JPH04338830A - Fuzzy sensor controller - Google Patents

Abstract

PURPOSE:To check the data stored in a program memory holding a membership function and a fuzzy rule without deteriorating the operating speed of a fuzzy sensor controller. CONSTITUTION:A knowledge data deciding means 15 reads the knowledge data out of a program memory 7 and checks them during the waiting for the input given from an input interface 2 to which a signal is applied from a sensor and the waiting for the fuzzy inference of a fuzzy inference device. Thus the abnormality of the knowledge data is decided.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a fuzzy sensor controller that performs fuzzy inference depending on the output of a sensor and provides various outputs.

0002

2. Description of the Related Art Fuzzy sensor controllers are used which perform fuzzy inference using input signals from various sensors and provide control signals. Fuzzy sensor controller 1
As shown in FIG. 4, signals from sensors that provide on/off inputs and analog inputs for a plurality of sensors are provided to a signal processing unit 3 via an input interface (I/F) 2. The signal processing section 3 is realized by a microcomputer, and an internal memory 3a is provided inside the signal processing section 3. An operation mode setting section 4 for setting a program mode and an operation mode is connected to the signal processing section 3. Further, a higher-level computer 6, such as a personal computer, is connected via the communication interface 5. Signal processing section 3
A program memory 7 is connected as an external memory for storing membership functions, fuzzy rules, etc., and a fuzzy inference device 8 is connected to perform fuzzy inference based on the data in this memory. The fuzzy inference device 8 stores a memory 8a for storing membership functions and fuzzy rules (hereinafter referred to as knowledge data). The fuzzy reasoner 8 is a digital fuzzy element for performing fuzzy inference according to knowledge data and fuzzy rules. Further, on/off control output and analog control output are output to the signal processing section 3 via the output interface 9.

In such a conventional fuzzy sensor controller, data of membership functions of the antecedent part and the consequent part and fuzzy rules are given from the host computer 6 in the program mode. The signal processing section 3 once writes this data into the program memory 7 and ends the program mode. When the operation mode is reached, the membership function and fuzzy rule data held in the program memory 7 are transferred to the memory 8a of the fuzzy inference device 8, and then input data from the sensor obtained via the input interface 2 is read. , performs fuzzy inference based on the data and outputs a control signal from the output interface 9.

0004

[Problem to be Solved by the Invention] However, in such a conventional fuzzy sensor controller, knowledge data is transferred only once at the start of operation, so if the knowledge data in the program memory changes for some reason, the knowledge data is transferred only once at the start of operation. , the fault is that the incorrect knowledge data is then transferred to the fuzzy inference machine and fuzzy inference is executed.

The present invention has been made in view of the problems of the conventional fuzzy sensor controller, and includes:
The technical problem is to be able to check knowledge data in a program memory without reducing processing speed.

[0006]

[Means for Solving the Problems] The present invention is a fuzzy sensor controller that performs fuzzy inference based on the output of a sensor and outputs an inference result. A program memory that stores knowledge data including functions and fuzzy rules, a fuzzy inference device to which knowledge data in the program memory is transferred at the start of operation and performs fuzzy inference based on given feature data, and a sensor input from an input section. a signal processing means that takes in the feature data and provides it to the fuzzy inference machine at a predetermined timing, and outputs the inference result obtained from the fuzzy inference machine; and an output unit that outputs the inference result signal obtained from the signal processing means to the outside. and knowledge data determining means for determining an abnormality in the knowledge data in the program memory, at least either when waiting for input from the input unit to the signal processing means or during fuzzy inference by the fuzzy inference device. That is.

[0007]

[Operation] According to the present invention having such characteristics, membership functions and fuzzy rules are stored in the program memory in advance before starting fuzzy inference, and the knowledge data is used for fuzzy inference before starting fuzzy inference. Give it to the vessel. During fuzzy inference, the signal from the sensor obtained through the input section is periodically applied to the fuzzy inference section as feature data by the signal transmission means, and the inference result is received and output. The knowledge data determining means determines whether there is an abnormality in the knowledge data held in the program memory at least when waiting for sensor input from the input unit or when waiting for fuzzy inference results from the fuzzy inference device. ing.

[0008]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is a block diagram showing the overall configuration of a fuzzy sensor controller 11 according to an embodiment of the present invention, and the same parts as in the conventional example described above are given the same reference numerals and detailed explanations will be omitted. In this embodiment as well, inputs from on-off type sensors and analog sensors are provided to the signal processing section 12 via the input interface 2, which is an input section. The signal processing unit 12 is realized by a microcomputer, and in addition to the internal memory 13, as will be described later, signals from the input interface 2 at every predetermined period are given to the fuzzy inference unit 8 as feature data, and a signal for outputting an inference result is provided. A knowledge data determining means 15 is provided for determining the knowledge data in the program memory 7 while the processing means 14 and the signal are on standby. Further, an output interface 9 is connected to the signal processing section 12 as an output section, and outputs an on/off signal or an analog control signal to the outside. The rest of the configuration is the same as that of the conventional example described above, and the signal processing section 3 is connected to an operation mode setting section 4 for setting an operation mode and a program memory 7 for holding knowledge data. The program memory 7 is a nonvolatile memory or a memory backed up by a battery, and is capable of retaining knowledge data even when the power is cut off.

Next, the operation of this embodiment will be explained using the flowcharts shown in FIGS. 2 and 3. When the operation starts, initialization processing is first performed in step 21 of FIG. 2, and the process proceeds to step 22, where it is checked whether the program mode is set. If the program mode is set by the operation mode setting unit 4, a program write command is awaited from the host computer 6 in step 23. If the program write command is obtained, the process proceeds to step 24 to check whether there is a communication error, and if there is no communication error, the membership function (MSF) of the antecedent part of the fuzzy inference given following the write command and The membership function data of the consequent part, the fuzzy rule, and in addition to this, the total value of these data (so-called sum value) are written into the program memory 7. The process then proceeds to step 26 to generate a response and returns to step 22. If there is a communication error in step 24, communication error handling is performed (step 27) and the process ends.

After the knowledge data is thus written in the program memory 7 in the program mode, the operation mode setting section 4 sets the operation mode to the run mode. Then, proceed from step 22 to step 31 in Figure 3,
Check to see if you have just finished changing the mode. If the mode has just been changed, the knowledge data held in the program memory 7, that is, the membership function data of the antecedent part and the consequent part and the fuzzy rules, are transferred to the memory 8a in the fuzzy inference device 8 in step 32 ( Step 32). The process then proceeds to step 33, reads the input from the sensor, and waits for the reading to be completed. In many usage conditions, sensor input for a certain period of time is often required. For example, when determining the shape of an article flowing on a belt conveyor, it is necessary to continuously read sensor input while the sensor is detecting the article. In this way, reading input from the sensor requires a certain amount of time. Therefore, until the reading is completed, the process proceeds from step 34 to step 35 to read out the knowledge data in the program memory 7, that is, the membership function data, rule data, and sum value.
A sum check is performed to see if the sum of these data is equal to the sum value. Then, the process advances to step 36 to check whether there is an error, and if there is no error, the process returns to step 22 and repeats the same process.

[0011] When reading of the sensor input is completed in step 34, the routine proceeds to routine 37, where feature data is calculated and generated from the read sensor input. Then, the process proceeds to step 38 and the feature data is transferred to the fuzzy inference unit 8. Then, proceed to steps 39 and 40 and use the fuzzy inference machine 8.
Execute the inference by and wait for the inference to complete. Inference by the fuzzy inference device 8 requires a certain amount of time determined by membership functions, the number of rules, etc., for example, several milliseconds of inference time. If the inference has not been completed, the process proceeds to step 41, where the knowledge data is read out from the program memory 7 again and sum-checked, as in step 35. Then, in step 41, the presence or absence of an error is checked, and if there is no error, completion of the inference is awaited. When the inference is completed, an on/off signal or an analog control signal is output from the output interface 9, and the process is completed. If an error is detected in step 36 or 42,
Proceeding to step 44, sum check error processing is performed. This error processing involves displaying an error, outputting an error, and forcibly interrupting control.

[0012] In this way, in this embodiment, since the sum check of the knowledge data is performed between the time of reading the sensor input and the time when the inference of the fuzzy inference is completed, the knowledge data is checked for some reason without reducing the processing speed. You can check changes in knowledge data when changes occur. If there is a large number of rules and membership function data, and the sum check does not finish while waiting for sensor input or during fuzzy inference, the sum check is executed over a plurality of fuzzy control cycles. In this way, waiting time can be used effectively. Furthermore, reliability can be further improved by performing a sum check during initial processing after power is turned on. Here, the signal processing section 12 performs step 33.
, 34 and 37 to 39, 40, and 43 constitute a signal processing means 14 that supplies the sensor output obtained from the input interface 2 as feature data to the fuzzy inference unit 8, executes fuzzy inference, and outputs the inference result, Steps 35, 36 and steps 41, 42 achieve the function of the knowledge data determining means 15 which reads the knowledge data from the program memory and checks its contents during these processes.

[0013]

Effects of the Invention As described in detail above, according to the present invention, abnormalities in the knowledge data in the program memory are determined when waiting for input from a sensor and during fuzzy inference, thereby increasing the data processing speed. Knowledge data can be checked without degrading it. Furthermore, since the knowledge data is constantly checked, abnormalities can be easily discovered and highly reliable fuzzy inference can be realized.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the configuration of a fuzzy sensor controller according to an embodiment of the present invention.

[Figure 2] Flowchart showing the operation of this embodiment (Part 1)
).

[Figure 3] Flowchart showing the operation of this embodiment (Part 2)
).

FIG. 4 is a block diagram showing the configuration of a conventional fuzzy sensor controller.

[Explanation of symbols]

1, 11 Fuzzy sensor controller 2 Input interface 4 Operation mode setting unit 5 Communication interface 7 Program memory 8 Fuzzy inference unit 8a Memory 9 Output interface 12 Signal processing unit 13 Internal memory 14 Signal processing means 15 Knowledge data discrimination means

Claims (1)

[Claims] 1. A fuzzy sensor controller that performs fuzzy inference based on the output of a sensor and outputs an inference result, the controller comprising an input section to which a signal from the sensor is applied, and knowledge including a membership function and a fuzzy rule. a program memory that stores data; a fuzzy inference device to which knowledge data in the program memory is transferred at the start of operation and performs fuzzy inference based on given feature data; a signal processing means for providing data to the fuzzy inference device at a predetermined timing and outputting an inference result obtained from the fuzzy inference device; an output unit for outputting an inference result signal obtained from the signal processing means to the outside; knowledge data determining means for determining an abnormality in the knowledge data in the program memory at least either when waiting for an input from an input unit to the signal processing means or during fuzzy inference by the fuzzy inference device; Features a fuzzy sensor controller.