JP2008062733A - System and program for assisting railway facility maintenance and inspection - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To materialize labor-saving of inspection record of railway facilities without installing ground facilities for detecting a present position of an inspector. <P>SOLUTION: A predetermined area having a center on a measured present position is variably defined as a facility inspection area based on position measurement accuracy of the present position by a GPS device 8. Inspection facilities is narrowed down based on inspection process from inspection facilities of which installation positions are included in the facility detection area. Then, an inspection facilities relevant to spoken voice input from a microphone 7a is identified by voice identification process based on voice data corresponding to narrowed down inspection facilities. Then, spoken voice input from the microphone 7a is subjected to voice identification process and inspection result data are input to PDA2 and are stored as inspection results of the identified inspection facility. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a railway facility maintenance inspection support system and program.

  Enormous facilities such as civil engineering, track, electric power, signal communication, etc. exist along the railway, and regular maintenance inspections of these facilities are performed by inspectors, ensuring the safety of railway operation. Has been.

  In maintenance inspections, a method has been established in which an inspector writes the inspection results by hand in a notebook while performing the inspection, and returns to the ward to register in the computer. However, during nighttime, rainy weather, and snowy inspections, it is difficult for the inspector to write the test results in the notebook while putting their hands in the plastic bag, or to remember the test results and fill in the notebook later. There are actual situations that have been devised, and labor saving of inspection records is desired.

Although it is a technique related to maintenance inspection of plants with completely different technical fields, Patent Document 1 discloses a system that realizes an inspection record by voice input as a technique for realizing labor saving of the inspection record. In the system of Patent Literature 1, an inspector carries a terminal including a microphone, a display, and a transmission / reception unit, and voice input to the microphone is communicated to the server side, and the inspection result is recorded. In addition, the position of the inspector is detected by sensors installed at various locations in the plant facility and sent to the server side.
JP-A-8-320697

  Unlike a limited facility such as a plant facility as disclosed in Patent Document 1, the railway facility is long, and the railway equipment to be inspected is arranged along the railway line. Therefore, if it is going to apply the technique of patent document 1 simply for the inspection of a railway installation, it is necessary to install the ground equipment called a sensor in every place along a railway line, and to lay the communication line which connects the sensor and a server This is unrealistic due to troubles and costs.

  Also, unlike the plant facilities, the railway line has no roof and exists in the natural environment. For this reason, inspections must be performed while exposed to wind and snow. For example, it is difficult to operate the terminal itself in a cold region in winter such as a snowy region. Therefore, a simple and reliable specification is required.

  The present invention has been made in view of the above-described problems.

The first invention for solving the above problems is:
A microphone (for example, the microphone 7a in FIGS. 1 and 4);
Position acquisition means for acquiring the current position (for example, PDA 2 in FIG. 1 and FIG. 4, GPS receiver 8b; CPU 20 in FIG. 4; step A5 in FIG. 14);
An object information storage means (for example, an association position storage (for example, an installation position 624 in FIG. 5) of each inspection object and dictionary data for identifying the inspection object by a predetermined voice recognition process in association with each other. PDA2 in FIG. 1 and FIG. 4; flash ROM 60 in FIG. 4; inspection equipment DB 620, inspection equipment dictionary data 651);
Based on the positional relationship between the current position acquired by the position acquisition means and the installation position of each inspection object stored in the object information storage means, a narrowing means for narrowing down the next inspection object (for example, FIG. 1). PDA2 in FIG. 4; CPU 20 in FIG. 4; Steps A7 and A9 → A33 in FIG.
Inspection object specifying means (for example, FIG. 1) that specifies the inspection object corresponding to the speech voice input from the microphone by performing voice recognition processing based on dictionary data corresponding to the inspection object narrowed down by the narrowing means. PDA2 in FIG. 4; CPU 20 in FIG. 4; Step A35 in FIG. 14 to Step A45 in FIG. 15);
Test result input means (for example, PDA2 in FIGS. 1 and 4; CPU 20 in FIG. 4) for inputting test result data (for example, test result data 641 in FIG. 7) after performing speech recognition processing on the speech input from the microphone. Step A53) of FIG.
Recording means (for example, PDA 2 in FIGS. 1 and 4; CPU 20 in FIG. 4) that records the inspection result data input by the inspection result input means as the inspection result for the inspection object specified by the inspection object specifying means. Step A53) of FIG.
Is a portable railway facility maintenance inspection support system (for example, the railway facility maintenance inspection support system 1 of FIGS. 1 and 4).

As another invention,
A portable computer (for example, PDA 2 in FIGS. 1 and 4) to which a microphone (for example, microphone 7a in FIGS. 1 and 4) is connected,
The installation position of each inspection object included in the object information associated with the installation position of each inspection object and dictionary data for identifying the inspection object by a predetermined voice recognition process, and the current position Narrowing means to narrow down the next inspection object based on the positional relationship of
An inspection object specifying means for specifying an inspection object corresponding to an utterance voice input from the microphone by performing voice recognition processing based on dictionary data corresponding to the inspection object narrowed down by the narrowing-down means;
A test result input means for performing voice recognition processing on speech speech input from the microphone and inputting test result data;
Recording means for recording the inspection result data input by the inspection result input means as the inspection result for the inspection object specified by the inspection object specifying means,
A program (for example, the equipment maintenance inspection support program 610 in FIG. 4) may be configured.

  According to the first aspect of the invention, the next inspection object is narrowed down based on the positional relationship between the current position and the installation position of each inspection object included in the object information, and the utterance input from the microphone The inspection object corresponding to the voice is identified by performing voice recognition processing based on dictionary data corresponding to the narrowed inspection object. Then, the speech voice input from the microphone is subjected to voice recognition processing, inspection result data is input, and the input inspection data is recorded as the inspection result for the specified inspection object.

  For example, if it is set as the structure which acquires a present position by the autonomous positioning using GPS, it will become unnecessary to arrange ground facilities, such as a sensor for detecting the present position of an inspector. In addition, since the inspection result is recorded based on the uttered voice input from the microphone, the inspector does not need to manually input the inspection result, and labor saving of the inspection record is realized.

  In addition, since voice recognition processing is performed based on dictionary data corresponding to inspection objects narrowed down based on the positional relationship between the installation position of each inspection object and the current position, voice recognition is performed based on all dictionary data. Compared to the case where processing is performed, the processing load is reduced.

Moreover, as 2nd invention, it is the railway equipment maintenance inspection support system of 1st invention,
A remote switch (for example, the remote switch 6 in FIGS. 1 and 4);
A pressing operation detecting means for detecting the pressing operation of the remote switch (for example, PDA2 in FIGS. 1 and 4; CPU 20 in FIG. 4; step A15 in FIG. 14);
Speech voice reception control means (for example, PDA2 in FIGS. 1 and 4; CPU 20 in FIG. 4; step A17 in FIG. 14) that accepts the voice input from the microphone as valid when detected by the pressing operation detection means. When,
With
The inspection object specifying means and the inspection result input means may constitute a railway facility maintenance inspection support system that performs voice recognition processing on the voice received by the utterance voice reception control means.

  According to the second aspect of the present invention, the voice input from the microphone when the remote switch pressing operation is detected is accepted as valid, and the voice recognition process is performed on the voice. Therefore, it is not necessary to always accept the input voice as being valid, and the processing load is reduced. Further, the inspector can instruct the reception of the voice by a simple operation of pressing the remote switch, and a simple but reliable operation is realized.

Moreover, as 3rd invention, it is the railway equipment maintenance inspection support system of 1st or 2nd invention,
Accuracy determination means (for example, PDA2 in FIGS. 1 and 4; CPU 20 in FIG. 4; step A7 in FIG. 14) for determining the accuracy (for example, the DOP value in FIG. 2) acquired by the position acquisition means. In addition,
The narrowing-down means may constitute a railroad equipment maintenance inspection support system that narrows down the next inspection target by varying the narrowing-down conditions according to the accuracy determined by the accuracy determination means.

  According to the third aspect of the invention, the accuracy of the acquired current position is determined, the narrowing condition is varied according to the accuracy, and the next inspection object is narrowed down. If the accuracy of the acquired current position is low, inappropriate inspection objects are narrowed down based on the incorrect current position, and the voice recognition process may not be performed properly.

  However, for example, when the accuracy of the acquired current position is low, the inspection object included in a wide range is extracted from the current position, and when the accuracy of the current position is high, it is included in a narrow range from the current position. By narrowing the narrowing conditions so as to extract the inspection object to be detected, it is possible to realize appropriate narrowing of the inspection object.

Moreover, as 4th invention, it is the railway equipment maintenance inspection support system of any one of 1st-3rd invention,
The narrowing-down means may further constitute a railway facility maintenance inspection support system that narrows down the next inspection object based on the installation position of the inspection object that has been inspected immediately before.

  According to the fourth aspect of the invention, the next inspection object is narrowed down based on the installation position of the inspection object that is the inspection object immediately before. Therefore, for example, by narrowing down so as to extract the inspection object included in a narrow range from the installation position of the inspection object that is the inspection object immediately before, it is possible to realize appropriate narrowing of the inspection object. .

Further, as a fifth invention, the railway facility maintenance inspection support system according to any one of the first to fourth inventions,
Inspection order storage means (for example, PDA 2 in FIGS. 1 and 4; flash ROM 60 in FIG. 4; inspection process DB 630) for storing the inspection order of each inspection object (for example, inspection order 633 in FIG. 6);
The narrowing-down means may further constitute a railroad equipment maintenance inspection support system that narrows down the next inspection object based on the inspection order stored in the inspection order storage means.

  According to the fifth aspect, the next inspection object is narrowed down based on the inspection order of each inspection object. Therefore, for example, by narrowing down so as to extract inspection objects whose inspection order is later than the current inspection order, it is possible to realize appropriate narrowing of inspection objects.

Moreover, as a sixth invention, the railway facility maintenance inspection support system according to any one of the first to fifth inventions,
Inspectable time zone storage means (for example, PDA 2 in FIGS. 1 and 4; flash ROM 60 in FIG. 4; inspection process) for storing the inspectable time zone (eg, inspectable time zone 637 in FIG. 6) of each inspection object. DB630),
The narrowing-down means may further constitute a railroad equipment maintenance inspection support system that narrows down the next inspection object based on the inspectable time zone stored in the inspectable time zone storage means.

  According to the sixth aspect of the invention, the next inspection object is narrowed down based on the inspectable time zone of each inspection object. Therefore, for example, by narrowing down so as to extract the inspection object whose current time is included in the inspectable time zone, it is possible to realize appropriate narrowing down of the inspection object.

Moreover, as a seventh invention, the railway facility maintenance inspection support system according to any one of the first to sixth inventions,
The object information storage means stores a kilometer (for example, kilometer 626 in FIG. 5) as the installation position of the inspection object,
The position acquisition means performs a speech recognition process on the speech voice input from the microphone to acquire a kilometer indicating a current position (for example, PDA2 in FIGS. 1 and 4; CPU 20 in FIG. 4; Step A21) of FIG.
The narrowing down means narrows down the next inspection object based on the positional relationship between the kilometer distance acquired by the kilometer distance acquisition means and the kilometer distance of each inspection object stored in the object information storage means.
A railway equipment maintenance inspection support system may be configured.

  According to the seventh aspect of the present invention, the kilometer indicating the current position is acquired by performing voice recognition processing on the speech sound input from the microphone, and the positional relationship between the kilometer and the kilometer of each inspection object is obtained. Based on this, the next inspection object is narrowed down. Therefore, it is possible to realize appropriate narrowing down of inspection objects by performing narrowing down so as to extract inspection objects having a kilometer closer to the acquired kilometer.

Further, as an eighth invention, a railway facility maintenance inspection support system according to the seventh invention,
Receiving means (for example, the antenna 8a in FIGS. 1 and 4) that receives a predetermined external radio wave that is a radio wave that can identify the current position;
The position acquisition means includes
Current position specifying means (for example, GPS receiver 8b in FIGS. 1 and 4) for specifying the current position based on the external radio wave received by the receiving means;
Kilometer acquisition activation determination means (for example, PDA2 in FIGS. 1 and 4; FIG. 4) for determining whether or not to activate the kilometer acquisition means based on the specific success and / or accuracy by the current position specifying means. CPU 20; step A3) of FIG.
Have
The kilometer acquisition means may constitute a railway facility maintenance inspection support system that functions when it is determined to be activated by the kilometer acquisition activation determination means.

  According to the eighth aspect of the present invention, it is determined whether or not to acquire the kilometer based on the specific success or failure and / or accuracy of the current position based on the received external radio wave. The degree is acquired. Therefore, for example, when it is difficult to specify the current position based on the received external radio wave, or when the accuracy of the current position is low, it is possible to acquire about a kilometer.

Moreover, as a ninth invention, the railway facility maintenance inspection support system according to any one of the first to eighth inventions,
The recording unit corresponds to the inspection target specified by the inspection target specifying unit, with the speech data (for example, the utterance voice data 645 in FIG. 7) targeted by the test result input unit for voice recognition processing. A railway facility maintenance inspection support system having inspection result voice data recording means for recording (for example, PDA 2 in FIGS. 1 and 4; CPU 20 in FIG. 4; step A53 in FIG. 15) may be configured.

  According to the ninth aspect of the invention, speech voice data that is subject to voice recognition processing is recorded in association with the specified inspection object. Therefore, the inspector can rehearse the utterance content for the inspection object later.

Moreover, as a tenth invention, the railway facility maintenance inspection support system according to any one of the first to ninth inventions,
A speaker (for example, the headphones 7b in FIGS. 1 and 4);
Specific success / failure determination means (for example, PDA2 in FIGS. 1 and 4; CPU 20 in FIG. 4; step A47 in FIG. 15) for determining the specific success or failure of the inspection object by voice recognition processing of the inspection object specification means;
With
The inspection object specifying means, when it is determined that the specific success / failure determining means is unsuccessful, prompts re-input of the spoken voice and performs the voice recognition processing again (for example, the PDA2 in FIGS. 1 and 4). 4; step A47 in FIG. 15; No → step A37 in FIG. 14 to step A45 in FIG. 15);
When the number of failure determinations by the specific success / failure determination unit reaches a predetermined number, a reading control unit (for example, PDA 2 in FIGS. 1 and 4; FIG. 15; step A73 in FIG. 15; Yes → steps A79, A81 → A95, A81, steps A103, A105 → A111, A105) in FIG.
Success / failure input means (for example, microphone 7a, remote switch 6 in FIGS. 1 and 4) for sequentially inputting whether or not the inspection object sequentially read out by the reading control means is the next inspection object;
A rail facility maintenance inspection support system that identifies the next inspection object based on the input of the success / failure input means when the number of failure determinations by the specific success / failure determination means has reached a predetermined number of times. Also good.

  According to the tenth aspect, whether or not the test object is specifically determined by the voice recognition process is determined, and if it is determined to be unsuccessful, the speech recognition process is performed again by prompting re-input of the speech voice. In addition, when the number of failure determinations reaches a predetermined number, the sound for sequentially reading each inspection object is output from the speaker, and whether or not the inspection object sequentially read out is the next inspection object is a success or failure. The data are sequentially input by input means. And when the frequency | count of failure determination reaches predetermined number of times, the next test target object is specified based on the input of a success / failure input means.

  Therefore, even when voice recognition of the inspection object cannot be performed properly, the inspection object is read aloud sequentially and prompts the inspector to input “Yes” or “No” each time. Thus, it is possible to reliably specify the inspection object.

  According to the present invention, based on the positional relationship between the current position and the installation position of each inspection object included in the object information, the next inspection object is narrowed down and corresponds to the uttered voice input from the microphone. The inspection object is identified by voice recognition processing based on dictionary data corresponding to the narrowed inspection object. Then, the speech voice input from the microphone is subjected to voice recognition processing, inspection result data is input, and the input inspection data is recorded as the inspection result for the specified inspection object.

  For example, if it is set as the structure which acquires a present position by the autonomous positioning using GPS, it will become unnecessary to arrange ground facilities, such as a sensor for detecting the present position of an inspector. In addition, since the inspection result is recorded based on the uttered voice input from the microphone, the inspector does not need to manually input the inspection result, and labor saving of the inspection record is realized.

  In addition, since voice recognition processing is performed based on dictionary data corresponding to inspection objects narrowed down based on the positional relationship between the installation position of each inspection object and the current position, voice recognition is performed based on all dictionary data. Compared to the case where processing is performed, the processing load is reduced.

1. System Configuration FIG. 1 is a diagram showing a schematic configuration of a railway facility maintenance inspection support system 1 in the present embodiment. The railway facility maintenance inspection support system 1 includes a PDA (Personal Digital Assistants) 2 that is a computer device body, a remote switch 6, a headset microphone 7, and a GPS (Global Positioning System) device 8. .

  The PDA 2 is a portable information terminal for an inspector to input and record data such as inspection results of equipment, the operation key 3, the touch screen 4 in which the touch panel 4a and the display 4b are integrally formed, and wireless communication. And a device 5.

  The remote switch 6 is a button switch to be pressed when the inspector instructs the PDA 2 to accept voice input or makes a judgment / response to the voice guidance output from the PDA 2, and is connected to the PDA 2 by a cable. Yes.

  The headset microphone 7 is configured integrally with a microphone 7a for an inspector to input voice to the PDA 2 and a headphone 7b for outputting voice guidance from the PDA 2, and is connected to the PDA 2 by a cable. ing.

  The GPS device 8 is an antenna 8a for receiving a GPS signal (radio wave) transmitted from a GPS satellite, and a circuit unit for positioning the current position in the earth reference coordinate system based on the GPS signal received by the antenna 8a. The GPS receiver 8b is integrally formed and is connected to the PDA 2 by a cable.

  Here, four GPS satellites are arranged in each of the six orbital planes, and are operated so that four or more satellites can always be observed in a geometrical arrangement from anywhere on the earth. . The GPS device 8 performs positioning calculation based on GPS signals of four satellites selected from observable satellites.

  Usually, in the maintenance inspection of equipment, the inspector often uses both hands to perform work, and when manually inputting the inspection result to the PDA 2, it is necessary to temporarily suspend the work. Also, in maintenance inspections in cold regions in winter, the inspectors work with thick gloves, and it is difficult to manually input inspection results into the PDA 2.

  Therefore, the inspector, for example, stores the PDA 2 in the chest pocket of the work clothes, wears the headset microphone 7, and wears a helmet having the GPS device 8 disposed on the top of the head. Then, by speaking to the microphone 7a according to the voice guidance output from the PDA 2 through the headphones 7b, data is voice-input to the PDA 2 in a hands-free manner. In a cold region, there is a problem that the battery is consumed rapidly if the PDA 2 is exposed to the outside, but this problem can also be solved by storing the PDA 2 in work clothes.

  The inspector holds the remote switch 6 in one hand and speaks while pressing the remote switch 6. That is, the PDA 2 accepts the voice input from the microphone 7a as valid only when the pressing of the remote switch 6 is detected. Therefore, it is not necessary to always accept the input voice as being valid, and the processing load is reduced. In addition, even if the inspector is wearing thick gloves or hands are cramped due to cold, the inspector can instruct to accept the voice by a simple operation of pressing the remote switch 6, which is simple but reliable. Operation is realized.

  A GPS device 8 is disposed on the top of the helmet. When the GPS device 8 is built in the PDA 2, there is a possibility that the antenna 8a cannot receive the GPS signal satisfactorily because the PDA 2 is stored in the work clothes. However, this problem is solved by making the GPS device 8 separate from the PDA 2 and disposing it on the top of the helmet opened in the sky. Moreover, by using the helmet provided with the GPS device 8 as a helmet dedicated to inspection, the helmet is always worn.

2. Principle Next, the principle will be described.
In this embodiment, there are a plurality of facilities (hereinafter referred to as “inspection facilities”) to be inspected by the inspector on the upper and lower routes, and the inspector performs the inspection facility according to the inspection process determined in advance. It will be described as performing maintenance inspection.

  The inspector inspects predetermined items in each inspection facility, and records the inspection results such as measured values by voice input to the PDA 2. Specifically, the inspector first inputs the “facility ID”, which is identification information assigned to the inspection facility, into the PDA 2 by voice. When the equipment ID is recognized by the PDA 2, the inspector voice-inputs the inspection item to the PDA 2 and then voice-inputs the inspection result of the inspection item to the PDA 2.

  The PDA 2 is a dictionary for recognizing a test facility as a dictionary for speech recognition, a “test facility dictionary” that is a dictionary for recognizing a kilometer, “yes”, “no” "Dictionary dictionary" which is a dictionary for recognizing the judgment of an inspector such as "Inspection item dictionary" which is a dictionary for recognizing inspection items, and "Inspection result" which is a dictionary for recognizing inspection results "Dictionary" is memorized.

  The PDA 2 performs voice recognition while switching a dictionary (hereinafter referred to as a “voice recognition target dictionary”) that is a target of arithmetic processing in voice recognition. For example, when the inspector voice-inputs the equipment ID, the inspection equipment dictionary is set as a voice recognition target dictionary, and when the inspector voice-inputs the inspection results, the inspection result dictionary and the judgment dictionary are voice recognition target. Set in dictionary. In this way, by switching the speech recognition target dictionary according to the inspection stage, it is possible to narrow down the speech recognition target dictionary and to speed up speech recognition.

  In addition, when the inspection equipment dictionary is set as a voice recognition dictionary, the PDA 2 uses a vocabulary (hereinafter referred to as “voice recognition target”) that is subject to arithmetic processing in voice recognition from the vocabulary stored in the inspection equipment dictionary. Vocabulary ”is set, and speech recognition is performed based on the speech recognition target vocabulary. Hereinafter, a method for setting the speech recognition target vocabulary will be described in detail.

  First, the PDA 2 determines a range for detecting an inspection facility (hereinafter referred to as “equipment detection range”). The equipment detection range is set by different methods depending on whether positioning by the GPS device 8 is possible or impossible.

  When positioning by the GPS device 8 is possible, that is, when GPS signals can be received from four GPS satellites, the PDA 2 has a value of DOP (Dilution of Precision) based on the sky arrangement of the four GPS satellites. And the equipment detection range is determined based on the calculated DOP value and the position obtained by the positioning calculation of the GPS device 8 (hereinafter referred to as “positioning position”).

  Here, the DOP value is a scale of positioning accuracy based on the sky arrangement of GPS satellites, and means that the positioning accuracy is higher as the DOP value is smaller. There are a plurality of types of DOPs such as HDOP, VDOP, PDOP, and GDOP. In the present embodiment, these types are collectively referred to as “DOP”.

  Specifically, referring to FIG. 2, when the GPS satellites used for positioning calculation are evenly distributed in the sky as seen from the GPS device 8, the positioning accuracy is high and the DOP value is small (FIG. 2). (A)). On the other hand, when the GPS satellites are unevenly distributed in the sky, the positioning accuracy is low and the DOP value is large (FIG. 2B).

  When the DOP value is small, since the positioning accuracy is high, the positioning position is likely to be close to the true position of the GPS device 8. On the other hand, when the DOP value is large, since the positioning accuracy is low, there is a high possibility that the positioning position is deviated from the true position of the GPS device 8. Accordingly, when the DOP value is small, a narrow range from the positioning position is set as the equipment detection range, and when the DOP value is large, a wide range from the positioning position is set as the equipment detection range.

  Specifically, with reference to FIG. 3, when the positioning position is “P”, the inside of a circle having a radius “DOP value × k” centered on the positioning position “P” is set as the equipment detection range. However, “k” is an arbitrary constant and is a value set as appropriate. In FIG. 3, six inspection facilities represented by “56I”, “57I”, “58I”, “X46”, “KI240”, and “KI241” are included in the equipment detection range. Become.

  On the other hand, when positioning by the GPS device 8 is impossible, that is, when GPS signals cannot be received from the four GPS satellites, the PDA 2 outputs voice guidance (synthetic voice) and gives the inspector about a kilometer. Prompt for voice input. Here, the kilometer is indicated on, for example, a kilometer post disposed at a predetermined interval on the end of the track, and the inspector inputs the kilometer indicated on the kilometer post into the PDA 2 by voice.

  Then, when the kilometer input by voice is successfully recognized, the PDA 2 determines the facility detection range based on the kilometer. Specifically, the inside of a circle with a radius “R” centering on the position represented by the kilometer is set as the equipment detection range. However, “R” is a value set as appropriate. Further, if the recognition of the kilometer inputted by voice fails, the PDA 2 sets the range including all the inspection facilities, that is, the entire range as the facility detection range.

  When the equipment detection range is determined by the above-described procedure, the PDA 2 narrows down the inspection equipment based on the inspection process for the inspection equipment included in the equipment detection range. The inspection process includes the order of inspecting the inspection equipment (inspection order), the time period during which the inspection equipment can be inspected (inspectable time period), and whether the inspection equipment has been inspected (inspection state) ) Etc. are stored.

  PDA2 is the inspection order after the inspection equipment that was inspected immediately before, among the inspection equipment included in the equipment detection range, the current time is included in the inspectable time zone, and the inspection is still in progress. Narrow down inspection facilities by extracting unexamined inspection facilities. Then, among the vocabulary stored in the inspection equipment dictionary, the vocabulary corresponding to the narrowed inspection equipment is set as the speech recognition target vocabulary, and the speech recognition is performed based on the speech recognition target vocabulary, thereby specifying the inspection equipment. To do.

3. Functional Configuration Next, the functional configuration will be described.
FIG. 4 is a block diagram showing a configuration of the railway maintenance inspection support system 1. A description of the remote switch 6, the headset microphone 7, and the GPS device 8 described in FIG. 1 is omitted, and the functional configuration of the PDA 2 will be described below.

  The PDA 2 includes a CPU (Central Processing Unit) 20, an operation input unit 30, a display unit 40, a wireless communication unit 50, a flash ROM (Read Only Memory) 60, and a RAM (Random Access Memory) 70. Each unit is a computer system configured to be connected to each other via a bus 80 so that data communication is possible.

  The CPU 20 comprehensively controls each unit in accordance with a system program or the like stored in the flash ROM 60. Further, the CPU 20 performs facility maintenance inspection support processing in accordance with the facility maintenance inspection support program 610 stored in the flash ROM 60.

  The operation input unit 30 is an input device that receives an operation instruction from an inspector and outputs an operation signal corresponding to the operation to the CPU 20. This function is realized by hardware such as buttons and a touch panel, for example, and corresponds to the operation key 3 and the touch panel 4a in FIG.

  The display unit 40 is a display device that performs various displays based on a display signal input from the CPU 20. This function is realized by hardware such as LCD (Liquid Crystal Display) and ELD (Electro Luminescence Display), and corresponds to the display 4b in FIG.

  The wireless communication unit 50 is a communication device that performs wireless communication with other PDAs 2 and management servers based on the control of the CPU 20. This function is realized by a wireless communication module such as a wireless LAN based on IEEE802.11 or Bluetooth (registered trademark), and corresponds to the wireless communication device 5 in FIG.

  The flash ROM 60 is a readable / writable nonvolatile memory, and stores programs, data, and the like for realizing various functions provided in the PDA 2. In the present embodiment, in the flash ROM 60, an equipment maintenance inspection support program 610 read by the CPU 20 and executed as equipment maintenance inspection support processing (see FIGS. 14 to 16), an inspection equipment DB (Data Base) 620, An inspection process DB 630, an inspection result DB 640, and a voice recognition dictionary DB 650 are stored.

  The equipment maintenance inspection support process is a process in which the CPU 20 performs a speech recognition process on the uttered voice input from the microphone 7a while switching the speech recognition target dictionary according to the inspection status, and records the inspection result.

  Specifically, the CPU 20 determines the equipment detection range based on the positioning result by the GPS device 8 and the information about the kilometer inputted by voice, and performs the inspection process for the inspection equipment included in the equipment detection range. Narrow down the inspection equipment based. Then, the CPU 20 sets the vocabulary corresponding to the narrowed inspection facility as the speech recognition target vocabulary, and performs speech recognition of the inspection facility based on the speech recognition target vocabulary. This facility maintenance inspection support process will be described later in detail again using a flowchart.

  The inspection facility DB 620 is a DB in which inspection facility data 621 (621-1, 621-2, 621-3,...) Of all inspection facilities is stored, and an example of the data configuration is shown in FIG. In each inspection equipment data 621, the equipment ID 622, which is identification information of the inspection equipment, the vocabulary number 623 corresponding to the inspection equipment in the inspection equipment dictionary data 651, and the installation of the inspection equipment represented by latitude and longitude A position 624, an up / down identification 625 indicating whether the inspection facility is an upstream / downstream facility, and a kilometer 626 of the inspection facility are stored.

  For example, for the inspection facility of the inspection facility data 621-1, the facility ID 622 is “56a”, and the vocabulary number 623 of the corresponding inspection facility dictionary data 651 is “A1”. Further, the installation location 624 of the inspection facility is “Latitude 38 ° 25: 18” N, “140 ° 48: 32” east longitude, “up / down” identification 625 is “up”, and kilometer 626 is “10.3”. It is.

  In the equipment maintenance inspection support process, the CPU 20 determines the equipment detection range based on the positioning result by the GPS device 8 and the calculated DOP value when the GPS signals from the four GPS satellites can be received. Based on the installation position 624 of the inspection facility, the inspection facility included in the facility detection range is specified.

  On the other hand, when it is impossible to receive GPS signals from the four GPS satellites, the CPU 20 determines the facility detection range based on the kilometer distance voice-input by the inspector. Based on the installation position 624 of the inspection facility, the inspection facility included in the facility detection range is specified.

  The inspection process DB 630 is a DB in which inspection process data 631 (631-1, 631-2, 631-3,...) Of all inspection processes is stored, and a data configuration example is shown in FIG. The types of inspection process data 631 include, for example, data in which inspection facilities that need to be inspected once a week are determined, data that inspect inspection facilities that need to be inspected once a month, 1 Data that defines inspection facilities that need to be inspected once a year is stored.

  Each inspection process data 631 stores a process ID 632 that is identification information of the inspection process. The inspection process data 631 includes an inspection equipment inspection order 633 in the inspection process, an equipment ID 634 of the inspection equipment to be inspected in the inspection order 633, an up / down identification 635 of the inspection equipment, and a kilometer of the inspection equipment. In step 636, an inspectable time zone 637, which is a time zone in which the inspection facility can be inspected, and an inspection state 638 indicating whether or not the inspection facility has been inspected are stored in association with each other.

  However, two or more inspection facilities may be set for one inspection order 633. This is a case where there are a plurality of inspection facilities in the vicinity of a route in a certain kilometer, and it is up to the inspector to determine the order of inspection of these inspection facilities.

  For example, in the inspection process of the inspection process data 631-1, the process ID 632 is “P1”. In the inspection process, the inspection equipment “56i” on the “up” route is the first inspection equipment, and “up” The route inspection facilities “58i” and “60i” are the second inspection facilities. In addition, the inspection facility “56a” is inspected at “7:00 to 12:00”, and the inspection facilities “58i” and “60a” are inspected at “8:00 to 13:00”. The inspection facility “56a” is “inspected”, the inspection facility “58i” is “uninspected”, and the inspection facility “60a” is “inspected”.

  In the equipment maintenance inspection support process, the CPU 20 refers to inspection process data 631 corresponding to the inspection process on the day of the inspector. And, from the inspection equipment included in the set equipment detection range, the inspection order 633 after the inspection equipment that was inspected immediately before, the current time is included in the inspectable time zone 637, and The inspection equipment whose inspection state 638 is “not yet” is extracted. Of the vocabulary stored in the inspection facility dictionary data 651, the vocabulary corresponding to the extracted inspection facility is set as the speech recognition target vocabulary.

  The inspection result DB 640 is a DB in which inspection result data 641 (641-1, 641-2, 641-3,...) For all inspection facilities is stored, and a data configuration example is shown in FIG. Each inspection result data 641 stores a facility ID 642 of the inspection facility and a vertical identification 643 of the inspection facility.

  The inspection result data 641 includes an inspection item 644 that is an inspection item, utterance voice data 645 that is voice data input by the inspector regarding the inspection item 644, and the inspection item 644 in the current inspection. A current measurement value 646 that is a measurement value, a previous measurement value 647 that is a measurement value in the previous inspection of the inspection item 644, and an appropriate range 648 that is an appropriate range of the measurement value of the inspection item 644 are associated with each other. It is remembered.

  For example, the inspection result of the inspection result data 641-1 is data on the inspection result of the inspection facility “56i” on the “up” route. In addition, the speech data 645 regarding “Tongler contact” in the inspection item 644 is “s1.wav”, the current measurement value 646 is “0.0”, and the previous measurement value 647 is “0.0”. The appropriate range 648 is “0-3”.

  In the facility maintenance inspection support processing, the CPU 20 specifies the inspection item 644 by causing the inspector to input the inspection item by voice and performing voice recognition processing on the input voice. Next, the CPU 20 causes the inspector to input the measurement value for the inspection item 644 by voice, and stores the input voice data as the speech voice data 645. Further, the CPU 20 performs a voice recognition process on the speech voice data 645 to specify a measurement value, and stores the measurement value as the current measurement value 646.

  When the inspection result recording process ends, the CPU 20 performs the following process as the inspection result suitability determination process. When the current measurement value 646 stored in the inspection result recording process has not changed by more than a predetermined ratio (for example, 20%) from the previous measurement value 647 and is included in the appropriate range 648, the CPU 20 Judge that it is appropriate. In other cases, it is determined that the inspection result is inappropriate, and a warning message or the like is output as voice on the spot.

  The speech recognition dictionary DB 650 is a DB in which dictionary data for speech recognition is stored, and inspection facility dictionary data 651 that is dictionary data for inspection facility recognition, and kilometer dictionary data 652 that is dictionary data for kilometer recognition. Determination dictionary data 653 which is dictionary data for determination recognition, inspection item dictionary data 654 which is a dictionary for inspection item recognition, and inspection result dictionary data 655 which is dictionary data for inspection result recognition are stored. Yes.

  The RAM 70 forms a work area for temporarily storing a system program executed by the CPU 20, various processing programs, data being processed in various processes, processing results, and the like. In this embodiment, the RAM 70 includes positioning data 710, DOP value data 720, speech voice data 730, voice recognition target dictionary data 740, voice recognition target vocabulary data 750, and voice recognition failure frequency data 760. Remembered.

  FIG. 8 is a diagram illustrating a data configuration example of the positioning data 710. The positioning data 710 stores the positioning position of the GPS device 8 expressed by latitude and longitude. In FIG. 8, “north latitude 39 degrees 54 minutes 27 seconds” and “east longitude 141 degrees 18 minutes 46 seconds” are stored as positioning positions. In the equipment maintenance inspection support process, the CPU 20 acquires the positioning position obtained by the GPS device 8 by positioning calculation, and updates the positioning data 710 at the positioning position.

  FIG. 9 is a diagram illustrating a data configuration example of the DOP value data 720. The DOP value data 720 stores a DOP value based on the sky arrangement of the four GPS satellites used for the positioning calculation. In FIG. 9, “5” is stored as the DOP value. In the equipment maintenance inspection support process, the CPU 20 calculates a DOP value based on a combination of GPS satellites (more specifically, a combination of arrangement positions of GPS satellites) used by the GPS device 8 for positioning calculation, and uses the DOP value. The DOP value data 720 is updated.

  FIG. 10 is a diagram illustrating a data configuration example of the speech audio data 730. The utterance voice data 730 stores the utterance voice data input by the inspector. In FIG. 10, “S1.wav” is stored as the speech voice data. In the facility maintenance inspection support process, when a voice is input from the microphone 7a while the remote switch 6 is being pressed, the CPU 20 updates the speech voice data 730 with the voice data.

  FIG. 11 is a diagram illustrating a data configuration example of the speech recognition target dictionary data 740. The voice recognition target dictionary data 740 stores a dictionary set as the voice recognition target dictionary. In FIG. 11, “inspection equipment dictionary” is stored as a speech recognition target dictionary. In the facility maintenance inspection support process, the CPU 20 sets the speech recognition dictionary as the speech recognition target dictionary according to the inspection status, and updates the speech recognition target dictionary data 740.

  FIG. 12 is a diagram illustrating a data configuration example of the speech recognition target vocabulary data 750. The speech recognition target vocabulary data 750 stores a vocabulary set as the speech recognition target vocabulary. In FIG. 12, “56i”, “X46”, and “ki 240” are stored as the speech recognition target vocabulary. In the facility maintenance inspection support process, the CPU 20 sets the vocabulary corresponding to the inspection facility narrowed down based on the inspection process as the speech recognition target vocabulary, and updates the speech recognition target vocabulary data 750.

  FIG. 13 is a diagram illustrating a data configuration example of the speech recognition failure frequency data 760. The voice recognition failure frequency data 760 stores the number of voice recognition failures. In FIG. 13, “3” is stored as the number of voice recognition failures. In the facility maintenance inspection support process, when the voice recognition fails in the voice recognition process, the CPU 20 increments the number of voice recognition failures and updates the voice recognition failure number data 760.

4). Process Flow Next, the process flow will be described.
14 to 16 are flowcharts showing the flow of the equipment maintenance inspection support process executed in the PDA 2 when the equipment maintenance inspection support program 610 is read and executed by the CPU 20.

  First, the CPU 20 sets the voice recognition failure frequency to “0” (step A1), and updates the voice recognition failure frequency data 760 in the RAM 70. Next, the CPU 20 determines whether or not the GPS device 8 can receive GPS signals from four GPS satellites (step A3). If it is determined that the GPS device 8 can receive GPS signals (step A3; Yes), the GPS device 8 acquires the positioning position obtained by the positioning calculation (step A5), and updates the positioning position data 710 in the RAM 70.

  Next, the CPU 20 calculates the DOP value of the combination of the four GPS satellites used for the positioning calculation (step A7), and updates the DOP value data 720 in the RAM 70. And CPU20 determines an equipment detection range based on the positioning position acquired by step A5, and the DOP value calculated by step A7 (step A9).

  Specifically, the inside of a circle having a radius “DOP value × k” centered on the positioning position “P” is determined as the equipment detection range. Then, with reference to the inspection equipment data 621 stored in the flash ROM 60, the inspection equipment included in the equipment detection range is specified based on the installation position 624 of each inspection equipment.

  On the other hand, if it is determined in step A3 that the GPS signal cannot be received (step A3; No), the CPU 20 sets the kilo-distance dictionary as the speech recognition target dictionary (step A11), and the speech recognition target dictionary in the RAM 70. Data 740 is updated.

  Then, the CPU 20 causes the headphone 7b to output a voice that instructs the inspector to utter about a kilometer (step A13). Specifically, a synthesized voice such as “Please speak about a kilometer” is output from the headphones 7b.

  Next, the CPU 20 determines whether or not the remote switch 6 has been pressed by the inspector (step A15). If it is determined that the remote switch 6 has not been pressed (step A15; No), the CPU 20 returns to step A13. If it is determined that the remote switch 6 has been pressed (step A15; Yes), the CPU 20 accepts voice input (step A17) and uses the input voice data as utterance voice data, and the utterance voice in the RAM 70. Data 730 is updated.

  Thereafter, the CPU 20 determines whether or not the remote switch 6 has been released by the inspector (step A19). If it is determined that the remote switch 6 has not been released yet (step A19; No), the CPU 20 returns to step A17. If it is determined that the remote switch 6 has been released (step A19; Yes), the CPU 20 performs voice recognition processing (step A21).

  Next, the CPU 20 determines whether or not the voice recognition is successful (step A23). However, although the voice recognition success / failure determination will be described as one step here, in reality, when the inspection equipment can be specified by the voice recognition processing, the inspection equipment is output from the headphones 7b as a sound (repeated). This prompts the inspector to make a determination such as “yes” or “no”.

  Specifically, a synthesized voice such as “press the remote switch once if yes, press the remote switch twice if no” is output. For example, if the specified inspection facility is “56I”, a synthesized voice such as “Is the inspection facility 56I?” Is output, and the remote switch 6 is pressed “once” by the inspector. If “Yes” or “twice” is pressed, it is determined that “No” is determined.

  If the inspection equipment cannot be specified in the voice recognition process (recognition error), or if the inspector makes a “no” determination when the inspection equipment is read back (false recognition), It is determined that recognition has failed. The same applies to the voice recognition success / failure determination in the following description.

  If it is determined in step A23 that the speech recognition is successful (step A23; Yes), the CPU 20 determines the equipment detection range based on the kilometer (step A25). More specifically, the inside of a circle with a radius “R” centering on the kilometer position identified in the speech recognition process is set as the equipment detection range. Then, with reference to the inspection equipment data 621 stored in the flash ROM 60, the inspection equipment included in the equipment detection range is specified based on the installation position 624 of each inspection equipment.

  If it is determined in step A23 that voice recognition has failed (step A23; No), the CPU 20 increments the number of voice recognition failures (step A27) and updates the voice recognition failure number data 760 in the RAM 70.

  Then, the CPU 20 determines whether or not the number of speech recognition failures stored in the speech recognition failure frequency data 760 is “3 times” or more (step A28), and when it is determined that the number is not “3 times” or more. (Step A28; No), the kilo-speech utterance instruction voice output from the headphones 7b is changed to a voice for re-input instruction (step A29). Specifically, the utterance instruction voice about kilo kilometers is changed to a voice such as “Please speak another kilometer”. Then, the CPU 20 returns to step A13.

  On the other hand, when it is determined that the number of voice recognition failures is “three times” or more (step A28; Yes), the CPU 20 determines a range (all range) including all inspection facilities as a facility detection range (step S28). A30).

  Therefore, when the number of voice recognition failures is less than the threshold value, the voice input of repeated kilometer is accepted, but when the number of voice recognition failures is more than the threshold value, the entire range is forcibly determined as the equipment detection range. Become. Note that the threshold “3 times” for the number of voice recognition failures in step A28 is merely an example.

  When the equipment detection range is determined in any one of steps A9, A25, and A30, the CPU 20 sets the inspection equipment dictionary as the voice recognition target dictionary (step A31), and updates the voice recognition target dictionary data 740 in the RAM 70.

  Then, the CPU 20 narrows down inspection facilities based on the inspection process for inspection facilities included in the facility detection range (step A33). Specifically, in the inspection process data 631 stored in the inspection process DB 630 of the flash ROM 60, the inspection process data 631 corresponding to the inspection process on the day of the inspector is referred to.

  And the inspection order 633 after the inspection facility in which the inspection result data 641 is stored immediately before, the current time is included in the inspectable time zone 637, and the inspection state 638 is “not yet”. To extract.

  Next, the CPU 20 sets the vocabulary of the inspection facility dictionary corresponding to the inspection facility narrowed down in step A33 as the speech recognition target vocabulary (step A35), and updates the speech recognition target vocabulary data 750 in the RAM 70.

  Then, the CPU 20 causes the headphone 7b to output a sound for instructing the utterance of the inspection facility to the inspector (step A37). Specifically, a synthesized voice such as “Please speak the equipment ID” is output from the headphones 7b.

  Next, the CPU 20 determines whether or not the remote switch 6 has been pressed by the inspector (step A39). If it is determined that the remote switch 6 has not been pressed (step A39; No), the CPU 20 returns to step A37. If it is determined that the remote switch 6 has been pressed (step A39; Yes), the CPU 20 accepts voice input (step A41), and the voice data in the RAM 70 is set as voice data for the voice input. Data 730 is updated.

  Thereafter, the CPU 20 determines whether or not the press of the remote switch 6 has been released by the inspector (step A43). If it is determined that the remote switch 6 has not been released yet (step A43; No), the CPU 20 returns to step A41. If it is determined that the remote switch 6 has been released (step A43; Yes), the CPU 20 performs a voice recognition process (step A45).

  Next, the CPU 20 determines whether or not the voice recognition is successful (step A47). If it is determined that the voice recognition is successful (step A47; Yes), the inspection item dictionary is set as the voice recognition target dictionary (step A48). The voice recognition target dictionary data 740 in the RAM 70 is updated.

  Then, the CPU 20 causes the headphone 7b to output a sound that instructs the inspector to utter the inspection item (step A49). Specifically, a synthesized voice such as “Please speak the inspection item” is output from the headphones 7b. And CPU20 specifies a test | inspection item by performing a speech recognition process with respect to the input audio | voice (step A50).

  Next, the CPU 20 sets the inspection result dictionary and the determination dictionary as the speech recognition target dictionary (step A51), and updates the speech recognition target dictionary data 740 in the RAM 70. Then, the CPU 20 causes the headphone 7b to output a sound that instructs the inspector to speak the test result (step A52). Specifically, a synthesized voice such as “Please speak the test result” is output from the headphones 7b.

  Thereafter, the CPU 20 performs an inspection result recording process (step A53). Specifically, the CPU 20 sets the input voice data as utterance voice data 645 and stores it in the test result data 641 of the flash ROM 60 in association with the test item 644 identified in step A50. Since the utterance voice data 645 is stored in association with the inspection item 644, the inspector again listens to his / her voice (the content of the utterance for each inspection item 644) after the inspection is completed, and the input is incorrect. It is possible to make a final confirmation of whether there is any.

  Thereafter, the CPU 20 performs a voice recognition process on the stored utterance voice data 645 to identify a measurement value. Then, the specified measurement value is set as the current measurement value 646 and stored in the inspection result data 641 of the flash ROM 60 in association with the inspection item 644.

  When the inspection result recording process is finished, the CPU 20 performs an inspection result suitability determination process (step A55). Specifically, the current measurement value 646 stored in the inspection result data 641 in the inspection result recording process does not change by a predetermined ratio (for example, 20%) or more from the previous measurement value 647 and is included in the appropriate range 648. If it is, the test result is determined to be appropriate. On the other hand, when the current measurement value 646 has changed by a predetermined ratio (for example, 20%) or more from the previous measurement value 647, or when it is not included in the appropriate range 648, it is determined that the inspection result is inappropriate.

  If the CPU 20 determines that the inspection result is appropriate (step A57; appropriate), the process proceeds to step A65. On the other hand, if it is determined that the inspection result is inappropriate (step A57; inappropriate), the CPU 20 causes the headphone 7b to output a sound prompting the inspector to confirm the inspection result (step A59). Specifically, a synthesized voice such as “Is there an error in the test result?” Is output from the headphones 7b.

  Next, the CPU 20 determines whether or not re-recording is instructed by the inspector (step A61). Specifically, when the remote switch 6 is pressed by the inspector and a determination of “no” is input to the confirmation voice of the inspection result output in step A59, re-recording is instructed. judge.

  If it is determined that re-recording has been instructed (step A61; Yes), the CPU 20 erases the utterance voice data 645 stored in association with the inspection item 644 of the inspection result data 641 (step A63). Return to Step A52.

  On the other hand, when it is determined in step A61 that re-recording has not been instructed (step A61; No), the CPU 20 refers to the inspection result data 641 and determines whether or not the recording of the inspection results for all the inspection items 644 has ended. (Step A65), and when it is determined that the process has not been completed (step A65; No), the process returns to step A48.

  On the other hand, if it is determined in step A65 that the recording of the inspection result has been completed (step A65; Yes), the CPU 20 outputs a sound prompting the inspector to confirm the completion of the inspection from the headphones 7b (step A67). Specifically, a synthesized voice such as “Do you want to end the examination?” Is output from the headphones 7b.

  Then, the CPU 20 determines whether or not the inspection end is instructed by the inspector (step A69). Specifically, it is determined whether or not the remote switch 6 is pressed by the inspector and a determination of “Yes” is input by voice.

  If it is determined that the inspection end is not instructed (step A69; No), the CPU 20 returns to step A1, and if it is determined that the inspection end is instructed (step A69; Yes), the equipment maintenance inspection support process Exit.

  On the other hand, if it is determined in step A47 that voice recognition has failed (step A47; No), the CPU 20 increments the number of voice recognition failures (step A71), and updates the voice recognition failure number data 760 in the RAM 70.

  Then, the CPU 20 determines whether or not the number of speech recognition failures stored in the speech recognition failure frequency data 760 is “5 times” or more (step A73), and when it is determined that the number is not “5 times” or more. (Step A73; No) returns to Step A37.

  On the other hand, if it is determined that the number of voice recognition failures is “five times” or more (step A73; Yes), the CPU 20 sets the judgment dictionary as the voice recognition target dictionary (step A75), and the voice recognition target dictionary in the RAM 70. Data 740 is updated. The threshold value “5 times” for the number of voice recognition failures in step A73 is merely an example.

  After that, the CPU 20 refers to the inspection process data 631 of the flash ROM 60 and extracts inspection equipment whose inspection state 638 is “not yet” (hereinafter referred to as “uninspected equipment”) (step A77).

  Then, the CPU 20 selects the first uninspected facility among the extracted uninspected facilities (step A79), and causes the headphone 7b to output a sound prompting the inspector to confirm the uninspected facility (step A81). Specifically, for example, if the selected uninspected equipment is “58a”, a synthesized voice such as “Is the inspection equipment 58a?” Is output from the headphones 7b.

  Next, the CPU 20 determines whether or not the remote switch 6 has been pressed by the inspector (step A83). If it is determined that the remote switch 6 has not been pressed (step A83; No), the CPU 20 returns to step A81. If it is determined that the remote switch 6 has been pressed (step A83; Yes), the CPU 20 accepts voice input (step A85), and the voice data in the RAM 70 is set as voice data for the voice input. Data 730 is updated.

  Thereafter, the CPU 20 determines whether or not the remote switch 6 has been released by the inspector (step A87). If it is determined that the remote switch 6 has not been released yet (step A87; No), the CPU 20 returns to step A85. On the other hand, if it is determined that the remote switch 6 has been released (step A87; Yes), the CPU 20 performs a voice recognition process that identifies whether the input voice is “yes” or “no”. (Step A89).

  Next, the CPU 20 determines whether or not the speech recognition is successful (step A91). If it is determined that the speech recognition is successful (step A91; Yes), the recognition result is determined (step A93). If it is determined that the recognition result is “yes” (step A93; yes), the process proceeds to step A48.

  When it is determined in step A93 that the recognition result is “No” (step A93; No), the CPU 20 selects the next uninspected facility among the uninspected facilities extracted in step A77 (step A95). Return to Step A81.

  If it is determined in step A91 that voice recognition has failed (step A91; No), the CPU 20 increments the number of voice recognition failures (step A97) and updates the voice recognition failure number data 760 in the RAM 70.

  Then, the CPU 20 determines whether or not the number of speech recognition failures stored in the speech recognition failure frequency data 760 is “10 times” or more (step A99), and when it is determined that the number is not “10 times” or more. (Step A99; No) returns to Step A81.

  On the other hand, if it is determined that the number of voice recognition failures is “10 times” or more (step A99; Yes), the CPU 20 causes the headphone 7b to output a voice instructing the inspector to make a determination using the remote switch 6. (Step A101).

  Specifically, a synthesized voice such as “press the remote switch once if yes, press the remote switch twice if no” is output from the headphones 7b. Note that the threshold “10 times” for the number of voice recognition failures in step A99 is merely an example.

  Then, the CPU 20 selects the first uninspected facility among the uninspected facilities extracted in Step A77 (Step A103), and causes the headphone 7b to output a sound prompting the inspector to confirm the uninspected facility (Step A105). ). Specifically, for example, if the selected uninspected equipment is “58a”, a synthesized voice such as “Is the inspection equipment 58a?” Is output from the headphones 7b.

  Next, the CPU 20 determines whether or not the remote switch 6 has been pressed by the inspector (step A107). If it is determined that the remote switch 6 has not been pressed (step A107; No), the CPU 20 returns to step A105. If it is determined that the remote switch 6 has been pressed (step A107; Yes), the CPU 20 determines the number of times that the remote switch 6 has been pressed (step A109).

  If it is determined that the number of times of pressing is “once” (step A109; once), the CPU 20 determines that the determination by the inspector is “yes”, and the process proceeds to step A48. To do.

  If it is determined in step A109 that the number of times of pressing is “twice” (step A109; twice), the CPU 20 determines that the determination by the inspector is “no” and the extraction is performed in step A77. The next uninspected facility is selected from the uninspected facilities (step A111), and the process returns to step A105.

5. According to the present embodiment, based on the positioning accuracy of the current position by the GPS device 8, a predetermined range centered on the current position measured is variably determined as the equipment detection range. And the inspection equipment based on the inspection process is narrowed down for the inspection equipment whose installation position is included in the equipment detection range. Then, the inspection equipment corresponding to the uttered voice input from the microphone 7a is identified through voice recognition processing based on the voice data corresponding to the narrowed inspection equipment. Then, the speech voice input from the microphone 7a is subjected to voice recognition processing, and the inspection result data is input to the PDA 2 and recorded as the inspection result for the specified inspection facility.

  Since the current position is acquired by autonomous positioning using GPS, there is no need to arrange ground facilities such as a sensor for detecting the location of the inspector. In addition, since the inspection result is recorded based on the uttered voice input from the microphone 7a, the inspector does not need to manually input the inspection result. Therefore, it is possible to save labor for inspection records of railway facilities without additionally arranging ground facilities.

  Further, narrowing down of inspection facilities is performed based on inspection processes such as the inspection order of inspection facilities, the inspection possible time zone, and the inspection state. Accordingly, since the inspection equipment is appropriately narrowed down by excluding the inspection equipment that is not the inspection target, the processing load related to voice recognition is reduced.

  Also, when the number of voice recognition failures reaches the predetermined number, instead of letting the inspector input the inspection equipment by voice, the inspection equipment is read aloud sequentially, and a voice of “Yes” or “No” is judged each time. By prompting the inspector for input, the inspection equipment is specified. If the number of voice recognition failures still increases, the determination of “Yes” or “No” is made by depressing the remote switch 6. Therefore, even when voice recognition cannot be performed properly, it is possible to reliably specify the inspection facility.

6). Modification 6-1. Although the present embodiment has been described as detecting the current position using GPS in this embodiment, the current position can also be detected by other methods. For example, the PDA 2 may be provided with a communication device such as a mobile phone or a PHS, and the current position may be detected by communicating with a mobile phone / PHS base station.

  For example, when there is one base station with which the mobile phone / PHS can communicate, the location within the communication range of the location of the base station is the current location, and there are two base stations with which the mobile phone / PHS can communicate. In this case, the position obtained by the three-point positioning or the overlapping range within the communication range of the two base stations is the current position.

  Therefore, since the accuracy of position detection can be determined according to how many communication base stations can communicate with, it is possible to variably determine the equipment detection range and appropriately narrow down the inspection equipment. It becomes.

6-2. Narrowing down inspection equipment Based on the position of the inspection equipment inspected immediately before by the inspector, the next inspection equipment may be narrowed down. Specifically, the inspection facility is narrowed down so as to extract inspection facilities existing within a circle having a predetermined radius centered on the position of the inspection facility inspected immediately before by the inspector. As a result, for example, it is possible to extract only the inspection objects included in the vicinity of the installation position of the inspection facility inspected immediately by the inspector, so that appropriate inspection objects can be narrowed down. Is possible.

  Further, the inspection equipment may be narrowed down based on the upper / lower identification 635 included in the inspection process data 631. For example, there is a case where it is determined in advance that an inspection facility for an upstream line in the morning and an inspection facility for a downstream line in the afternoon. In this case, if the current time is in the morning, narrowing down is performed so that only the inspection equipment whose up / down identification 635 is “up” is extracted, and if the current time is in the afternoon, the up / down identification 635 is “down”. Narrow down to extract only equipment.

  In any case, it is more preferable to narrow down in combination with the inspection order 633, the inspection possible time zone 637, and the inspection state 638.

6-3. Determination of equipment detection range In this embodiment, when GPS signals cannot be received from four GPS satellites, voice input of about a kilometer by an inspector is accepted, and the equipment detection range is determined based on the kilometer. Explained as In other words, when positioning by GPS is impossible, acquisition of about a kilometer is activated.

  However, instead of adopting such a configuration, acquisition of about a kilometer may be activated based on the positioning accuracy by GPS. For example, when the calculated DOP value is equal to or greater than a predetermined threshold, acquisition of about a kilometer is activated. Thereby, even when the positioning accuracy is low, the facility detection range can be set appropriately.

6-4. Specifying Inspection Items When specifying the inspection items spoken by the inspector, items that have already been inspected by the inspector may be excluded from the target of speech recognition. Specifically, by referring to the inspection result data 641, the inspection item 644 for which the current measurement value 646 is not stored, that is, the inspection item 644 that has not been inspected by the inspector is specified. Then, among the vocabulary stored in the test result dictionary data 654, the vocabulary corresponding to the test item 644 is set as the speech recognition target vocabulary.

6-5. Determination of inspection result input Depending on the inspection item, there is a measurement value that can be input as an inspection result, which can be selected by a choice such as binary or ternary value. In such a case, the input determination of the inspection result may be performed by pressing the remote switch 6. Specifically, for example, when the measurement values that can be input are limited to two values “0” and “1”, the number of times the remote switch 6 is pressed is “0” and 2 times. A case is defined as “1”.

  In the test result recording process, when the number of voice recognition failures reaches a predetermined threshold, the input acceptance is switched from voice input to operation input, and “0”, 2 when the remote switch 6 is pressed once. If pressed once, it is determined that “1” has been input. When the measured value is ternary, the input determination may be performed in the same manner based on the number of times the remote switch 6 is pressed once to three times.

The figure which shows schematic structure of a railway equipment maintenance inspection support system. Explanatory drawing of a DOP value. Explanatory drawing of an equipment detection range. The figure which shows the function structure of a railway equipment maintenance inspection assistance system. The figure which shows the data structural example of inspection equipment DB. The figure which shows the data structural example of inspection process DB. The figure which shows the data structural example of equipment result DB. The figure which shows the data structural example of positioning data. The figure which shows the data structural example of DOP value data. The figure which shows the data structural example of speech audio | voice data. The figure which shows the data structural example of speech recognition object dictionary data. The figure which shows the data structural example of speech recognition object vocabulary data. The figure which shows the data structural example of speech recognition failure frequency data. The flowchart which shows the flow of an equipment maintenance inspection support process. The flowchart which shows the flow of an equipment maintenance inspection support process. The flowchart which shows the flow of an equipment maintenance inspection support process.

Explanation of symbols

1 Railway equipment maintenance inspection support system 2 PDA
3 Operation Key 4 Touch Screen 4a Touch Panel 4b Display 5 Wireless Communication Device 6 Remote Switch 7 Headset Microphone 7a Microphone 7b Headphone 8 GPS Device 8a Antenna 8b GPS Receiver 20 CPU
30 Operation Input Unit 40 Display Unit 50 Wireless Communication Unit 60 Flash ROM
610 Equipment maintenance inspection support program 620 Inspection equipment DB
630 Inspection process DB
640 Inspection result DB
650 Speech recognition dictionary DB
70 RAM
710 Positioning data 720 DOP value data 730 Speech voice data 740 Voice recognition target dictionary data 750 Voice recognition target vocabulary data 760 Voice recognition failure frequency data 80 bus

Claims (11)

With a microphone,
Position acquisition means for acquiring the current position;
An object information storage means for storing the installation position of each inspection object and dictionary data for identifying the inspection object by a predetermined voice recognition process;
Based on the positional relationship between the current position acquired by the position acquisition means and the installation position of each inspection object stored in the object information storage means, a narrowing means for narrowing down the next inspection object;
An inspection object specifying means for specifying an inspection object corresponding to an utterance voice input from the microphone by performing voice recognition processing based on dictionary data corresponding to the inspection object narrowed down by the narrowing down means;
A test result input means for performing voice recognition processing on speech speech input from the microphone and inputting test result data;
A recording unit for recording the inspection result data input by the inspection result input unit as an inspection result for the inspection target specified by the inspection target specifying unit;
A portable railway equipment maintenance inspection support system equipped with A remote switch,
A pressing operation detecting means for detecting a pressing operation of the remote switch;
Utterance voice reception control means for accepting voice input from the microphone as valid when detected by the pressing operation detection means;
With
The railway facility maintenance inspection support system according to claim 1, wherein the inspection object specifying unit and the inspection result input unit perform voice recognition processing on the voice received by the utterance voice reception control unit. Further comprising accuracy determining means for determining the accuracy of the current position acquired by the position acquiring means;
The railway facility maintenance inspection support system according to claim 1, wherein the narrowing-down unit narrows down a next inspection target by varying a narrowing-down condition according to the accuracy determined by the accuracy determination unit.   The railway facility maintenance inspection support according to any one of claims 1 to 3, wherein the narrowing down means further narrows down the next inspection object based on the installation position of the inspection object that has been inspected immediately before. system. Further comprising inspection order storage means for storing the inspection order of each inspection object;
The railway facility maintenance inspection support system according to any one of claims 1 to 4, wherein the narrowing down means further narrows down the next inspection object based on the inspection order stored in the inspection order storage means. Further comprising an inspectable time zone storage means for storing the inspectable time zone of each inspection object;
The railway facility maintenance inspection according to any one of claims 1 to 5, wherein the narrowing down means further narrows down a next inspection object based on an inspectable time zone stored in the inspectable time zone storage means. Support system. The object information storage means stores a kilometer as the installation position of the inspection object,
The position acquisition means includes a kilometer acquisition means for acquiring a kilometer indicating a current position by performing voice recognition processing on the speech sound input from the microphone,
The narrowing down means narrows down the next inspection object based on the positional relationship between the kilometer distance acquired by the kilometer distance acquisition means and the kilometer distance of each inspection object stored in the object information storage means.
The railway equipment maintenance inspection support system according to any one of claims 1 to 6. A receiving means for receiving a predetermined external radio wave that is a radio wave capable of specifying the current position;
The position acquisition means includes
Current position specifying means for specifying the current position based on the external radio wave received by the receiving means;
Kilometer distance acquisition activation determining means for determining whether or not to activate the kilometer distance acquisition means based on specific success / failure and / or accuracy by the current position specifying means;
Have
The railway facility maintenance inspection support system according to claim 7, wherein the kilometer acquisition unit functions when it is determined that the kilometer acquisition activation determination unit is activated.   The recording means comprises test result voice data recording means for recording the speech data that is the object of the voice recognition processing by the test result input means in association with the test object specified by the test object specifying means. The railway facility maintenance inspection support system according to any one of claims 1 to 8. Speakers,
Specific success / failure determination means for determining the specific success / failure of the inspection object by the voice recognition processing of the inspection object specifying means;
With
The inspection object specifying unit includes a re-identifying unit that prompts re-input of the uttered voice and performs voice recognition processing again when it is determined that the specific success / failure determining unit has failed,
Read-out control means for performing output control from the speaker of a sound that sequentially reads out each inspection object when the number of times of failure determination by the specific success / failure determination means reaches a predetermined number of times,
Success / failure input means for sequentially inputting whether or not the inspection object sequentially read out by the reading control means is the next inspection object;
The next inspection object is specified based on the input of the success / failure input means when the number of times of failure determination by the specific success / failure determination means reaches a predetermined number. Railway equipment maintenance inspection support system described in 1. A portable computer with a microphone,
The installation position of each inspection object included in the object information associated with the installation position of each inspection object and dictionary data for identifying the inspection object by a predetermined voice recognition process, and the current position Narrowing means to narrow down the next inspection object based on the positional relationship of
An inspection object specifying means for specifying an inspection object corresponding to an utterance voice input from the microphone by performing voice recognition processing based on dictionary data corresponding to the inspection object narrowed down by the narrowing-down means;
A test result input means for inputting test result data by performing voice recognition processing on the speech voice input from the microphone;
Recording means for recording the inspection result data input by the inspection result input means as the inspection result for the inspection object specified by the inspection object specifying means;
Program to function as.

Images