CN116010066A - RPA robot and implementation method - Google Patents

Abstract

The invention provides an RPA robot and an implementation method, which belongs to the technical field of RPA robot software, wherein the RPA robot is in communication connection with a local system and comprises: the design device comprises: writing rpa flow according to the service, and storing the flow as a flow file after writing is completed; carrying out parameter maintenance on the flow, executing the flow and carrying out log tracking; an actuator: automatic handling of the driving business; providing an interface for calling a local system and acquiring flow data through the interface; feeding back an execution result to the local system; and (3) a controller: receiving a flow file, creating a flow task queue, distributing the task queue to an executor, and synchronizing the flow file to the executor; OCR technology is adopted to identify and analyze the popup characters, and machine learning algorithm is adopted to identify and classify the popup characters. The invention integrates a plurality of artificial intelligence technologies, and realizes the RPA robot specially used for automatically transacting the business aiming at the characteristics of the business.

Description

RPA robot and implementation method

Technical Field

The invention relates to the technical field of RPA robot software, in particular to an RPA robot and an implementation method.

Background

Each enterprise needs to pay social security for the employee. After the enterprise is participating in the insurance, the business to be declared is required to be transacted on the official website of the social security every month. For large human resource enterprises, the personnel fluctuation amount per month is hundreds or thousands, and a great deal of manpower is required to be spent on the declaration business of social security. Since most of the social security systems in the whole country do not provide program access interfaces to the outside, even if some enterprises have human resource systems, the enterprises cannot realize the butt joint with the social security systems.

With the development of RPA technology, RPA robots are applied in the fields of finance, insurance, finance and the like. Chinese invention patent name: an RPA robot process automation realization system and method, patent number: CN113254174a discloses a system and a method for implementing RPA robot process automation, the system is deployed at a cloud, the system comprises: the system comprises a flow design module, a control management module and at least one RPA robot cluster; the process design module is used for receiving the operation of a user on at least one graphic component through a browser, designing an automatic process which comprises at least one task node and represents a target industry problem solution, wherein at least one node in the automatic process is represented by a graphic corresponding to the graphic component, and one graphic component displayed in the browser is represented by an RPA robot in the cloud; and the control management module is used for calling the RPA robot represented by the graphic component in the automatic flow and executing the task in the automatic flow. However, in the application of the social security field, only a few places are used for realizing partial work automation of the alternative social security business sponsor department, and no mature application cases exist nationwide in the aspect of realizing automatic social security declaration instead of enterprises and reducing the workload of enterprise social security specialists. The reasons are mainly as follows:

1. the number of websites needing to be docked is large, the flow is large, and the websites are frequently updated;

2. the website has a plurality of popup windows and a plurality of flow branches;

3. the human resource system connected with the enterprise is difficult to dock, and the whole process of automatic declaration of social security cannot be completely realized;

4. the RPA flow is difficult to operate and maintain;

for the above reasons, the RPA robot products which are popular in the market at present have no landable solution in the scenes of more processes, large regional difference, frequent upgrading, poor stability, human resource system docking and the like facing social security business reporting, and even if an automatic process can be barely prepared, the operation and maintenance cost for keeping the process running stably is far greater than economic benefit.

Disclosure of Invention

Aiming at the problems in the prior art, the invention merges various artificial intelligence and automation technologies, and aims at the characteristics of the online declaration service, thereby realizing the RPA robot product specially used for automatic declaration and being capable of perfectly helping enterprises to conduct automatic business handling.

The technical scheme adopted by the invention is as follows:

an RPA robot in communication with a local system, the RPA robot comprising:

the design device comprises: the method is used for compiling rpa flow according to the business, and storing the flow as a flow file after compiling is completed; carrying out parameter maintenance on the flow, executing the flow and carrying out log tracking;

an actuator: the automatic transaction is used for driving the business; providing an interface for calling a local system and acquiring flow data through the interface; feeding back an execution result to the local system;

and (3) a controller: the method comprises the steps of receiving a flow file and performing scheduling execution on the flow; creating a flow task queue, distributing the task queue to an executor, and synchronizing a flow file to the executor; and (3) identifying and analyzing the popup characters by adopting an OCR technology, storing the popup characters into a knowledge base managed by the controller, and identifying and classifying the popup characters by adopting a machine learning algorithm.

An implementation method of an RPA robot comprises the following steps:

the robot designer operates the rpa procedure according to the business configuration, store the operation rpa procedure to the robot controller;

the controller controls and manages rpa flow and the robot executor, receives a flow execution request sent by the local system, controls the executor to execute according to the requested flow information, and receives feedback information of the executor;

the robot executes rpa flow, acquires the latest flow from the controller, invokes a local system interface to acquire flow data, automatically transacts business, and feeds back the transacting result to the local system.

The beneficial effects of the invention are as follows:

1. according to the invention, abstract and model each declaration business process, business key points are extracted, other non-keys are abandoned, a popup window shielding assembly is realized by adopting selenium+embedded js, all non-key information is shielded and ignored, normal operation of declaration business can be kept, and the problems of more websites, more processes and frequent website upgrading which need to be docked are solved;

2. the robot collects various reminding, prompting, alarming, confirming and informing information in the process of executing the flow, recognizes and analyzes characters by using an OCR technology, stores the characters into a knowledge base managed by a controller, recognizes and classifies the information by adopting a machine learning algorithm, analyzes out which useful information is fed back to a user by the robot, and solves the problem of a plurality of popup windows in a website;

3. the invention provides a plurality of fixed declaration interface components and supports custom development interface components, which are used for supporting a robot to be in butt joint with a local system;

4. according to the invention, the Java CV technology is used for realizing the screen recording of the robot operation, and the screen recording video and the operation date are uploaded to the server in a multithreaded segmented uploading mode, so that operation and maintenance personnel can monitor the running state of the flow and cause of problems.

Drawings

FIG. 1 is a schematic diagram of an RPA robot designer of the present invention;

FIG. 2 is a schematic diagram of the RPA robot actuator of the present invention;

FIG. 3 is a schematic diagram of the RPA robot controller of the present invention;

FIG. 4 is a flow chart of an implementation method of the RPA robot of the invention;

FIG. 5 is a schematic diagram of the workflow of the RPA robot designer of the present invention;

FIG. 6 is a schematic diagram of the workflow of the RPA robot controller of the present invention;

FIG. 7 is a schematic diagram of the workflow of the RPA robot actuator of the present invention;

fig. 8 is a flow chart of a method for processing popup windows by the RPA robot controller according to the present invention.

Detailed Description

The invention is further illustrated by the following examples in conjunction with the accompanying drawings: in order to clearly illustrate the technical features of the present invention, the present invention will be described in detail below with reference to the following detailed description and the accompanying drawings. The following disclosure provides many different embodiments, or examples, for implementing different structures of the invention. In order to simplify the present disclosure, components and arrangements of specific examples are described below. Furthermore, the present invention may repeat reference numerals and/or letters in the various examples. This repetition is for the purpose of simplicity and clarity and does not in itself dictate a relationship between the various embodiments and/or configurations discussed. It should be noted that the components illustrated in the figures are not necessarily drawn to scale. Descriptions of well-known components and processing techniques and processes are omitted so as to not unnecessarily obscure the present invention.

The invention integrates Sikulix, selenium, electron, paddleOCR, javaCV, BPMN and other artificial intelligence and automation technologies, takes the local human resource system to declare social security as an example, and describes the technical scheme of the invention.

As shown in fig. 1 to 3, the RPA social security reporting robot provided by the present invention follows the RPA design principle and is divided into three parts: controller, designer, executor.

The design device comprises: the designer is an application of a pure front-end technical stack, is used for the developer to design social security reporting process, mainly uses electron, bmpn and other technologies, and can directly call an executor to carry out debugging execution of the process. The designer provided by the invention mainly comprises: the system comprises a flow design module, a code debugging module and a flow publishing module.

Flow design module flow creation: the method is mainly used for creating rpa business processes, and generating and storing process files;

drawing a flow chart: providing a drawing tool for drawing a business flow chart;

code editing: and configuring required components and coding codes for each node of the service flow chart.

Code debugging module parameter maintenance: the system is used for maintaining various parameters which are required to be input in the business process;

the flow is executed: the method is used for running the business process needing to be debugged;

log tracking: various logs in the running process of the flow are displayed, including error report logs, warning logs, information record logs and the like.

Compiling a flow release module code: compiling a flow code, and giving a prompt if an error exists;

and (3) packaging the flow: packaging the flow to generate a flow file which can be identified and operated by an executor;

and (3) issuing a flow: and issuing the flow to the controller.

The open source technique used by the designer of the invention comprises: electron front-end reference frame: electron is a desktop application framework that uses JavaScript, HTML and CSS to build a cross-platform. The designer uses Electron to construct the system main body, and the development of each function is completed. BPMN flow diagram framework: BPMN, namely business process modeling and labeling, is a markup language which is understood by users based on business process diagram definition, builds a standardized bridge between business process design and process realization, and is suitable for numerous groups such as business analyzers, software developers, business managers, monitors and the like. The designer uses a standard BPMN flow diagram framework as a tool to draw the flow diagrams.

An actuator: the system is deployed on a machine or a server on a user site and used for driving a social security official system to realize automatic business handling, an interface for calling a human resource system can be provided, social security declaration data can be obtained through the interface, an executor needs to be developed mainly by using java, various automatic execution and flow driving technical stacks are integrated, and finally, the system is packaged into an exe file which can be installed locally on the user. The actuator mainly comprises: an executor interface, flow management, business component management, execution engine management and the like. The part encapsulates a plurality of fixed components for social security declaration, including: a system login component, a popup screen component, a file upload component, a file download component, a service call component, and the like.

Actuator interface login authentication: the method is mainly used for logging the executor on the controller. The executor calls the controller interface to carry out login authentication, and after the authentication is passed, the flow task is obtained from the executor and is scheduled by the executor;

history viewing: the method is mainly used for inquiring the execution history record and the log;

and (3) flow management: for managing the business processes downloaded locally from the executor, this function may be used to initiate execution of the process.

Service component module code logic component: various general business logic processing code blocks of the package;

service operation component: various generic business operation code components packaged;

the predefined function: various general and service-independent general functions of the package.

Business process module block management: each node for managing rpa flow;

and (3) flow sequence management: the method is used for managing the sequence of each node of the rpa process;

and (3) circulation condition management: for managing branching conditions of rpa flow branching nodes.

The execution engine manages a screen-based execution engine: and identifying the executor for sending the screenshot. Inputting identification of a picture verification code logged in by a system;

web-based execution engine: various element information in the web page is identified. For example, identifying the location of a user name entry box;

the flow execution engine: the method is used for driving the business process to be executed step by step according to the actual business data;

a screen recording engine: the method is used for recording the operation of the current screen into a video, so that the execution history can be conveniently checked;

script execution engine: and the method is used for executing the java script in the flow chart.

The open source technology adopted by the actuator of the invention comprises the following steps: sikuliX automation tool open source framework: any application software interface which can be seen by human eyes can be automatically controlled, so that automation of various software operation steps is realized. Among them, the actuator is mainly used as a screen-based execution engine to identify various element information on a screen. The Flowable workflow engine: the workflow engine can drive and execute the flow chart based on the bpmn standard, and the main function of the workflow engine in an executor is to drive the step-by-step execution of the flow task. Selenium automation web tool framework: is an open-source Web automatic testing tool. The primary role in the executor is to identify the handle of the web page element. Java CV computer vision library open source packaging library: a variety of common libraries and utility classes in the field of computer vision are packaged. In the actuator, the main user records the operation video of the screen.

And (3) a controller: the method comprises the steps of deploying on a cloud server, scheduling and executing flows, and scheduling an executor and a controller. The flow designed by the designer is issued through the controller, the executor is registered, and the flow is synchronized to the executor. Certain flow operations are performed that require execution on the server (e.g., ocr identification, etc.). The controller is a web engineering of a BS architecture, and is responsible for providing expansion capability for the executor, reporting the issuing process and other functions, and providing page realization to manage the upstream and downstream. The system mainly comprises a web application module, a web service module, an execution engine management module and the like. The controller is internally provided with a knowledge base and a machine learning engine, and various popup window contents in the running process of the system are collected and a processing scheme is formulated.

web application module system login: a function page for a manager to log in the controller;

and (3) flow management: for managing various rpa procedures. The method comprises the steps of newly building, uploading, logging out, upgrading and the like of the flow;

and (3) managing flow tasks: the method is used for managing various rpa processes to execute tasks, creating a task queue and distributing specific executors to execute;

configuration management: configuring various parameter information;

knowledge base management: for managing various information collected in the knowledge base and for manually maintaining its way of handling.

web service module rights authentication: when the executor calls the controller, the validity and the authority of the executor need to be checked;

calling an executor interface: the controller calls an actuator interface for controlling the execution condition of the actuator;

the capability interface provides: the controller provides various capability interfaces including a flow downloading interface, a state inquiring interface and the like.

The execution engine manages the OCR engine: the method is used for screen capture identification and verification code screen capture identification;

knowledge base analysis engine: the system is used for collecting and analyzing various popup window data characteristics and matching execution results;

machine learning engine: for formulating a solution based on the data characteristics.

The open source technology adopted by the controller of the invention comprises the following steps: the PaddleOCR recognition library is a ocr recognition tool library based on hundred-degree flyers, and is mainly used for recognizing screenshot information and verification code screenshot information in the system. The robot deployment adopts a SaaS cloud deployment mode, and can be used for simultaneously docking human resource systems of a plurality of companies.

As shown in fig. 4, the present invention provides a method for implementing an RPA robot, which takes a declaration social security as an example, and the method includes:

s100: operating an rpa process of the official platform service according to service configuration by using a designer, and releasing the rpa process of the social security service to a controller after the design is finished, wherein the rpa process of the social security service comprises the following steps: personnel participating in the insurance process, personnel stopping the insurance process, payment verification process, payment bill printing process, participating in the insurance proof printing process and the like. Each declaration service corresponds to an rpa flow;

s200: the method comprises the steps of managing a flow and an actuator by using a controller, providing an interface for the actuator and an external system (local human resource management system) to call, receiving a flow execution request sent by the external system, searching for a proper actuator to execute according to the requested flow information, and receiving feedback information of the actuator;

s300: starting rpa flow, acquiring the latest flow from the controller, acquiring corresponding parameter information from the human resource system, automatically operating the social security website to transact business, and feeding back the transacting result to the human resource system.

The RPA robot automatic reporting social security is mainly aimed at medium and large enterprises, and the enterprises generally have local human resource systems. The automatic declaration service is initiated from the human resource system. The business manager operates the local human resource system instead of directly operating the robot, so that the execution progress and the execution result of the robot can be monitored in real time through the local human resource system. Most RPA robot products realize the simulation of human operators to operate official websites, source data is obtained by reading fixed files, databases or crawling websites, other interfaces are provided less, and information cannot be fed back in real time. Thus, seamless connection with a local human resource system is difficult.

The invention specifically exemplifies each step of the personnel participating in and reporting business:

the method disassembles, refines and encapsulates the social security declaration process, forms common components for public parts, and supports regional customized processing. The method is used for carrying out modularized processing on login parts, exception handling parts, file downloading, uploading, browser closing and the like, and all business processes are shared. Corresponding customized components are set for each component according to regions, for example, a user name and password mode login in city A, a mobile phone and verification code mode in city B and a payment device code login in city C are required to be customized into different components. The technology unifies the social security services of all regional differences in terms of large flow, and shields the regional differences. Abstract and model each declaration business process, extract business key points, discard other non-keys, change business process anyway, and the RPA process is not needed to be changed as long as the key points are not changed. The core of the staff-participating business is, for example: logging in the system, opening the augmentation page, importing the augmentation file, feeding back the error reporting information, and storing the correct information. This technique minimizes the impact of system upgrades on the process because system upgrades are only operational level changes and rarely result in changes to the subject business process. A plurality of popup windows irrelevant to the business process are frequently arranged on the social security website, and the popup window shielding assembly is realized by adopting the selenium+embedded js, so that all non-key information is shielded and ignored, and the normal operation of the declaration business can be kept. The technology has good coping effect on the conditions of social security website upgrading, large regional difference and the like.

The configuring rpa process described in S100, as shown in fig. 5, takes the office to participate in the insurance declaration service as an example, includes the following contents:

s110, newly-built business flow: a new process is created by using a designer (new process), named as a social security participating process (taking personnel participating and paying as an example), and stored in a designated folder, at this time, a folder named as a social security participating process is generated in the folder, and various process files including a process configuration file, a process graphic file and the like are arranged in the folder.

S120, drawing a business flow chart: according to the flow of the participation and protection business of the staff logging in the official website of the social security, a designer is used for drawing a flow chart of the whole handling process. Comprising the following steps: logging in the system, opening the participating and protecting function, importing participating and protecting files, feeding back error information, storing correct information, popup window shielding processing, executing abnormal processing and other nodes, and connecting the nodes by using a connecting wire. And designing the parameter entering and exiting information of each node.

Logging in a system node: entering into the ginseng: social security website address, user login parameters. And (3) ginseng: success flag, exception information.

Opening the participating and protecting function node: entering into the ginseng: and no. And (3) ginseng: success flag, exception information.

Importing the participating and protecting file node: entering into the ginseng: the imported excel personnel participating in the file. And (3) ginseng: the participation and protection check result, the success mark and the exception information (excel file format is provided by a social security website and is generated in a human resource system, and the rpa flow does not carry out special treatment).

And (3) feeding back error reporting information nodes: entering into the ginseng: none, go out ginseng: an excel table (one row for each person) of error information, a success flag and exception information. The error reporting information refers to that some information which does not meet the participation and protection conditions in the participation and protection process is fed back to staff, and is not only abnormal information processing.

Saving the correct information node: entering into the ginseng: none, go out ginseng: an excel list of successful personnel (one row for each person), a success flag and exception information.

Exception handling node: entering into the ginseng: abnormal information, parameter: exception code, exception information.

Bullet window shielding node: entering into the ginseng: none, go out ginseng: the flow execution flag (1 continues execution 2 waits for 3 to end execution).

S130 configures general components: opening each flow node, starting execution steps in the design node, opening the flow for the flow nodes supported by the general component such as system login and exception handling, dragging the flow into the general component, and setting handles in websites corresponding to each operation in the component. Taking login website as an example, the handle to be set includes: a user name input box handle, a password input box handle, a verification code picture handle, a verification code input box handle, a login button handle, and the like.

S140, writing a flow code: for code and business logic within the authoring process that is not handled by the generic components, it is also necessary to identify handles to web site related content.

S150, debugging and publishing the flow: setting default entry of the flow, debugging the operation of the flow by using a flow debugging function, and releasing the flow to the controller by using a flow release function after the debugging is completed.

The execution of the scheduling task in S200, as shown in fig. 6, takes the case of transacting personnel participation and declaration business through the personnel resource system call controller, includes the following contents:

s210 receives a task: and receiving the dispatching command and parameters transmitted by the personnel system. And checking flow information to be executed, and establishing a flow task. The parameters include: process id, tenant id, region code to be executed;

s220 assigns tasks: and searching the task queues of all the executors meeting the execution conditions. The task queue is preferably selected to be empty, and then the actuator with the shortest task queue is selected to hang the task on the corresponding queue. The task queue is arranged to execute the previous task according to the first-in first-out principle;

s230 monitors the task: and monitoring the execution condition of the task. And collecting abnormal information in the running process. Such as no response by the server on which the actuator resides, network disruption, request timeout, etc. The state of the actuator can be queried in real time through the query function provided by the controller.

The scheduling task execution of S300 is as shown in fig. 7 below: taking the example of transacting social security participation and declaration business through an executor, the method comprises the following contents:

s310 starts the flow: calling a controller service, acquiring a to-be-executed flow task from a current execution queue, judging the validity of the task after acquiring the to-be-executed flow task, and downloading the latest version of flow from an executor to a local area;

s320, acquiring flow execution parameters: and calling the personnel system to acquire parameters of flow execution, including website login information, participant information and the like. The information of the participants is Base64 code, converted into excel file and stored in the C disc fixed directory.

The website login information includes: company id, official website login mode, user name, password, company name, reporting month, official website name, official website number, official website address, flow id, flow category.

The information of the participating staff comprises: valid certificate number, certificate type, name, gender, date of birth, ethnicity, job format, cell phone number, nature of the household, residence location, augmentation year and month, payment wages, business endowments, industry loss insurance, industrial injury insurance, file date of birth, contact name, contact phone, personal identity, postal code, communication address, cultural degree, academic specialty, residence province level, residence location city level, residence location county level, residence location supplemental information, health status, marital status, professional technical job level, administrative job, national professional qualification level, personal number, joining date, affiliated streets, affiliated communities, etc.

S330 flow performs: starting to gradually execute according to a preset flow: the method comprises the steps of system login, participating in protection function opening, participating in protection file importing, error information feedback, submitting and storing and the like. The specific implementation process is as follows:

the system login is specifically as follows:

(1) Opening a google browser;

(2) Inputting a website address in a browser address bar, turning to the website, and opening a login page;

(3) After the executor judges that the login page appears, respectively inputting a user name and a password in a user name and password frame;

(4) Intercepting a picture corresponding to the picture verification code, and calling the verification code in other pictures served by the controller;

(5) Filling the verification code received with feedback into a verification code input box;

(6) Identifying and clicking a login button;

(7) When the occurrence of the system home page is detected, the unit name is acquired from the upper right corner of the system, whether the unit name is consistent with the company name in the input parameters is checked, if the unit name is inconsistent, the exception is thrown out, and if the unit name is consistent, the login is successful.

The function of opening the participation and protection is specifically as follows:

(1) Clicking the social security-social security reporting function of the left menu in sequence to enter a social security reporting page;

(2) Identifying whether an increaser declaration link exists on the page, and if so, waiting to enter the increaser declaration page;

(3) And after judging that the page opening of the increaser declaration is completed, the function of opening the participation and protection is completed.

The imported participating files are specifically as follows:

(1) Identifying and clicking a leading-in newspaper disc function at the bottom of the page, and waiting to enter batch newspaper disc pages;

(2) After judging that the pages of the batch newspaper boards are opened, clicking the function of importing newspaper board files at the bottom of the batch newspaper boards;

(3) After clicking the newspaper disc function, identifying and clicking the selection file in the popped selection box, then inputting the path and name of the newspaper disc excel file, and clicking the determination button. The file starts to be imported, and a progress bar appears;

(4) Waiting for the progress bar on the page to disappear. And importing the newspaper file.

The feedback error information is specifically as follows:

(1) The current page switches tab pages into an error information table component, judges whether records exist or not, and enters a submitting and storing flow if not;

(2) If the right button is clicked on the error information table component, selecting another storage mode, and storing the information into a specified excel table under a specified directory;

(3) And calling the human resource system service to feed back the excel table to the human resource system.

The submitting and saving is specifically as follows:

(1) And switching the tab page to the correct information table component by the current page, judging whether records exist or not, and ending the flow if not. The feedback information is 'no person to be added';

(2) If yes, judging and clicking a save button, and waiting for entering an uploading material page;

(3) After entering the uploading material page, judging and clicking a next button, and waiting for entering a waiting material auditing page;

(4) After entering the page waiting for checking materials, the page waiting for about several seconds automatically disappears, if the page waiting for 10 seconds does not disappear, the checking result button is identified and clicked, and the page waiting for entering the checking result page is waited;

(5) Entering a checking result page, clicking a printing button, printing the participating and protecting registry, and generating a pdf file as a result;

(6) Calling the human resource system service to feed back the pdf file to the human resource system;

(7) And (5) finishing storing the participating and protecting data.

S340 popup shielding: and capturing various popup windows in the system execution process and influencing the flow execution.

In the process of executing the flow, the robot collects various reminding, prompting, alarming, confirming and informing information, recognizes and analyzes characters by using an OCR technology, stores the characters into a knowledge base managed by the controller, recognizes and classifies the information by adopting a machine learning algorithm, analyzes out which is useful information, and can feed back the useful information to a user by the robot. For example: prompts that may occur in logging into the social security website to the transacting augmentation include: modifying a password confirmation frame (yes/no), subsidy pickup reminding, website upgrading notification, last business handling result notification, confirmation frame (yes/no) of whether a person is transacted for the age close to retirement, and the like, wherein the feasible occurrence of the participating business is not identical every time, and as the system is upgraded, new information possibly occurs, the first occurrence of the new information is processed by adopting a default processing rule, the robot receives the information and sends the information to the background, and a background machine learning algorithm performs analysis (manual intervention analysis if necessary) after forming a processing rule for the information, and the next time the new information is processed according to the same processing rule. Through machine learning, the correct processing mode of each type of popup window can be automatically identified, shielding processing is carried out by adopting a popup window shielding component which is completely irrelevant to a business process, and processing branches which are relevant to the business and cannot be communicated according to specific scenes are processed. The implementation principle is shown in fig. 8: the invention mainly completes the identification and processing of popup window contents by three background engines: the flow execution engine is responsible for collecting popup content, the knowledge base recognition engine is responsible for matching execution schemes, and the machine learning engine is responsible for formulating the execution schemes and enriching the knowledge base, and manual intervention is adopted if necessary. By the processing method, the RPA robot becomes more intelligent, various emergency situations can be processed correctly, and the trend of various flows can be processed only by few flow branches. The processing process of the popup window comprises the following steps:

(1) A flow execution engine in the executor captures the popup;

(2) Judging the type of the popup window, and closing the popup window if the popup window belongs to the condition that the flow is not influenced;

(3) If the influence exists, acquiring popup window content;

(4) The popup window content is sent to the controller and is processed by a knowledge base analysis engine of the controller; the machine learning engine analyzes the popup content characteristics;

(5) Matching the solution according to the characteristic information;

(6) If no solution is found, processing by a machine learning engine of the controller;

(7) The machine learning engine finds out an optimal execution scheme according to the content characteristics and various execution rules built in advance, and feeds the execution scheme back to the knowledge base analysis engine;

(8) The knowledge base analysis engine stores the content characteristics and the processing results into a knowledge base and sends the processing results to the flow execution engine;

(9) The flow execution engine judges according to the result: if the judgment result is 'continue operation', closing the popup window and continuing the next step;

(10) If the judging result is 'stop executing', closing the closing window and stopping the flow operation.

For example, when transacting a parametrics, if the date of birth is close to 60 years, there is a popup prompt, "about to arrive at retirement age, not the parametrics still to be transacted? Yes/no if the choice is yes, proceed, and if the choice is no, operation is terminated.

Type of popup window:

(1) Notification class popup window: for presenting certain notification notices, has no effect on business process operation at all, but obscures later element recognition if the window is not closed. Such windows are identified by their window pop-up closing mode, typically the window pop-up closing mode 'X', 'ok' button, and 'close' button. Intelligent identification is needed;

(2) Selecting a bullet-like window: the common window comprises a confirm selection box, wherein 'yes/no', 'determine/cancel' and the like can be selected, and corresponding prompt contents are arranged on the window. It is necessary to intelligently identify which button to click depending on the content;

(3) Warning type popup window: the common method comprises an error reporting frame, a warning frame and a prompt frame which are realized by adopting an alert mode. Typically there is only one "ok" button with corresponding prompts on the window. Whether clicking can be continued after determining, whether intelligent identification is continued according to content is needed.

S350, exception handling: and processing various abnormal information generated by summarizing the execution process, and influencing the flow execution. When the current executing process is abnormal, the current executing process automatically jumps to the abnormal processing module, and can be directly transferred to the abnormal module according to certain judging conditions. And after the exception handling is completed, automatically jumping to a flow ending module. The exception handling means that the flow caused by all unexpected emergencies cannot be continued, for example, the network suddenly breaks, so that errors are reported when service call occurs, and the exception handling is required, and many other situations, such as memory overflow, interface call failure, database crash, web page crash and the like, belong to the exception handling.

The S360 flow ends: closing the browser, feeding back the execution state of the controller, and feeding back the execution result to the resource system.

Preferably, the invention uses the Java CV technology to realize the screen recording of the robot operation, and uses a multithreaded subsection uploading mode to upload the screen recording video and the operation date to the server, so that the operation and maintenance personnel can monitor the running state of the flow and cause of the problem.

Finally, it should be noted that: the above embodiments are only for illustrating the technical aspects of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the above embodiments, it should be understood by those of ordinary skill in the art that: modifications and equivalents may be made to the specific embodiments of the invention without departing from the spirit and scope of the invention, which is intended to be covered by the claims.

Claims (10)

1. An RPA robot in communication with a local system, the RPA robot comprising:

the design device comprises: the method is used for compiling rpa flow according to the business, and storing the flow as a flow file after compiling is completed; carrying out parameter maintenance on the flow, executing the flow and carrying out log tracking;

an actuator: the automatic transaction is used for driving the business; providing an interface for calling a local system and acquiring flow data through the interface; feeding back an execution result to the local system;

and (3) a controller: the method comprises the steps of receiving a flow file and performing scheduling execution on the flow; creating a flow task queue, distributing the task queue to an executor, and synchronizing a flow file to the executor; and (3) identifying and analyzing the popup characters by adopting an OCR technology, storing the popup characters into a knowledge base managed by the controller, and identifying and classifying the popup characters by adopting a machine learning algorithm.

2. The RPA robot of claim 1, wherein the robot abstracts and models the flow, encapsulates common parts of the flow into components for supporting the robot to interface with a local system, and obtains flow data; the assembly comprises: the code logic component is used for packaging the general service logic processing code blocks; service operation component: for encapsulating a generic business operation code component; a predefined function component: for encapsulating common functions.

3. The RPA robot of claim 1, wherein the recognition and analysis of the popup text using OCR technology, the storage in a knowledge base managed by the controller, and the recognition and classification of the popup text using a machine learning algorithm, specifically comprises:

the controller receives popup window content sent by the executor;

the controller adopts OCR technology to recognize and analyze the popup text, adopts machine learning algorithm to analyze popup content characteristic information, and matches the solution according to the characteristic information; if the feature information is not matched with the corresponding solution, the controller adopts a machine learning algorithm to match the optimal execution scheme according to the popup window content feature information and a preset execution rule, and feeds back the execution scheme to a controller knowledge base.

4. The RPA robot implementation method is characterized by comprising the following steps:

the robot designer operates the rpa procedure according to the business configuration, store the operation rpa procedure to the robot controller;

the controller controls and manages rpa flow and the robot executor, receives a flow execution request sent by the local system, controls the executor to execute according to the requested flow information, and receives feedback information of the executor;

the robot executes rpa flow, acquires the latest flow from the controller, invokes a local system interface to acquire flow data, automatically transacts business, and feeds back the transacting result to the local system.

5. The RPA robot implementation method of claim 4, wherein the robot operating RPA flow according to the service configuration specifically comprises:

newly building a business process in a robot designer, and designating a process file storage path;

according to the manual business handling process, the robot designer makes a handling process diagram, and sets the input participation parameter information of each node in the handling process;

setting an execution step in each node, if a flow supported by a general component exists, directly dragging the general component into the flow, and setting a handle in a website corresponding to each operation in the general component; otherwise, setting business logic and identifying the handle of the website content.

6. The RPA robot implementation method of claim 4, wherein the controller controlling and managing RPA processes and the robot executor, the controller receiving a process execution request sent by the local system, controlling the executor to execute according to the requested process information, and receiving feedback information of the executor specifically includes:

the method comprises the steps that a robot controller receives a scheduling command and parameters transmitted by a local system, verifies flow information to be executed, and establishes a flow task;

the robot controller searches an executor task queue meeting execution conditions and hangs the flow task to the corresponding queue;

the robot controller monitors the execution condition of the task and collects abnormal information in the running process.

7. The RPA robot implementation method of claim 6, wherein the first priority of the task queue of the actuator is an actuator with an empty task queue and the second priority is an actuator with a shortest task queue; the task queue is set to first in first out.

8. The RPA robot implementation method of claim 4, wherein the robot executing RPA procedure, obtaining a latest procedure from the controller and calling a local system interface to obtain procedure data, automatically transacting business, and feeding back a transacting result to the local system specifically comprises:

the robot executor calls the controller service, acquires a flow task to be executed from the current execution queue, and downloads the latest version of flow definition information to the local;

the robot executor calls the local system to acquire the flow execution parameters, and gradually executes the flow tasks according to the preset flow.

9. The RPA robot implementation method of claim 4, wherein the executor is further configured to capture various popups when executing the RPA procedure, specifically:

the executor captures the popup, judges whether the popup type has influence on the flow, and acquires the popup content if the popup type has influence on the flow; if not, closing the popup window;

the executor sends the popup window content to the controller, the controller adopts a knowledge base and a machine learning algorithm to intelligently judge a processing algorithm, if the judgment result is 'continue execution', the popup window is closed, and the next operation is continued to be executed; if the judgment result is 'stop execution', closing the popup window stop flow.

10. The RPA robot implementation method of claim 9, wherein the popup window type comprises: a notification type popup, a selection type popup and a warning type popup, wherein if the popup is the notification type popup, no influence on the flow is judged; if the popup type is the selected popup type or the warning popup type, the popup type is judged to have influence on the flow, and the executor sends popup content to the controller to obtain a processing result.

Images