Abstract: Based on the concept of big data and artificial intelligence of Huaneng Group Company, combined with the actual work needs of Dongfang Power Plant, financial work must be deeply digitalized and intelligentized. As the digital transformation and upgrading of finance accelerates, the level of informatization continues to improve, and internal information The continuous deepening of interconnection has put forward higher requirements for work efficiency and economic benefits, and the demand for industry and financial integration continues to increase. Major group companies have established financial shared service centers to strengthen centralized management and control, resulting in a large number of centralized economic businesses. It has become possible to apply RPA robots to actual financial work. For highly repetitive, time-consuming, and inefficient transactional work, replacing manual processing with robots will greatly free up human resources and lead to a change in the functions of financial personnel.
Dongfang Power Plant used the material settlement and payment process as a pilot project to carry out practical exploration of RPA robots. Specifically, it started from the five links of invoice review, three-order matching, invoice pre-production, invoice posting, and withdrawal of payment orders to solve the problem of clear rules, high repeatability, and heavy workload. The business is completed by robots. This is a major practical exploration by Dongfang Power Plant to promote the digitalization and intelligence of Hainan Company's financial work.
Keywords: financial robot; automation; practical exploration
01 Preface
1.1 Practical background and significance
Huaneng Dongfang Power Plant is affiliated to the Hainan Branch of China Huaneng Group. It is a comprehensive energy power enterprise focusing on power generation and is responsible for about 20% of the power generation tasks in Hainan Province.
Enterprises use SAP to achieve integrated business and financial management. Due to the large volume of material procurement business, a large amount of human resources are required to invest in purchase order invoice review, invoice, order, contract audit matching, invoice pre-recording, and creation of payment applications. As business volume gradually increases, labor costs and management costs are getting higher and higher.
The significance of introducing financial robots is mainly to consider that such work can be simulated by robots, which can judge and execute the set processes on their own, thereby improving the efficiency and accuracy of financial work and reducing the company's labor and management costs.
RPA is not only a powerful tool for enterprises to reduce costs and increase efficiency, but also plays an important guarantee and promotion role in optimizing business processes, compliance, security, auditing and confidentiality. The benefits it brings are extensive and long-term.
1.2 Innovation points and key problem solving
1. Possible innovations
Huaneng Group currently has few practical cases of financial robots. The units that carry out this business basically optimize and improve on the basis of ready-made templates. However, the material financial process robot of Dongfang Power Plant is completely an exploration and practice from scratch. Regardless of whether it is a group, regional company or grass-roots unit, it is achieved based on seeking truth from facts, scientific demonstration and small steps of trial and error. Judging from the results, the original goal has been achieved. It is indeed a major innovative practice attempt that will continue to expand its applications in different businesses and scenarios.
2. Problems that need to be solved
For this practice, what needs to be focused on is which computer language to imitate manual operations according to established processes, how to identify some special scenarios, establish a certain error-tolerance and correction mechanism, and the interaction between humans and machines. The problem.
For the processing of various non-standard businesses, existing processes need to be continuously modified and optimized so that they can cover most businesses, or even achieve full coverage. This also requires a process of accumulation.
02 Basic connotation of robotic process automation
RPA (Robotic Process Automation), also known as Robotic Process Automation, also known as "digital virtual employees", replaces manual processing of many highly repetitive and logical transactional tasks.
The RPA used in this practice is a set of software tools that can realize uninterrupted automated computing, data storage and business operations around the clock, and automate business processes such as finance, taxation, human resources, and supply chain management. The establishment of RPA does not require changes to the original system functions, and the process can be quickly established in a short time and generate benefits.
03 Implementation ideas and design process
From many processes, after many rounds of exploration, the material and financial check and payment businesses were selected as the entry point to carry out machine automation implementation work. The Python data programming language was used to automate human labor by executing repetitive and judgmental instructions. , process-oriented, focusing on automating the five processes of invoice review, three-order matching, invoice pre-production, invoice posting, and payment orders.
3.1 Invoice review
The material department initiates the process and issues a list of invoice information that needs to be inspected. At the same time, the financial department exports a detailed list of uncertified invoices from the tax system. The robot obtains the necessary information for invoice inspection (for example: invoice number, invoicing date, invoice issuance date) by comparing the two tables. Amount, etc.), query the specific content of the corresponding invoice on the National Value-Added Tax Inspection Platform of the State Administration of Taxation, and store the inspection results in an Excel table and transmit them to relevant business personnel.
3.2 Three orders matching
3.2.1 The keyword of the three-order matching link is the purchase order number. The material procurement department needs to require the supplier to note the corresponding purchase order number in the remarks column when issuing an invoice.
3.2.2 The robot automatically logs in to the SAP system based on the purchase order number in the remark of the invoice information, and queries the corresponding procurement information query report based on the purchase order number.
3.2.3 The robot automatically logs in to the legal system based on the contract number in the procurement information query report, downloads the corresponding procurement contract based on the contract number, and parses out the part of the procurement information specified in the procurement contract, and temporarily stores it in the memory.
3.2.4 The robot performs data comparison based on invoice information, purchase orders, and contract information, and outputs the comparison results to an Excel table. The three order verification information includes: purchase order number, material description, specifications, unit price, total price, and quantity.
Fill in the invoice verification results into the form
Matching results are stored in logs and a classification result table is generated.
Secondary classification of classification results and merger number marking
Invoice posting after automatic adjustment of non-special accounts
3.3 Invoice pre-production
3.3.1 The robot performs preliminary pre-classification based on the item number in the purchase order, sends the classified and unclassified results to the material business personnel for manual secondary classification, and manually determines whether the same supplier and the same cost need to be processed Merge order tags to reduce the number of documents for approval; return the results to the robot after classification is completed.
3.3.2 The robot receives the information table containing the classification results and enters it into the SAP system in sequence according to the specified rules.
3.4 Invoice posting
3.4.1 The robot reads the invoice pre-production result, including information such as the invoice number generated by the pre-production system.
3.4.2 The robot uses the SAP public account to log in and post the invoice.
3.4.3 Automatically check the invoice information and balance information, directly post the orders without balance, automatically adjust the posting for orders with non-special expense difference within 2 yuan, and record the difference and adjustment information into the log form and Feedback to business personnel through OA.
3.4.4 When the process is running, the posting amount of each expense in this batch and monthly summary will be counted simultaneously, so that business personnel can control the use of funds for each expense.
3.5 Payment order
3.5.1 The robot generates a demand intermediate table based on the posting log information and automatically logs in to the SAP system.
3.5.2 Automatically select the supplier, fill in the payment amount, check the bank payment information, fill in the fund budget account, PS project number and other information.
3.5.3 Automatically submit payment orders for documents with an amount of less than 100,000 yuan, intercept documents with an amount of more than 100,000 yuan and mark them as excess, prompt the business staff to upload attachments, and finally form a bill of lading log and feed it back to the business staff.
Automatically check bank payment information and fill in the fund budget account and PS project number
04 Estimation and solutions of possible problems encountered
1. During the development process, fault-tolerant processing is set up for errors that have occurred or may occur to ensure that correct documents are approved to the greatest extent. During the running process, if an error is found, the process will skip the error message and proceed to the next task until all the preset target tasks of the process are completed.
2. For problems that cannot be solved through the fault tolerance mechanism, the process settings will notify the relevant personnel of the operation results by email, mark them in red in the operation log and note the cause of the error, and feedback the operation status of the process to the business personnel as soon as possible. Manually verify the cause of the error, correct it, and execute it when the next process task is executed.
05 Achievements achieved by financial robots
Since the financial robot was put into operation in June 2021, the accuracy of automatically running business recognition has reached 95%. The biggest highlight of the financial robot is to integrate the complete material procurement and payment process between materials and finance, realizing invoice review, three orders Automate the matching, invoice pre-production, invoice posting, and payment order processing steps to achieve higher operational efficiency, save time and release human resources.
Taking a single batch of 10 material purchase orders as an example, the entire manual process requires 100 minutes (including 40 minutes of manual review, 30 minutes of cost determination and pre-production, 10 minutes of invoice entry, and 20 minutes of submission of payment orders). The robot only takes 30 minutes to complete the process accurately. Completed instantly, work efficiency increased three times. Calculated based on the estimated 2,400 material purchase orders for the whole year, excluding data migration and 5% manual processing time for the entire business process, the material purchase and payment business can be completed in 120 hours per year.
In terms of job staffing, as long as the internal control checks and balances are not violated, the material department and the financial forecasting department each have one operator to achieve the normal operation of the financial robot business.
RPA's powerful business processing capabilities can effectively face the sudden increase in workload caused by the continued growth of business in the future. It will also make the adjustment and optimization of existing business smoother, and can calmly face short-term workload bottlenecks. .
This successful practice has accumulated certain experience and lessons for the group company and each system unit in deploying and launching RPA. These business processes and experiences can be quickly copied to each unit, greatly shortening the time and process of launching RPA. Accelerate the business automation and intelligence process of the entire group.
06 Implementation experience summary
For many scenarios with clear rules, high repeatability, and heavy workload, robot interaction language can be used to achieve automation and intelligence of target tasks. The implementation experience is summarized as follows:
1. Build system data sharing. There are many systems running inside and outside the enterprise. Digital links and sharing between each system and business sector need to be built. Robots can switch between different software and platforms according to manual settings. This time, the financial robot realized the link sharing and mutual verification of data from the State Administration of Taxation’s national value-added tax inspection platform, SAP system and legal system.
2. Build automation of various processes. Due to the crossover of business functions between various business departments within the enterprise, a business process is transferred between departments, resulting in a time lag in the handover link, which has a great impact on the smooth flow of business. This time, the financial robot has changed the intersection of material and financial business. A series of tasks are automatically run by the robot according to the set process.
3. Set up real-time human-computer interaction, the original front-end business is not standardized, and error information is fed back in real time during the robot recognition process. Whether the operation is successful or not, the robot will send the results of each process to the corresponding operator in the form of OA email, and the operator will make corrections as soon as possible according to the prompts until the process ends normally.
07 Development Outlook
Ernst & Young's RPA survey report on Fortune 500 companies in Greater China pointed out that 78% of the companies surveyed have started the RPA robot process, and nearly 70% of the companies expressed their hope to expand the application scale of RPA robots.
In the future, financial robots will be combined with artificial intelligence to apply to a wider range of fields. They will be combined with big data to perform model predictions and strategic adjustments to form larger modules. Through simple "drag and drop", simple standardized operations can be realized more quickly. process.
With the rapid development of AI technology, AI+RPA will serve as the technical basis for enterprise business process automation. Enterprises can fully use AI+RPA technology to realize automation and intelligence in finance, procurement, customer service and other scenarios, thereby moving towards a higher overall enterprise level. Advance towards digitalization, automation, and intelligence to obtain greater strategic benefits.
Recommended unit: Huaneng Hainan Power Generation Co., Ltd. Dongfang Power Plant
Authors of this article: Chen Mingchun, Zeng Hui
If there is any infringement, please contact us to delete it.