Real-time rule based processing system

• US 5,487,134 A
• Filed: 02/23/1995
• Issued: 01/23/1996
• Est. Priority Date: 02/25/1993
• Status: Expired due to Fees

• Current Assignee: Reticular Systems Incorporated
• Original Assignee: Reticular Systems Incorporated
• Inventors: Ballard, Dan
• Primary Examiner(s): Downs, Robert W.
• Application Number:
  US08/393,241
• Time in Patent Office: 334 Days
• Field of Search: 395/51, 395/52, 395/67
• US Class Current: 706/46
• CPC Class Codes: G06F 30/30 Circuit design
• CPC Class Codes: G06N 5/04 Inference or reasoning models
• CPC Class Codes: G06N 5/046 Forward inferencing; Production systems
• CPC Class Codes: G06N 5/047 Pattern matching networks; Rete networks
• CPC Class Codes: Y10S 706/916 Electronic or computer, internal or network, circuit

Abstract

A rule based processing system that is optimized for implementation in hardware. The system includes a set of logic gates that are used to implement a rule network connected to a working memory composed from a set of memory components. When a set of inputs are presented to the logic gates, a set of inferences are formed and the results are stored in the memory components. The inferences that are stored in the working memory are used to determine the state of the system at any given time. These inferences can be output to an external device to report on the status of the system. The rule based production system is designed to be implementable on a high performance semiconductor micro-chip and is thus suitable for use in embedded real-time applications.

Claims

1. An inference circuit which simultaneously forms a plurality of inferences from a plurality of input variables, said inference circuit comprising:

a set of logic gates representative of a rule network comprising a plurality of rule network portions, wherein each said portion comprises one or more of said logic gates connected together, wherein each said portion is capable of forming one of said inferences, and wherein each rule network portion simultaneously forms a one of said inferences;

a set of working memory storage elements representative of a working memory comprising a plurality of working memory groups, wherein each said group comprises one or more of said elements, each said element capable of storing one of said inferences;

a first set of signal lines connecting an output of each of the rule network portions to an input of a respective one of the working memory storage elements so that each inference is simultaneously stored in one of said working memory storage elements; and

a second set of signal lines connecting the output of each working memory storage element from a selected one of said working memory groups to an input of a respective selected group of said rule network portions so that the inferences stored in said selected working memory group are available for subsequent inferences by the selected group of said rule network portions.

2. The inference circuit defined in claim 1, wherein said inference circuit comprises a single semi-conductor chip.

3. The inference circuit defined in claim 1, wherein said set of logic gates are programmable to enable said rule network to be modified during the operation of said inference circuit.

4. The inference circuit defined in claim 1, wherein the set of logic gates and connections in each said rule network portion are selected to perform one or more AND/OR operations associated with rule antecedents.

5. The inference circuit defined in claim 1, wherein each working memory storage element is a delay flip-flop.

6. The inference circuit defined in claim 1, additionally comprising a plurality of input working memory storage elements, each said input working memory storage element capable of storing an input signal.

7. In a system comprising a rule network and a working memory, a method of simultaneously forming a plurality of inferences from a plurality of input variables each inference cycle, said method comprising the steps of:

storing the condition of each of said input variables in a respective working memory storage element of the working memory;

performing a first set of logical operations representative of a portion of the rule network on a portion of the working memory storage elements, each said rule network portion capable of forming an inference;

storing each of the inferences in a respective working memory storage element so that said inferences are available for subsequent logical operations; and

simultaneously with the first set of logical operations, performing a second set of logical operations which receive said stored inferences.

8. The method defined in claim 7, further comprising the step of:

programming the logical operations that are performed on the portion of said memory storage elements to represent the rule network.

9. The method defined in claim 7, wherein the first rule network portion is receivably connected to a set of input memory storage elements.

10. The method defined in claim 7, wherein the working memory storage elements storing each inference from a rule network portion, form part of a group of working memory elements.

11. The method defined in claim 10, wherein a rule network portion comprises one or more logic gates.