characteristics, which describe the conditions that may affect the growth of one or more crops on the field or the characteristics of one or more crops or both. In addition, the agronomic model may include suggestions based on agronomic factors, such as crop suggestions, irrigation suggestions, planting suggestions, fertilizer suggestions, fungicide suggestions, pesticide suggestions, harvesting suggestions and other crop management suggestions. Agronomic factors can also be used to estimate one or more crop-related results, such as agronomic yields. The agronomic yield of a crop is an estimate of the quantity of the crop produced, or in some examples, the income or profit obtained from the crop produced.
[0108] In one embodiment, the agricultural intelligence computer system 130 may use a pre-configured agronomic model to calculate agronomic characteristics related to the location and crop information of one or more fields currently received. The pre-configured agronomic model is based on previously processed field data, which includes but is not limited to identification data, harvest data, fertilizer data, and weather data. The pre-configured agronomic model may have been cross-validated to ensure the accuracy of the model. Cross-validation may include a comparison with ground truth, which compares the predicted results with the actual results on the field, such as comparing rainfall estimates with rain gauges or sensors that provide weather data for the same or nearby locations, or comparing estimates of nitrogen content with soil sample measurements.
[0109] Figure 3 A programmed process is shown by which an agricultural intelligence computer system generates one or more pre-configured agronomic models using field data provided by one or more data sources. Figure 3 Algorithms or instructions may be used to program the functional elements of the agricultural intelligence computer system 130 to perform the operations now described.
[0110] At box 305, the agricultural intelligence computer system 130 is configured or programmed to implement agronomic data preprocessing of field data received from one or more data sources. The field data received from one or more data sources may be preprocessed for the purpose of removing noise, distortion effects, and confounding factors within the agronomic data, including measurement outliers that may adversely affect the received field data values. Embodiments of agronomic data preprocessing may include, but are not limited to, removing data values that are typically associated with outlier data values, specific measurement data points that are known to unnecessarily skew other data values, data smoothing, aggregation, or sampling techniques for removing or reducing additive or multiplicative effects from noise, and other filtering or data derivation techniques for providing clear distinctions between positive and negative data inputs.
[0111] At box 310, the agricultural intelligent computer system 130 is configured or programmed to perform data subset selection using the preprocessed field data to identify data sets useful for initial agronomic model generation. The agricultural intelligent computer system 130 can implement data subset selection techniques, including but not limited to genetic algorithm methods, all subset model methods, sequential search methods, stepwise regression methods, particle swarm optimization methods, and ant colony optimization methods. For example, the genetic algorithm selection technique uses an adaptive heuristic search algorithm based on the evolutionary principles of natural selection and genetics to determine and evaluate data sets within the preprocessed agronomic data.
[0112] At box 315, the agricultural intelligent computer system 130 is configured or programmed to implement field data set evaluation. In one embodiment, a specific field data set is evaluated by creating an agronomic model and using a specific quality threshold for the created agronomic model. One or more comparison techniques (such as, but not limited to, leave-one-out cross-validation root mean square error (RMSECV), mean absolute error, and mean percentage error) can be used to compare and / or validate the agronomic model. For example, RMSECV can cross-validate the agronomic model by comparing the predicted agronomic attribute values created by the agronomic model with the historical agronomic attribute values collected and analyzed. In one embodiment, the agronomic data set evaluation logic is used as a feedback loop, where an agronomic data set that does not meet the configured quality threshold is used during a future data subset selection step (box 310).
[0113] At block 320, the agricultural intelligence computer system 130 is configured or programmed to implement agronomic model creation based on the cross-validated agronomic data set. In one embodiment, the agronomic model creation may implement a multivariate regression technique to create a pre-configured agronomic data model.
[0114] At box 325 , the agricultural intelligence computer system 130 is configured or programmed to store the pre-configured agronomic data model for use in future field data evaluations.
[0115] 2.5. Implementation Example - Hardware Overview
[0116] According to one embodiment, the techniques described herein are implemented by one or more special-purpose computing devices. Special-purpose computing devices may be hardwired to perform these techniques, or may include digital electronic devices such as one or more application-specific integrated circuits (ASICs) or field programmable gate arrays (FPGAs) that are permanently programmed to perform these techniques, or may include one or more general-purpose hardware processors that are programmed to perform these techniques according to program instructions in firmware, memory, other storage, or a combination. Such special-purpose computing devices may also combine customized hard-wired logic, ASICs, or FPGAs with customized programming to implement these techniques. Special-purpose computing devices may be desktop computer systems, portable computer systems, handheld devices, networking devices, or any other device containing hard-wired and / or program logic to implement these techniques.
[0117] For example, Figure 4 4 is a block diagram illustrating a computer system 400 upon which embodiments of the present invention may be implemented. Computer system 400 includes a bus 402 or other communication mechanism for communicating information, and a hardware processor 404 coupled to bus 402 for processing information. Hardware processor 404 may be, for example, a general purpose microprocessor.
[0118] The computer system 400 also includes a main memory 406, such as a random access memory (RAM) or other dynamic storage device, coupled to the bus 402 for storing information and instructions to be executed by the processor 404. The main memory 406 may also be used for storing temporary variables or other intermediate information during execution of instructions to be executed by the processor 404. Such instructions, when stored in a non-transitory storage medium accessible to the processor 404, render the computer system 400 a special-purpose machine dedicated to performing the operations specified in the instructions.
[0119] Computer system 400 also includes a read only memory (ROM) 408 or other static storage device coupled to bus 402 for storing static information and instructions for processor 404. A storage device 410, such as a magnetic disk, optical disk, or solid state drive, is provided and coupled to bus 402 for storing information and instructions.
[0120] The computer system 400 may be coupled via the bus 402 to a display 412, such as a cathode ray tube (CRT), for displaying information to a computer user. An input device 414, including alphanumeric and other keys, is coupled to the bus 402 to communicate information and command selections to the processor 404. Another type of user input device is a cursor control 416, such as a mouse, trackball, or cursor direction keys, for communicating directional information and command selections to the processor 404 and for controlling movement of a cursor on the display 412. The input device typically has two degrees of freedom in two axes, a first axis (e.g., x) and a second axis (e.g., y), which allows the device to specify a position in a plane.
[0121] Computer system 400 may implement the techniques described herein using custom hardwired logic, one or more ASICs or FPGAs, firmware, and / or program logic that, in conjunction with the computer system, makes computer system 400 a special purpose machine or programs computer system 400 to be a special purpose machine. According to one embodiment, the techniques herein are performed by computer system 400 in response to processor 404 executing one or more sequences of one or more instructions contained in main memory 406. Such instructions may be read into main memory 406 from another storage medium such as storage device 410. Execution of the sequences of instructions contained in main memory 406 causes processor 404 to perform the process steps described herein. In alternative embodiments, hardwired circuitry may be used in place of or in combination with software instructions.
[0122] The term "storage medium" as used herein refers to any non-transitory medium that stores data and / or instructions that cause a machine to operate in a particular manner. Such storage media may include non-volatile media and / or volatile media. Non-volatile media include, for example, optical disks, magnetic disks, or solid-state drives, such as storage device 410. Volatile media include dynamic memory, such as main memory 406. Common forms of storage media include, for example: floppy disks, flexible disks, hard disks, solid-state drives, magnetic tapes or any other magnetic data storage media, CD-ROMs, any other optical data storage media, any physical media with a pattern of holes, RAM, PROM, and EPROM, FLASH-EPROM, NVRAM, any other memory chip or cassette tape.
[0123] Storage media are distinct from transmission media, but can be used in conjunction with transmission media. Transmission media participate in the transfer of information between storage media. For example, transmission media include coaxial cables, copper wire, and optical fiber, including the wires that make up bus 402. Transmission media can also take the form of sound or light waves, such as those generated during radio wave and infrared data communications.
[0124] Various forms of media may be involved in carrying one or more sequences of one or more instructions to processor 404 for execution. For example, the instructions may initially be carried on a disk or solid-state drive of a remote computer. The remote computer may load the instructions into its dynamic memory and send the instructions over a telephone line using a modem. A modem local to computer system 400 may receive the data on the telephone line and convert the data into an infrared signal using an infrared transmitter. An infrared detector may receive the data carried in the infrared signal, and appropriate circuitry may place the data on bus 402. Bus 402 carries the data to main memory 406, from which processor 404 retrieves and executes the instructions. The instructions received by main memory 406 may optionally be stored on storage device 410 before or after execution by processor 404.
[0125] Computer system 400 also includes a communication interface 418 coupled to bus 402. Communication interface 418 provides bidirectional data communication coupling to a network link 420 connected to a local network 422. For example, communication interface 418 can be an integrated services digital network (ISDN) card, a cable modem, a satellite modem, or a modem for providing a data communication connection to a telephone line of a corresponding type. As another example, communication interface 418 can be a local area network (LAN) card for providing a data communication connection to a compatible LAN. A wireless link also can be implemented. In any such implementation, communication interface 418 sends and receives electrical signals, electromagnetic signals, or optical signals that carry digital data streams representing various types of information.
[0126] The network link 420 typically provides data communications through one or more networks to other data devices. For example, the network link 420 may provide a connection through a local network 422 to a host computer 424 or to data equipment operated by an Internet Service Provider (ISP) 426. The ISP 426, in turn, provides data communication services through the global packet data communication network now commonly referred to as the "Internet" 428. Both the local network 422 and the Internet 428 use electrical, electromagnetic, or optical signals that carry digital data streams. The signals through the various networks and the signals on the network link 420 and through the communication interface 418 are example forms of transmission media that carry the digital data to and from the computer system 400.
[0127] Computer system 400 can send messages and receive data, including program code, through one or more networks, network link 420 and communication interface 418. In the Internet example, server 430 can send code for the requested application through Internet 428, ISP 426, local network 422 and communication interface 418.
[0128] The received code may be executed by processor 404 as it is received and / or stored in storage device 410 or other non-volatile storage for later execution.
[0129] 3. Systems for executing machine learning algorithms
[0130] In one embodiment, a computer system for executing a machine learning algorithm includes hardware and software programmed to execute various machine-based models. In some embodiments, independent components can be selected, modified, and grouped to form a combination of components configured to support the execution of a model. Components can be combined to form a sequential, parallel, or hybrid combination that can facilitate the execution of a model. This can be illustrated using the following example: Assume that the model repository includes components that implement various processes for planting corn and various processes for estimating corn planting cycles. In addition, it is assumed that the model execution infrastructure includes a group of CPUs, a group of memory units, storage devices, communication buses, and communication networks. Using components from the model repository, crop growers can select or build a model that is configured to determine how much nitrogen is recommended to farmland to obtain the maximum possible corn harvest from the field, or when and how to fertilize the field, or which type of corn should be planted, or how many days it takes for the corn in the field to mature. Crop growers can provide input to the model that specifies information about the soil, climate, and location of the field. Then, growers can request to execute the model on the model execution infrastructure and according to the input provided to generate output. The output can include nitrogen and corn type recommendations for the field. Outputs could also include forecasts for corn harvest dates, etc.
[0131] Fig. 7A An exemplary computer-based platform 700 for executing machine learning algorithms is depicted. In the depicted example, the platform 700 includes a management processor 705, a model repository 710, a model execution infrastructure 720, and a model history log 730. The management processor 705 can be implemented in a computing device configured to receive, process, and transmit any type of data.
[0132] In one embodiment, the infrastructure 720 includes one or more storage devices 722, one or more cloud system computers 723, one or more cloud-based networks 724, one or more computer servers 725, one or more computer workstations 726, and one or more communication networks 727. The model execution infrastructure 720 is Fig. 8A In other embodiments, infrastructure 720 includes fewer or more components.
[0133] The data may be transmitted and received via one or more communication networks 740 and one or more computing devices configured to transmit and receive data (such as a terminal 750, a tablet computer 752, a mobile device 754, a server computer 756, and other computing devices ( Fig. 7A , such as cloud storage, cloud-based networks, etc.) to facilitate access to the platform 700. In other embodiments, the platform 700 may include fewer or more devices.
[0134] In one embodiment, the platform 700 is configured to catalog, define, and organize various machine-based models, store them in a repository 710, execute them on a model execution infrastructure 720, and log the execution data of the models in a model history log 730. The platform 700 is superior to other platforms known in the industry for building and storing models in providing comprehensive and extensive capabilities (described below) to track models, manage modifications and execution of models, and make the history and log of models available to users. For example, the platform 700 provides comprehensive version control capabilities for tracking versions, modifications, and runtime characteristics of models, as well as comprehensive cataloging capabilities for browsing and querying models and their definitions.
[0135] In one embodiment, the model repository 710 is configured to store and catalog the definitions and metadata of machine-based models. The definition of a model describes the characteristics specific to the model, while the metadata associated with the model describes the usage and execution characteristics of the model.
[0136] In one embodiment, model execution infrastructure 720 provides the ability to abstract model execution and model input so that infrastructure 720 can be reused for executing various models.
[0137] In one embodiment, the model history log 730 is configured to store a historical log of the execution of the model. The log may be useful in restoring an old or different version of a model, executing different versions of the same model, and comparing results generated by different versions of a model. The model history log 730 allows for the storage of metrics about the execution of various models on the same or different components of the model execution infrastructure 720 and on different input data. This may be useful in historical analysis of runtime metrics, expected accuracy metrics, and expected execution duration metrics of the execution model.
[0138] 3.1. Example Model Repository
[0139] In one embodiment, the model repository 710 is a computer-based database configured to store information about machine learning models. The model repository 710 can be used to store various types of models and various types of information for each model.
[0140] Figure 7BAn exemplary computer-based model repository 710 is depicted. In the depicted example, the repository 710 includes examples of machine-based models that may be useful in agricultural applications. The repository 710 may also include Figure 7B Other models not depicted in the diagram. The models stored in the repository 710 may be indexed, organized, and categorized. The index of the model may be stored in an index data structure 719, which may be organized using data pointers, data tables, and the like.
[0141] In the context of agricultural applications, examples of machine-based models may include operational research models 712 , machine learning models 714 , deep learning models 716 , and physical models 718 .
[0142] Despite Figure 7B Not shown, but machine-based models may include models specific to non-agricultural applications, such as human resource models, billing models, etc.
[0143] Operational research model 712 is a special model configured to handle complicated mathematical equations to solve optimization problems.Usually, operational research model 712 is handled based on constraint programming.The operational research model can accept the input comprising one or more constraints.After receiving the input, the model can perform an optimization algorithm to draw the solution that the input value of the constraint is optimized.The operational research model can be used to optimize the number of employees and the number of robot machines in the genetics laboratory that for example handles the required genetic related data in the shortest possible time period.
[0144] The operational research model 712 has broad applicability in agricultural technology to optimize seed planting schedules, optimize fertilization schedules, and optimize harvest schedules. The operational research model 712 can also be applied to logistics and transportation.
[0145] Machine learning models 714 typically include artificial intelligence ("AI") models. They are typically used to learn the behavior of a system and, based on the learned behavior of the system, predict the system's response to new inputs and data. Machine learning models 714 can be structured and customized for specific systems and specific tasks.
[0146] The machine learning model 714 can be widely used in agricultural technology. For example, a machine learning model can be trained on a set of plant pictures showing certain plant diseases. The training can include assigning names of diseases to the pictures. For example, the training can include building associations between disease characteristics depicted in the pictures and the names of the diseases. Once the model is trained, the model can be deployed and can be used to identify plant diseases depicted in plant photos provided by users.
[0147] The deep learning model 716 includes a machine learning model that is generally more advanced and complex than the machine learning model 714. The deep learning model 716 may use, for example, a random forest algorithm, a linear regression algorithm, a neural network algorithm, and other complex algorithms.
[0148] Physical model 718 is a physical model of a software component that can be executed in model execution infrastructure 720. For example, physical model 718 can include software libraries, instructions, and programmable calls that can be combined and selected to form an executable file for a specific model, such as an operational research model, a machine learning model, or a deep learning model.
[0149] The models stored in model repository 710 are typically stored in association with their respective definitions and metadata. Each model may be associated with its own set of definitions and its own metadata. Thus, each operational research model in operational research models 712 may have its own definition and metadata 712A, each machine learning model in machine learning models 714 may have its own definition and metadata 714A, each deep learning model in deep learning models 716 may have its own definition and metadata 716A, and each physical model in physical models 718 may have its own definition and metadata 718A.
[0150] Figure 7C An example data structure maintained by a computer-based model repository is depicted. The depicted example shows an example physical model data structure 718B and an example physical model 718A (in Figure 7B ) associated definitions and metadata structure 718BB. Figure 7C Example data structures are depicted only for physical models, but other models may have similar data structures stored in model repository 710 .
[0151] like Figure 7C As shown, model repository 710 may store a physical model data structure 718A for physical model 718. Data structure 718A may be used to store definitions and metadata associated with physical model 718.
[0152] Data structure 718A may be used to store definitions and metadata associated with physical model 718. Definitions are typically related to static characteristics of the model, while metadata are typically related to characteristics related to the execution of the model. In some cases, the distinction between definitions and metadata may be unclear because both definitions and metadata may relate to, for example, expected runtime metrics. For example, expected runtime metrics may be described in both definitions and metadata.
[0153] exist Figure 7CIn the example shown, the definition and metadata 718A includes a model name, input definition, output definition, infrastructure requirements, runtime metrics, expected execution duration, expected accuracy, status (such as development status, test status, or production status), model classification, model version, model type, and model owner. In other examples, the data structure 718A may include fewer or more elements.
[0154] In one embodiment, the definition and metadata 718A includes a model artifact. An artifact is a binary file that is built for a machine-based model and can be executed to model certain behaviors of a system. Once a model is trained on training data, a binary file is usually generated. Once the training phase is completed, the trained model is captured into an actual binary file. The binary file includes executable instructions that, when executed by a computer processor, cause the model to predict system behaviors consistent with the training data. The binary file (i.e., the artifact) can be deployed to production and executed in the model execution infrastructure 720.
[0155] The elements included in the machine-based model definition and metadata 718A can vary and can depend on the type of model and whether the model has been executed or modified. For example, the definition and metadata associated with a machine learning model will typically include the specific type of model, quality metrics of the model, and the expected accuracy of the model. The expected accuracy of the model can be expressed as a score (such as 95%) or similar form.
[0156] Figure 7C The data structure depicted in can be managed by Fig. 7A The processor 705 is shown to create, query, update and maintain.
[0157] Access to the data structures maintained in the model repository 710 may be via one or more communication networks 740 (in Fig. 7A shown in Figure 7C 750 ) and facilitated via one or more computing devices (such as terminal 750 , tablet computer 752 , mobile device 754 , server computer 756 , and other computing devices configured to transmit and receive data).
[0158] The model repository 710 can be updated and modified. For example, a new model can be added to the model repository 710, and / or a new version of a model can be created and stored in the repository.
[0159] To create a new machine-based model or a new version of a model, a user or administrator can build a new model from the building blocks available in the model repository 710. Once built, an initial definition is created for the new model and stored in the model repository 710 to indicate that the new model is under development. Once the new model is tested, another definition is created and stored in the model repository 710 to indicate that the new model is being tested. Once the new model is deployed to production, additional definitions are created for the new model and stored in the model repository 710 to indicate that the new model has been put into production. Each definition can be stored in association with the new model itself, using, for example, index 719 (in Figure 7B as shown) to be indexed and made available for querying and reporting.
[0160] Once a new model is ready to execute, additional definitions are created and stored in the model repository 710 to indicate, for example, expected execution duration, expected accuracy, and execution infrastructure requirements.
[0161] Once the new model is executed, metadata is created for the model and stored in the model repository 710 to indicate, for example, runtime metrics, state data, status indicators, etc. Additionally, an execution log of the new model is created and stored in the model history log database 730 .
[0162] When a new model is executed on new input data, new metadata is created for the model and stored in the model repository 710 to indicate new runtime metrics, state data, status indicators, etc. Additionally, a new execution log of the model is created and stored in the model history log database 730 .
[0163] This process is repeated each time a model is created, modified, executed, or queried. Over time, the model repository 710, model execution infrastructure 720, and model history log database 730 become a central repository that can be extended and tested by scientists, researchers, and agricultural professionals, and used and compared by crop growers.
[0164] Fig.7D An example flow chart for building, storing, and cataloging machine-based models is depicted. This example is provided only to illustrate one of many ways to incorporate a model into the repository 710, and one of many ways to integrate a model into the platform 700. The depicted process can be performed by including Fig. 7A The processing may be executed by any computer-based processor, including the management processor 705 shown.
[0165] In step 780, the computer-based processor receives a request to store a machine-based model. The request may be received from any terminal device such as Fig. 7A , 7C Devices 750-756) are shown being received.
[0166] In step 782, the processor parses the request and identifies model artifacts and information about the model. The artifacts may be provided in the request itself; alternatively, the request may include a hyperlink to a storage space where the artifact is stored, and the processor may retrieve the artifact from the storage space. Similarly, information about the model may be included in the request itself, or the request may include a hyperlink to a storage space, and the processor may retrieve the definition and metadata of the model from the storage space.
[0167] In step 784, the processor determines the inputs for the model. The inputs and outputs may be provided by the user. Alternatively, a hyperlink may be provided to a storage space from which the inputs / outputs may be retrieved. The inputs may include specific input values that will be processed to generate output values when provided to the model. The inputs may include, for example, information about the soil, climate, and location of the farmland. The outputs may include a request to provide a recommendation on the type of fertilizer that, if applied to the field, would achieve the maximum yield in the field.
[0168] In step 788, the processor determines the infrastructure requirements for executing the model. The requirements may have been provided in the definition associated with the model. Alternatively, the requirements may be obtained from the model repository 710 that has stored a record of the model or a data record of a similar model. The requirements may include CPU specifications, memory requirements, storage requirements, etc.
[0169] In step 790, the processor determines the state, version, expected accuracy, and estimated execution time of the model. This information may already be available in the definition associated with the model. Alternatively, this information may be obtained from the model repository 710 where a record of the model has been stored or records of similar models are stored.
[0170] In step 792 , the processor generates or updates, definitions and metadata for the model, and stores the updated definitions and metadata in the model repository 710 in association with the model.
[0171] In step 794, the processor associates the model artifacts with the model's definition and metadata. For example, the processor may create a physical model data structure 718A for the model. The data structure 718A is stored in Figure 7C is described in .
[0172] In step 795 , the processor stores the association between the model artifacts and the definition and metadata of the model in the model repository 710 .
[0173] In step 796, the processor creates or updates a record for the model in the model history log database 730. This may include checking whether a record has already been created for the model, and if so, updating the record with the information included in the definition and metadata. However, if a record has not yet been created for the model, the processor may create a record in the model history log database 730 and store the definition and metadata of the model in the record.
[0174] If the processor receives a new request in step 797, the processor proceeds to step 782 to process the new request. Otherwise, the processor ends execution in step 799.
[0175] 3.2. Example Model Execution Infrastructure
[0176] Fig. 8A An exemplary computer-based model execution infrastructure 720 for executing machine learning algorithms is depicted.
[0177] In one embodiment, infrastructure 720 is a collection of hardware and software components that can be used in combination or individually to facilitate the execution of machine-based models stored in model repository 710. Infrastructure 720 may include, for example, computer processors, memory units, caches, communication buses, and computer programming instructions that together provide an execution platform for executing models. The computer programming instructions may include instructions for an operating system, system calls, cross-platform application programming interfaces, programming libraries, and the like.
[0178] The infrastructure 720 can be accessed via any type of computing device. In the depicted example, the user 102 can access the infrastructure 720 from a mobile device 754. The mobile device 754 can be configured to establish a communication session and communicate (808) with various components of the infrastructure 720, and communicate (810) with the physical machine 1102, and communicate (816) with the model repository 710 and the model history log database 730. The infrastructure 720 can provide (812) recommendations to the mobile device 754. The infrastructure 720 can also be configured to provide (814) instructions to machines 1102 such as tractors, combines, planters, seed drills, and the like.
[0179] In one embodiment, the user 102 provides input to the infrastructure 720 to cause the infrastructure 720 to obtain agricultural recommendations for the field based on the input, to cause the infrastructure 720 to transmit the recommendations to the user 102, and / or to transmit the recommendations directly to the agricultural machine 1102 to cause the machine 1102 to cultivate the field according to the recommendations obtained from the infrastructure 720. For example, upon receiving a request for agricultural recommendations for the field from the user 102, the mobile device 754 transmits the request to the model execution infrastructure 720 (808), and the infrastructure 720 selects a particular machine model, or allows the user 102 to select a particular machine model from the model repository 710. Once execution of the particular machine model is complete and a recommendation is generated, the infrastructure transmits the recommendation directly (814) to the machine 1102, and transmits the recommendation directly (812) to the mobile device 754. Upon receiving the recommendation from the infrastructure 720, the computer implemented in the machine 102 uses the recommendation to generate instructions and causes the machine 102 to execute the instructions to cultivate the field. The transmission of the request to the infrastructure 720 (808), the transmission of the recommendation to the machine 1102 (814), and the transmission of the recommendation to the mobile device 754 (812) can be performed in real time.
[0180] The user can also use the terminal 750, the tablet computer 752, the server computer 756 ( Fig. 7A In the Fig. 8A ) and other computing devices configured to transmit and receive data to access infrastructure 720.
[0181] In one embodiment, infrastructure 720 includes one or more storage devices 722, one or more cloud system computers 723, one or more cloud-based networks 724, one or more computer servers 725, one or more computer workstations 726, and one or more communication networks 727. Each device 722-723, 725, and 726 may include one or more computer processors, one or more memory units, one or more caches, and one or more communication buses. The processors included in devices 722-723, 725, and 726 may vary and may have different processing capabilities. Similarly, the memory units included in devices 722-723, 725, and 726 may vary and may have different storage and processing capabilities.
[0182] Each device 722-723, 725, and 726 may store computer programming instructions that, when executed on a computer processor, cause the processor to execute a machine learning model stored in the model repository 710. In addition, each device 722-723, 725, and 726 may store computer programming instructions that, when executed on a computer processor, cause the processor to store information about the model in the model repository 710 and store information about the execution of the model in the model history log database 730.
[0183] Devices 722-723, 725, and 726 may be independent computing devices. Alternatively, some or all of devices 722-723, 725, and 726 may be implemented as a distributed system. The distributed system may be part of a cloud network 724, and components of the distributed system may communicate with each other via one or more communication networks 727.
[0184] In one embodiment, the model execution infrastructure 720 includes a Fig. 8A For example, the server 725 may be configured to communicate with a GPS system ( Fig. 8A Once the server 725 is in communication with, for example, a satellite-based GPS system ( Fig. 8A By establishing a communication connection (not shown), the server 725 can track the location and movement of the mobile device 754 that the user 102 is using. The server 725 can also track the location and movement of the tractor 1102 as it traverses the farmland and plants seeds on the field.
[0185] The components of the model execution infrastructure 720 may include general components, custom components, or both. These components may be used and reused to provide an execution platform for executing and re-executing various models.
[0186] If possible, certain components may be interchangeable. For example, if executing a model on a dual-core processor does not meet the expected execution time requirements, another processor (such as a quad-core processor) may be selected and used to execute the model, hoping that using the quad-core processor will reduce the execution time of the model.
[0187] Figure 8B An exemplary flow chart for selecting and executing an example machine learning model is depicted. This example is provided merely to illustrate one of many ways to select an execution infrastructure for a machine-based model, execute the model on the selected infrastructure, and utilize the output generated by executing the model. The output generated by executing the model can be used, for example, to control agricultural machinery, such as Fig. 8AThe tractor 1102 is shown. The depicted process may be performed by any computer-based processor, including Fig. 7A Management processor 705 is shown.
[0188] In step 880, the computer-based processor receives a request to browse a machine learning model stored in the model repository 710. The request may be received from, for example, a mobile device 754 operated by a user 102. The user 102 may be a scientist developing and testing models, or a crop grower who wants to know which seeds to plant in a field to obtain a high yield from the field.
[0189] In step 882, the processor generates and displays information about the machine learning model that has been stored in the model repository 710. This may include accessing the model repository 710, retrieving information about the model from the repository, generating a graphical representation of the retrieved information, and sending the graphical representation to the mobile device 754.
[0190] In some cases, the received request may provide information that the processor can use to limit the information about the models that should be retrieved from the model repository 710. For example, the request may specify that the user only wants to view operational research models or only wants to view deep learning models. According to another example, the request may specify that the user only wants to view models that were built or updated after a certain date, or models that have been built specifically to handle queries related to corn.
[0191] In step 884, the processor receives a selection of a model from a menu or list of models available to the user. The selection may indicate the model by providing a name, an identifier, or any other type of identification of the model.
[0192] In step 886, the processor determines the infrastructure requirements for executing the selected model. The requirements may specify, for example, the type of CPU that should be used to execute the selected model, or the storage space that needs to be allocated to execute the selected model. The requirements for the model may be retrieved from the model repository 710, or more specifically from metadata associated with the selected model stored in the model repository 710.
[0193] According to another example, the requirement can be extracted from the request itself. For example, the request can specify that the selected model needs to be executed on a quad-core processor, and the output generated by executing the selected model needs to be provided to the tractor 1102.
[0194] Additionally, during this step, the processor may generate a graphical representation of the infrastructure requirements and may transmit the graphical representation to the mobile device 754 .
[0195] In step 888, the processor determines or retrieves the definition and metadata of the selected model. This may include accessing the definition and metadata of the selected model stored in the model repository 710 and retrieving the stored definition and metadata. The definition may be used to determine, for example, the version of the selected model, input definitions, output definitions, model type, model classification, etc.
[0196] Also, in this step, the processor generates a graphical representation of the definition and metadata for the selected model and transmits the graphical representation to the mobile device 754. For example, the graphical representation may include information about the types of inputs that may be provided to the model, and the types of outputs that the model may generate.
[0197] In step 890, the processor receives inputs to be used to execute the selected model. The inputs may include, for example, specific values that the model can accept. The specific values may include soil type information for the farmland, geographic location information for the field, the type of crops to be planted on the field, etc.
[0198] If the received input satisfies the input definition included in the definition data structure saved for the selected model in the model repository 710, the processor proceeds to execute step 892. Otherwise, the processor may repeat step 888 and redisplay the input definition of the input accepted by the selected model.
[0199] In step 892, if the processor receives an indication or request to execute the selected model using the provided input, the processor proceeds to execute step 894. Otherwise, the processor proceeds to execute step 896.
[0200] In step 894, the processor initiates execution of the selected model using the input and on the model infrastructure. Once the execution of the model is complete, the model generates output information (referred to herein as output). The output may include, for example, a recommendation to address the request received in step 880. For example, if the request is for a sowing recommendation, the output may provide the type of seed recommended for the farmland as a recommendation, and the timing of sowing to obtain the highest yield. The recommendation may also include instructions for sowing, such as spacing instructions for sowing.
[0201] In step 895, the processor transmits the output to, for example, the mobile device 754. The output may also be transmitted to a computer or controller controlling a machine such as the tractor 1102. For example, if the output includes sowing instructions for sowing seeds using an agricultural planter, the output may be transmitted directly to a controller mounted on the planter to adjust the spacing between the sowing process pipes before the planter begins the sowing process.
[0202] If, in step 896, the processor determines that a request to modify the input of the model has been received, the processor proceeds to execute step 890. Otherwise, the processor proceeds to execute step 897.
[0203] If, in step 897 , the processor determines that a request to select a different model has been received, the processor proceeds to execute step 898 ; otherwise, the processor stops executing in step 899 .
[0204] In step 898, the processor receives a selection of another model and then proceeds to execute step 886, where the processor determines infrastructure requirements for the newly selected model.
[0205] This process is repeated each time the processor receives a request to select and execute a machine-based model. The request may be sent, for example, from a scientist developing and testing a new machine-based model or developing and enhancing the platform 700. The request may also be received, for example, from a crop grower seeking recommendations for improving agricultural practices on their fields.
[0206] 3.3. Example Model History Log
[0207] In one embodiment, the model history log 730 is used to store specific inputs and outputs for the execution of the machine-based model. For example, if a deep learning module is added to the model repository 710, a record for the model is created in the model history log 730. The inputs that the model is configured to accept and the outputs that the model can generate are stored in the record in the model history log 730. When the module is subsequently executed, the actual input values and the actual output values are added to the record stored in the model history log 730 of the model. In addition, when the model is executed on another input value, the new actual input value and the actual output value are added to the record stored in the model history log 730 of the model.
[0208] As a result, the records stored in the model history log 730 for a model are a comprehensive record of all inputs and all outputs used and generated by executing the model in the platform 700. The records generated and stored for all models form the model history log 730.
[0209] In one embodiment, the records of a model stored in the model history log 730 are aggregated, retrievable, searchable, and modifiable.
[0210] In one embodiment, the records for all models stored in the model history log 730 are aggregated, retrievable, searchable, and modifiable. The aggregate record may, for example, be searchable for specific outputs, specific recommendations, specific dates, etc. In addition, complex analysis may be performed on the aggregate record to determine whether the latest version of a particular model performs better than an older version of the model.
[0211] 4. Scalability considerations
[0212] In one embodiment, a method provides a wide range of scalability options. For example, the method allows determination of an execution infrastructure platform for a model, as well as one or more alternative execution infrastructure platforms on which the model may also be executed. It may be specified that the minimum execution infrastructure platform for a particular model may include, for example, dual CPUs, a 32GB memory card, a 16MB TransFlash card, and 500GB disk storage space. However, it may also be specified that an alternative execution infrastructure platform for a model may include a distributed cloud system equipped with a more powerful CPU and more storage space.
[0213] Scalability can also refer to using the same execution infrastructure platform to execute more than one machine-based model simultaneously. For example, a distributed cloud system including distributed supercomputers or the like can be used to execute two or more machine-based models simultaneously and in parallel, because such systems typically provide a large amount of computing and storage resources.
[0214] In one embodiment, the platform 700 described herein allows scientists to develop and deliver a variety of model-based solutions that are scalable, widely accessible, easily searchable, and modifiable.
[0215] In one embodiment, the model repository 710 is enhanced and extended. For example, additional features are added to the model repository 710 to enhance the searchability and applicability of the repository. In addition, the process from the design repository to the deployment repository is optimized so that the duration between design and deployment is relatively short. The model repository 710 can also be extended by adding new models and new versions of models that have already been added.
[0216] Enhancements may also include improvements to the model execution infrastructure 730. The improvements may include the addition of new hardware components, new software components, and the addition of the latest technologies.
[0217] Enhancements may include enhancements for enhancing the user experience. This includes adding polished interfaces, intelligent graphical user interfaces, and other tools to help users navigate, use, and modify the platform 700. This may also include adding automatic generation of new models and automatic generation of new versions of existing models.
[0218] Enhancements may also include modifications to the design of the model repository 710, model execution infrastructure 720, and model history log 730. The repository may be designed to provide easy access to models and efficient organization and categorization of models.
[0219] Enhancements may also include improvements for designing platform 700 as a turnkey system that is robust enough to meet a wide range of expectations and requirements. A turnkey system may include advanced components that can take advantage of new technological advances.
[0220] 5. Example Implementation
[0221] Fig. 9 An example diagram of a user interface for executing a machine learning algorithm is depicted.In one embodiment, the platform 700 is accessible from a user device 901 configured with a graphical user interface 910 .
[0222] Graphical user interface 910 is generated to provide interactivity with platform 700. Graphical user interface 910 can be used to facilitate inputting queries to platform 700 and obtaining responses to the queries from platform 700. Graphical user interface 910 can be a separate component from platform 700, such as Fig. 9 Alternatively, graphical user interface 910 may be a part of platform 700 .
[0223] In one embodiment, graphical user interface 910 is configured to generate and display graphical objects that can be used to provide queries to platform 700. Queries can include requests to browse (920) model repository 710, browse model history log 720, request information about models stored in model repository 710, modify (940) a request sent to platform 700, request (960) execution of a particular model in model execution infrastructure 720, and the like.
[0224] Graphical user interface 910 may also be configured to receive and display responses to queries, including, for example, responses 930 to browse requests, responses 950 to modify requests, and / or responses 970 to execute requests.
[0225] In one embodiment, graphical user interface 910 is configured to communicate received queries to platform 700 and, when responses to the queries are received, to communicate the received responses to the requestor.
[0226] The graphical user interface 910 may be configured to receive various recommendations, instructions, and notifications from the platform 700 and to display the recommendations / instructions / notifications on a display device. For example, the graphical user interface 910 may be configured to receive a recommendation generated by executing a model on the model execution infrastructure 720 and indicating a recommended strategy for a farm field. The recommendation may be displayed on a display device such as a mobile device 754. In one embodiment, the recommendation may also be transmitted to a computer or controller that controls agricultural equipment such as a planter, combine, harvester, or plow.
[0227] 6. Benefits of Certain Embodiments
[0228] In one embodiment, a method and tool use a computer-based platform to store, catalog, and execute machine-based models. The platform provides a computer foundation for cataloging common definitions and metadata of models, storing models in a repository, and browsing and selecting models from the repository.
[0229] In one embodiment, a method for executing a machine learning algorithm provides functionality for customizing agricultural models, cataloging, maintaining, and managing a large number of customized models, and facilitating execution of the models to provide recommendations to crop growers.
[0230] In one embodiment, a platform for storing, executing, and cataloging machine-based models includes a model repository, a model execution infrastructure, and a model history log. The model repository is a computer-based digital data repository configured to store models. The model execution infrastructure includes hardware and / or software components that collectively provide a platform for executing machine-based models. The model history log is a computer-based digital log containing information about the execution of machine-based models.
[0231] In one embodiment, a method for executing a machine learning algorithm is implemented for an agricultural application. For example, the method may allow a crop grower to browse models stored in a model repository, select a model from the repository, provide input for the selected model, and, using the provided input, initiate execution of the selected model on a model infrastructure to provide recommendations for the grower.
[0232] 7. Extensions and Replacements
[0233] In the foregoing specification, embodiments of the present disclosure have been described with reference to many specific details that may vary depending on the implementation. Accordingly, the specification and drawings are to be regarded as illustrative rather than restrictive. The scope of the present disclosure and what the applicant intends as the sole and exclusive indication of the scope of protection of the present disclosure is the literal and equivalent scope of the claims issued from this application, in the specific form of those claims, including any subsequent amendments.
Claims
1. A computer-implemented data processing method providing an improvement in performing a machine learning process on digital data representing physical properties related to agriculture, the method comprising: Incorporate machine learning models into digital model repositories, including: Receive a request to store a machine learning model; parsing the request and identifying model artifacts and information about the machine learning model; determining inputs and outputs for the machine learning model; determining infrastructure requirements for executing the machine learning model; storing the infrastructure requirements for the machine learning model in the digital model repository; determining a definition and metadata for the machine learning model; associating the definition and metadata with the machine learning model; storing the machine learning model in association with the definition and metadata in the digital model repository, receiving, from a computing device, a request to browse a machine learning model stored in the digital model repository; retrieving from the digital model repository and transmitting to the computing device information about the machine learning model stored in the digital model repository; receiving, from the computing device, a selection of a particular model from among the machine learning models and receiving a particular input for the particular model; executing the specific model on the specific input to generate a specific output using resources available in a model execution infrastructure platform; and The specific output is transmitted to a computer configured on the agricultural machine to control the agricultural machine when the agricultural machine performs an agricultural task in a farm field.
2. The method according to claim 1, further comprising: transmitting the specific output to the computing device; determining whether any modification to the particular input for the particular model is requested; In response to receiving an indication that modification of the particular input is requested, receiving a modified input for the particular model, executing the particular model on the modified input to generate a modified particular output, and transmitting the modified particular output to the computing device.
3. The method of claim 1 , wherein the information about the machine learning model stored in the digital model repository includes model artifacts and definitions and metadata associated with the model artifacts; wherein the model artifact comprises one or more of the following: an operational research model artifact, a machine learning model artifact, a deep machine learning model artifact, or a physical model artifact; The definitions and metadata associated with a particular model artifact include one or more of the following: model name, input definitions, output definitions, infrastructure requirements, runtime metrics, expected execution duration, expected accuracy, state information, model classification, model version, model type, or model owner name.
4. The method according to claim 1, further comprising: Create a record for the machine learning model and store the record in a model history log; wherein the record for the machine learning model includes one or more of the following: inputs to executing the machine learning model, outputs to executing the machine learning model, or information specific to executing the machine learning model.
5. The method according to claim 4, further comprising: receiving update information for the record of the machine learning model; wherein the update information includes new inputs and new outputs for executing the machine learning model; In response to receiving the update information, the record for the machine learning model is updated using the new input and the new output.
6. The method according to claim 1, further comprising: Receive new versions of machine learning models; wherein the new version of the machine learning model includes new model artifacts and new definitions and metadata; In response to receiving the new version of the machine learning model, storing the new model artifact and the new definition and metadata in the digital model repository; and updating a record for the machine learning model in a model history log.
7. The method of claim 1, wherein the computing device is any type of device configured to generate, transmit, and receive digital data; wherein the computing device comprises any of the following: a personal computer, a laptop, a smart phone, a tablet, or a computer server.
8. The method of claim 1, wherein the model execution infrastructure platform comprises one or more of: a computer processor, a memory unit, a storage device, a communication bus, a communication network, a cloud storage device, a cloud distributed system, a computer server, or a mainframe computer.
9. The method according to claim 2, further comprising: The computing device is caused to: generate and display a graphical user interface, generate a graphical representation of the modified specific output, and display the graphical representation of the modified specific output using the graphical user interface.
10. One or more non-transitory computer-readable storage media storing one or more computer instructions that, when executed by one or more computer processors, cause the one or more computer processors to perform: Incorporate machine learning models into digital model repositories, including: Receive a request to store a machine learning model; parsing the request and identifying model artifacts and information about the machine learning model; determining inputs and outputs for the machine learning model; determining infrastructure requirements for executing the machine learning model; storing the infrastructure requirements for the machine learning model in the digital model repository; determining a definition and metadata for the machine learning model; associating the definition and metadata with the machine learning model; storing the machine learning model in association with the definition and metadata in the digital model repository, receiving, from a computing device, a request to browse a machine learning model stored in the digital model repository; retrieving from the digital model repository and transmitting to the computing device information about the machine learning model stored in the digital model repository; receiving, from the computing device, a selection of a particular model from among the machine learning models and receiving a particular input for the particular model; executing the specific model on the specific input to generate a specific output using resources available in a model execution infrastructure platform; as well as The specific output is transmitted to a computer configured on the agricultural machine to control the agricultural machine when the agricultural machine performs an agricultural task in a farm field.
11. The one or more non-transitory computer-readable storage media of claim 10 storing additional instructions for: transmitting the specific output to the computing device; determining whether any modification to the particular input for the particular model is requested; In response to receiving an indication that modification of the particular input is requested, receiving a modified input for the particular model, executing the particular model on the modified input to generate a modified particular output, and transmitting the modified particular output to the computing device.
12. The one or more non-transitory computer-readable storage media of claim 10, wherein the information about the machine learning model stored in the digital model repository comprises model artifacts and definitions and metadata associated with the model artifacts; wherein the model artifact comprises one or more of the following: an operational research model artifact, a machine learning model artifact, a deep machine learning model artifact, or a physical model artifact; The definitions and metadata associated with a particular model artifact include one or more of the following: model name, input definitions, output definitions, infrastructure requirements, runtime metrics, expected execution duration, expected accuracy, state information, model classification, model version, model type, or model owner name.
13. The one or more non-transitory computer-readable storage media of claim 10 storing additional instructions for: Create a record for the machine learning model and store the record in a model history log; wherein the record for the machine learning model includes one or more of the following: inputs to executing the machine learning model, outputs to executing the machine learning model, or information specific to executing the machine learning model.
14. The one or more non-transitory computer-readable storage media of claim 13 storing additional instructions for: receiving update information for the record of the machine learning model; wherein the update information includes new inputs and new outputs for executing the machine learning model; In response to receiving the update information, the record for the machine learning model is updated using the new input and the new output.
15. The one or more non-transitory computer-readable storage media of claim 10 storing additional instructions for: Receiving a new version of a machine learning model; wherein the new version of the machine learning model includes new model artifacts and new definitions and metadata; In response to receiving the new version of the machine learning model, storing the new model artifact and the new definition and metadata in the digital model repository; and updating a record for the machine learning model in a model history log.
16. One or more non-transitory computer-readable storage media according to claim 10, wherein the computing device is any type of device configured to generate, transmit and receive digital data; wherein the computing device includes any of the following: a personal computer, a laptop computer, a smart phone, a tablet computer or a computer server.
17. The one or more non-transitory computer-readable storage media of claim 10, wherein the model execution infrastructure platform comprises one or more of: a computer processor, a memory unit, a storage device, a communication bus, a communication network, a cloud storage device, a cloud distributed system, a computer server, or a cloud computing system.
18. The one or more non-transitory computer-readable storage media of claim 11 storing additional instructions for: The computing device is caused to: generate and display a graphical user interface, generate a graphical representation of the modified specific output, and display the graphical representation of the modified specific output using the graphical user interface.
Citation Information
Patent Citations
Systems, methods and apparatus for multi-row agricultural implement control and monitoring
US20150094916A1
Methods for agronomic and agricultural monitoring using unmanned aerial systems
US20160050840A1
Generating agricultural information products using remote sensing
US8712148B2
Planter monitor system and method
US8738243B2
Automated soil measurement device
US8767194B2
