Data processing: artificial intelligence – Neural network – Learning method
Reexamination Certificate
2000-04-05
2003-07-08
Davis, George B. (Department: 2121)
Data processing: artificial intelligence
Neural network
Learning method
C706S021000
Reexamination Certificate
active
06591255
ABSTRACT:
TECHNICAL FIELD
This invention relates to computer-based forecasting systems and, more particularly, relates to a neural-network forecasting system that automatically extracts input data over the Internet, corrects the input data for errors, forecasts output values based on the corrected input data, and returns the forecast over the Internet. In particular, the system may be used to automatically extract weather forecast input data and return electricity demand output values.
BACKGROUND OF THE INVENTION
Many occupations can benefit from the availability of reliable and accurate forecasts of various types. For example, electric utilities can use electricity demand forecasts to schedule the operation of electric power plants, commodity traders can use commodity price forecasts to buy and sell commodity future contracts, municipal governments can use weather forecasts to schedule snow plows, hospitals can use patient forecasts to schedule medical personnel, manufactures can use product demand forecasts to schedule the purchase of raw materials, and so forth.
However, reliable and accurate forecasts can be difficult and expensive to obtain. The timeliness of the forecast is often a critical element. Once new input data becomes available, processing that data to obtain an updated forecast of important values in a timely manner may be imperative. For certain critical applications, such as weather and electricity demand forecasting, very expensive and sophisticated forecasting systems have been developed. For example, some of the most powerful, expensive, and sophisticated computers in the world are dedicated to the task of weather forecasting. In some cases, more than twenty hours of computing are required to produce a five-day weather forecast for the continental United States. In addition, sophisticated neural-network parallel processing hardware and software has been developed to forecast electricity demand based on weather forecasts and other information. Obviously, every entity that could benefit from reliable and accurate forecasts cannot afford these types of expensive forecasting systems.
In recent years, utility regulators in the United States have decreased price regulation and increased the numbers and types of electricity and gas services that are open for competitive bidding. The basic idea is to allow the market forces of supply and demand determine the cost to the end-user for these utility services. A similar type of government-sponsored price deregulation reshaped the airline, trucking, and telephone industries in previous decades. A practical result of this trend toward deregulation is to place the purchasing decisions for basic utility services out of the hands of regulated utilities and into the hands of competitive suppliers and the end-users that they serve.
This has proliferated the number and types of utility service options available for end users, such as industries, farms, co-operatives, municipalities, and the like. To plan their operations and make informed purchasing decisions, individual and groups of utility suppliers and end users have an increased need for reliable and accurate forecasts of their own utility needs. Increased competition in electricity procurement thus provides both suppliers and purchasers of electricity with increased opportunities to use electricity forecasts in their day-to-day operation.
Similar competitive forces are also increasing the need for forecasts by natural gas, telephone, computer service, and other utilities and their customers. Indeed, the availability or reliable and accurate forecasts of critical information can improve the efficiency and profitability of a very wide variety of activities. Although the preceding discussion focuses on the impact of deregulation on utility purchasing, entities involved in a wide range of activities have a similar need for reliable and accurate forecasts. For example, agricultural production from seed to shelf involves the use of forecasts to predict supply and demand for all sorts commodities. The manufacture of goods also involves the use of forecasts to predict the supply and demand for all sorts goods. Real estate development, health care, banking, personal investing, and many other occupations also use forecasts in some aspect of their operations.
In many cases, access to reliable and accurate forecasts may determine the profitability and ultimate viability of individual participants in these markets. The unfortunate truth is that the playing field may be far from level. Large entities with expensive and sophisticated forecasting systems may often have the important advantage of timely, reliable and accurate forecasts to guide their decisions. Smaller entities, who rely on older and less accurate information, may not be able to compete effectively simply because they cannot afford to purchase or develop sophisticated forecasting systems.
Thus, there is a general need in the art for a cost-effective mechanism for providing interested parties with the benefits of sophisticated forecasting systems. There is a further need for improved systems for generating and delivering timely, reliable and accurate forecasts for a wide range of applications.
SUMMARY OF THE INVENTION
The present invention meets the needs described above in a business process and computer system known as the “Rapid Learner Client Service” (RLCS) system. This system allows a large number of end-users to obtain the benefits of a sophisticated neural-network forecasting system. Rather than purchasing or developing a forecasting system of their own, RLCS clients subscribe to a forecasting service performed by forecasting equipment located at a remote site. This allows a single highly sophisticated forecasting system to meet the forecasting needs of a large number of subscribers.
This forecasting service is performed by an RLCS server that periodically and automatically accesses the subscriber's computer to obtain a fresh set of input data. This input data is then downloaded to the RLCS server, where it is checked and corrected for errors by imputing values for missing or deviant input values. Alternatively, the subscriber's computer may contact the RLCS server to initiate the process. The error-corrected input data is then used to compute a forecast of output values, which are downloaded to the client's computer. The RLCS server also computes and downloads a set accuracy statistics for the client's review.
In an electricity demand forecasting application, for example, the input data downloaded or received from the client's computer may include weather forecast data and the client's actual electricity demand and weather data for a recent historical period. The RLCS server then uses this input data to compute an electricity demand forecast for a projected period, which is returned to the client's computer. This process is typically repeated hourly or daily, depending on the client's needs. However, the process could be repeated at different intervals and for different applications based on the needs of different clients. It should be understood, therefore, that the invention is generally applicable as a business method and computer-based forecasting system for a wide range of industries, and may be applied to forecasting systems for goods as well as services. Nevertheless, the particular system described in this specification is well suited to providing electricity demand forecasts on an hourly or daily basis based on historical weather and electricity demand and weather forecast data downloaded from the client's computer.
That the invention improves over the drawbacks of conventional forecasting systems and accomplishes the advantages described above will become apparent from the following detailed description of the exemplary embodiments and the appended drawings and claims, as well as the attached Exhibits.
REFERENCES:
patent: 5835902 (1998-11-01), Jannarone
patent: 5910987 (1999-06-01), Ginter et al.
patent: 6327677 (2001-12-01), Garg et al.
Jannarone Robert J.
Tatum John T.
Wilkinson, IV W. Clayton
Davis George B.
Mehrman Michael J.
Mehrman Law Office PC
Netuitive, Inc.
LandOfFree
Automatic data extraction, error correction and forecasting... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Automatic data extraction, error correction and forecasting..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Automatic data extraction, error correction and forecasting... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3050553