CN112363400B - Cable tunnel intrusion monitoring method based on optical fiber sensor signal and abnormal coding - Google Patents

Abstract

The invention relates to a cable tunnel intrusion monitoring method based on optical fiber sensor signals and abnormal codes. The regression prediction model based on abnormal condition codes instead of specific numerical values can be obtained by utilizing the method, the problem that the conventional prediction method cannot stably run for a long time due to the binding of the conventional prediction method and specific experimental numerical values can be solved, and the intrusion condition of the cable tunnel can be effectively monitored in a longer time range.

Description

Cable tunnel intrusion monitoring method based on optical fiber sensor signal and abnormal coding

Technical field:

The invention relates to a cable tunnel intrusion monitoring method based on optical fiber sensor signals and abnormal codes, which is used for monitoring the intrusion situation of cable tunnels, and belongs to the technical field of cable tunnel state safety monitoring.

Background technique:

Power cables are an important carrier of power transmission, and cable tunnel failures affect the performance of power grid construction. Criminals often invade cable tunnels, and the internal cross interconnection wires, grounding wires, grounding copper bars, etc. in the tunnel are often stolen and stolen, which not only causes huge economic losses, but also seriously affects the safe and effective operation of the power system. Therefore, It is very necessary to monitor the cable tunnel intrusion.

Currently, there are two ways to monitor the intrusion of cable tunnels: one is to have a special person check the tunnel situation at fixed intervals. This method not only requires huge labor costs, but also reduces the possibility of intrusion due to the long cable distance. There will be cases where the detection is not timely or missed. Second, the optical fiber sensor signals are used to reflect the situation in the tunnel, and corresponding artificial intelligence models are established to predict possible intrusions. These models are usually built based on laboratory data or cable tunnels. It is constructed from the data obtained at the initial stage, and its characteristic is that it is bound to a specific value, and the accuracy will be very high in the initial stage of the system operation; however, due to the fact that the tunnels with large geographical distribution will inevitably appear settlement and deformation during the long-term use process, So these numerical-based models will become more and more inaccurate, and the false positive rate will continue to rise after running for a period of time.

Therefore, it is necessary to propose a method, which can avoid the problem of binding traditional methods and specific experimental values, and monitor the intrusion of cable tunnels in a long time range.

SUMMARY OF THE INVENTION

Aiming at the problems existing in the prior art, the present invention provides a cable tunnel intrusion monitoring method based on optical fiber sensor signal and abnormal code description. For the input optical fiber sensor signal, the method can encode the abnormal situation, and then obtain the regression model based on the encoding, and then realize the monitoring of the intrusion of the cable tunnel.

A method for monitoring cable tunnel intrusion based on optical fiber sensor signals and abnormal codes according to the present invention includes the following steps:

S1, input the cable tunnel fiber optic sensor signal data list HistoryList in the normal state, input the data list ErrorList of the intrusion of the cable tunnel, obtain the number of HistoryList data entries HistoryNum, establish the normal state statistics array HArray; obtain the number of ErrorList data entries ErrorNum, establish the intrusion state statistics array EArray, and obtain the maximum value of intrusion position DistanceMax;

S101 , input the cable tunnel optical fiber sensor signal data list HistoryList in the normal state, and the fields of the HistoryList include:

HSData: Array of fiber optic sensor signals in normal state, whose data type is a floating-point array of 100 elements.

HRQ: The time corresponding to the optical fiber sensor to obtain the HSData data;

S102, input the data list ErrorList of the intrusion of the cable tunnel, and the fields of ErrorList include:

ESData: Array of optical fiber sensor signals in case of intrusion, whose data type is a floating-point array of 100 elements;

EDistance: The distance between the location where the intrusion occurs in the tunnel and the fiber optic sensor, the data type is floating point data;

ERQ: The time corresponding to the optical fiber sensor acquiring ESData data;

S103, HistoryNum=Number of data entries in HistoryList, ErrorNum=Number of data entries in ErrorList; HArray=100-element floating-point array; EArray=100-element floating-point array; DistanceMax==Maximum value of EDistance field of ErrorList;

S104, set the value of all elements of HArray to 0; set all elements of EArray to 0; sort HistoryList from small to large according to the value of HRQ; sort ErrorList from small to large according to the value of ERQ;

S105, history data counter HistoryNumCounter=1;

S106, HArray=HArray+(HistoryList[HistoryNumCounter].HSData)×(1+(HistoryNumCounter/HistoryNum-0.5) /100);

S107, HistoryNumCounter=HistoryNumCounter+1;

S108, if HistoryNumCounter>HistoryNum, go to S109, otherwise go to S106;

S109, HArray=HArray/HistoryNum;

S110, intrusion counter ErrorNumCounter=1;

S111, EArray=EArray+(ErrorList[ErrorNumCounter].ESData);

S112, ErrorNumCounter=ErrorNumCounter+1;

S113, if ErrorNumCounter>ErrorNum, go to S114, otherwise go to S111;

S114, EArray=EArray/ErrorNum;

S2, establish the optical fiber signal abnormal coding description operator OFeature, the input variable of OFeature is OFeatureInput, the type of OFeatureInput is a floating-point array with 100 elements, the output variable of OFeature is OFeatureOutput, and OFeatureOutput is a floating-point array with 400 elements;

S201 , an optical fiber signal abnormality encoding description operator OFeature is established, the input variable of OFeature is OFeatureInput, and the type of OFeatureInput is a floating-point array of 100 elements;

S202, establish a floating-point array of OFeature Output variable OFeatureOutput=400 elements, and set the value of all elements of the array to 0;

S203, the operator counter OFeatureCounter=1;

S204, if OFeatureInput[OFeatureCounter]>HArray[OFeatureCounter], then OFeatureOutput[OFeatureCounter]=1;

S205, if OFeatureInput[OFeatureCounter]>EArray[OFeatureCounter], then OFeatureOutput[OFeatureCounter +100]=1;

S206, the operator's first temporary storage variable OFeatureTemp1=0; the operator's second temporary storage variable OFeatureTemp1=0;

S207, OFeatureTemp1=abs(OFeatureInput[OFeatureCounter]-HArray[OFeatureCounter]), where abs is the absolute value of the calculation;

S208, OFeatureTemp2=abs(OFeatureInput[OFeatureCounter]-EArray[OFeatureCounter]), where abs is the calculated absolute value;

S209, if OFeatureTemp1>OFeatureTemp2,

this OFeatureOutput[OFeatureCounter+200]=1;

S210, if abs(OFeatureTemp1-OFeatureTemp2)>0.5, OFeatureOutput[OFeatureCounter+300]=1, where abs is the calculated absolute value;

S211, OFeatureCounter=OFeatureCounter+1;

S212, if OFeatureCounter>100, go to S213, otherwise go to S204;

S213, take OFeatureOutput as the output of OFeature;

S3, use the optical fiber signal feature description operator OFeature to construct the decision information table DecisionTable, and use the DecisionTable to train the regression model RegModel;

S301, a decision information table DecisionTable is established, and the field structure of DecisionTable is as follows:

DecisionInput: decision input, a floating-point array of 400 elements;

DecisionDistance: The distance of the intrusion situation, its data type is floating-point data;

S302, the first variable of the decision counter CreateCounter1=1;

S303, the decision temporary storage array variable CTempArray= is calculated by OFeature, and the operator input OFeatureInput=HistoryList[CreateCounter1].HSData;

S304, create a new record in DecisionTable; in the newly added record, the value of the DecisionInput field is CTempArray, and the value of DecisionDistance is -1;

S305, CreateCounter1=CreateCounter1+1;

S306, if CreateCounter1>HistoryNum, go to S307, otherwise go to S303;

S307, the second variable of the decision counter CreateCounter2=1;

S308, CTempArray=Calculate by OFeature, operator input OFeatureInput=ErrorList[CreateCounter2].ESData;

S309, create a new record in DecisionTable; in the newly added record, the value of the DecisionInput field is CTempArray, and the value of DecisionDistance is ErrorList [CreateCounter2].EDistance/DistanceMax;

S310,CreateCounter2=CreateCounter2+1;

S311, if CreateCounter2>ErrorNum, go to S312, otherwise go to S308;

S312, establish a neural network regression analysis model RegModel;

S313, use the DecisionInput field of the DecisionTable as the input of the RegModel model, and use the DecisionDistance field of the DecisionTable as the output of the RegModel model; input all the data of the DecisionTable to train the RegModel;

S4, input the data CurrentArray collected by the optical fiber sensor signal in the cable tunnel, and predict the intrusion situation;

S401, input the data CurrentArray collected by the optical fiber sensor signal in the cable tunnel, and CurrentArray is a floating-point array with 100 elements;

S402, predicting the temporary storage feature variable Feature= is calculated by OFeature, and the operator input OFeatureInput=CurrentArray;

S403, the prediction result variable PResult= takes Feature as the input of RegModel and obtains the output of RegModel;

S404, if PResult<0, output "no intrusion situation" and go to S408, otherwise go to S405;

S405, the final predicted distance ResultDis=PResult×DistanceMax;

S406, output "there is an intrusion situation", and output ResultDis as the location of the intrusion;

S407, go to S409;

S408, HArray=(CurrentArray-HArray)/(HistoryNum×10)+HArray;

S409, the prediction process ends.

The beneficial effects of the present invention are:

Provided is a cable tunnel intrusion monitoring method based on optical fiber sensor signal and abnormal code description. For the input optical fiber sensor signal, the method can encode the abnormal situation, and then obtain the regression model based on the encoding, and then realize the monitoring of the intrusion of the cable tunnel. Using the patent of the present invention, a regression prediction model based on abnormal condition codes rather than specific values can be obtained, which can avoid the problem of inability to run stably for a long time caused by the binding of traditional prediction methods and specific experimental values, and is effective in a long time range. monitoring of cable tunnel intrusion.

Detailed ways

The present invention is further described by the following examples, and does not limit the present invention in any way. On the premise of not departing from the technical solutions of the present invention, any changes or changes that are easily realized by those of ordinary skill in the art made by the present invention will be fall within the scope of the claims of the present invention.

Example 1

S1, input the cable tunnel fiber optic sensor signal data list HistoryList in the normal state, input the data list ErrorList of the intrusion of the cable tunnel, obtain the number of HistoryList data entries HistoryNum, establish the normal state statistics array HArray; obtain the number of ErrorList data entries ErrorNum, establishes the intrusion state statistics array EArray, and obtains the maximum value of intrusion position DistanceMax.

S101 , input the cable tunnel optical fiber sensor signal data list HistoryList in the normal state, and the fields of the HistoryList include:

HSData: Array of fiber optic sensor signals in normal state, whose data type is a floating-point array of 100 elements.

HRQ: The time corresponding to the acquisition of HSData data by the optical fiber sensor.

S102, input the data list ErrorList of the intrusion of the cable tunnel, and the fields of ErrorList include:

ESData: Array of fiber optic sensor signals in case of intrusion, whose data type is a floating-point array of 100 elements.

EDistance: The distance between the location where the intrusion occurs in the tunnel and the fiber optic sensor, and its data type is floating-point data.

ERQ: The time corresponding to the fiber optic sensor acquiring ESData data.

S103, HistoryNum=Number of data entries in HistoryList, ErrorNum=Number of data entries in ErrorList; HArray=100-element floating-point array; EArray=100-element floating-point array; DistanceMax==Maximum value of EDistance field of ErrorList;

S104, set the value of all elements of HArray to 0; set all elements of EArray to 0; sort HistoryList from small to large according to the value of HRQ; sort ErrorList from small to large according to the value of ERQ;

S105, history data counter HistoryNumCounter=1;

S106, HArray=HArray+(HistoryList[HistoryNumCounter].HSData)×(1+(HistoryNumCounter/HistoryNum-0.5) /100);

S107, HistoryNumCounter=HistoryNumCounter+1;

S108, if HistoryNumCounter>HistoryNum, go to S109, otherwise go to S106;

S109, HArray=HArray/HistoryNum;

S110, intrusion counter ErrorNumCounter=1;

S111, EArray=EArray+(ErrorList[ErrorNumCounter].ESData);

S112, ErrorNumCounter=ErrorNumCounter+1;

S113, if ErrorNumCounter>ErrorNum, go to S114, otherwise go to S111;

S114, EArray=EArray/ErrorNum.

Take the data of a cable tunnel in Jilin Province from 2018 to 2019 as an example

Enter the cable tunnel fiber optic sensor signal data list HistoryList in normal state. HistoryList contains 100 pieces of data. The contents are as follows:

Enter the data list ErrorList of the intrusion of the cable tunnel. The ErrorList contains 100 pieces of data. The content is as follows:

Get the number of HistoryList data entries HistoryNum=100

Create a normal state statistics array HArray=[0.21, 1.67, 1.23, 0.99, 1.38, 0.48, 1.11, 1.01, 1.18, 1.62, ......0.28, 0.28, 0.44, 0.00, 0.42, 0.25, 0.25, 0.39, 0.00, 2.40];

Get the number of ErrorList data entries ErrorNum=100;

Create an intrusion state statistics array EArray=[0.11, 1.49, 1.29, 1.25, 1.49, 0.48, 1.61,0.69, 1.25, 1.62, ......0.28, 0.28, 0.44, 0.00, 0.42, 0.25, 0.25, 0.39, 0.00, 1.18];

Get the maximum value of the invasion location DistanceMax=2349;

S2, establish an optical fiber signal abnormality encoding description operator OFeature, the input variable of OFeature is OFeatureInput, the type of OFeatureInput is a floating-point array of 100 elements, the output variable of OFeature is OFeatureOutput, and OFeatureOutput is a floating-point array of 400 elements.

S201 , an optical fiber signal abnormality encoding description operator OFeature is established, the input variable of OFeature is OFeatureInput, and the type of OFeatureInput is a floating-point array of 100 elements;

S202, establish a floating-point array of OFeature Output variable OFeatureOutput=400 elements, and set the value of all elements of the array to 0;

S203, the operator counter OFeatureCounter=1;

S204, if OFeatureInput[OFeatureCounter]>HArray[OFeatureCounter], then OFeatureOutput[OFeatureCounter]=1;

S205, if OFeatureInput[OFeatureCounter]>EArray[OFeatureCounter], then OFeatureOutput[OFeatureCounter +100]=1;

S206, the operator's first temporary storage variable OFeatureTemp1=0; the operator's second temporary storage variable OFeatureTemp1=0;

S207, OFeatureTemp1=abs(OFeatureInput[OFeatureCounter]-HArray[OFeatureCounter]), where abs is the absolute value of the calculation;

S208, OFeatureTemp2=abs(OFeatureInput[OFeatureCounter]-EArray[OFeatureCounter]), where abs is the calculated absolute value;

S209, if OFeatureTemp1>OFeatureTemp2, this OFeatureOutput[OFeatureCounter+200]=1;

S210, if abs(OFeatureTemp1-OFeatureTemp2)>0.5, OFeatureOutput[OFeatureCounter+300]=1, where abs is the calculated absolute value;

S211, OFeatureCounter=OFeatureCounter+1;

S212, if OFeatureCounter>100, go to S213, otherwise go to S204;

S213, take OFeatureOutput as the output of OFeature;

S3, use the optical fiber signal feature description operator OFeature to construct the decision information table DecisionTable, and use the DecisionTable to train the regression model RegModel;

S301, a decision information table DecisionTable is established, and the field structure of DecisionTable is as follows:

DecisionInput: decision input, a floating-point array of 400 elements;

DecisionDistance: The distance of the intrusion situation, its data type is floating-point data;

S302, the first variable of the decision counter CreateCounter1=1;

S303, the decision temporary storage array variable CTempArray= is calculated by OFeature, and the operator input OFeatureInput=HistoryList[CreateCounter1].HSData;

S304, create a new record in DecisionTable; in the newly added record, the value of the DecisionInput field is CTempArray, and the value of DecisionDistance is -1;

S305, CreateCounter1=CreateCounter1+1;

S306, if CreateCounter1>HistoryNum, go to S307, otherwise go to S303;

S307, the second variable of the decision counter CreateCounter2=1;

S308, CTempArray=Calculate by OFeature, operator input OFeatureInput=ErrorList[CreateCounter2].ESData;

S309, create a new record in DecisionTable; in the newly added record, the value of the DecisionInput field is CTempArray, and the value of DecisionDistance is ErrorList [CreateCounter2].EDistance/DistanceMax;

S310,CreateCounter2=CreateCounter2+1;

S311, if CreateCounter2>ErrorNum, go to S312, otherwise go to S308;

S312, establish a neural network regression analysis model RegModel;

S313, use the DecisionInput field of the DecisionTable as the input of the RegModel model, and use the DecisionDistance field of the DecisionTable as the output of the RegModel model; input all the data of the DecisionTable to train the RegModel;

S4, input the data CurrentArray collected by the optical fiber sensor signal in the cable tunnel, and predict the intrusion situation;

S401, input the data CurrentArray collected by the optical fiber sensor signal in the cable tunnel, and CurrentArray is a floating-point array with 100 elements;

S402, predicting the temporary storage feature variable Feature= is calculated by OFeature, and the operator input OFeatureInput=CurrentArray;

S403, the prediction result variable PResult= takes Feature as the input of RegModel and obtains the output of RegModel;

S404, if PResult<0, output "no intrusion situation" and go to S408, otherwise go to S405;

S405, the final predicted distance ResultDis=PResult×DistanceMax;

S406, output "there is an intrusion situation", and output ResultDis as the location of the intrusion;

S407, go to S409;

S408, HArray=(CurrentArray-HArray)/(HistoryNum×10)+HArray;

S409, the prediction process ends;

When inputting the collected data CurrentArray=[0.37, 1.68, 0.76, 0.83, 1.62, 0.16,1.63, 1.13, 0.70, 1.48, ......0.28, 0.28, 0.44, 0.00, 0.42, 0.25, 0.25, 0.39 , 0.00, 0.41];

Get PResult=-0.3 to output "no intrusion";

When inputting the collected data CurrentArray=[0.26, 1.30, 0.88, 0.74, 1.27, 0.49,1.40, 1.22, 0.73, 1.29, ......0.34, 0.54, 0.48, 0.24, 0.65, 0.28, 0.43, 0.62 , 0.00, 2.98];

Obtain PResult=0.834 and output "intrusion situation exists", and output ResultDis=0.834×2349=1959.1 as the intrusion position.

Example 2:

The method proposed by the present invention conducts actual operation test of a cable tunnel in a certain area of Jilin Province. The tunnel is put into use in 2016, and the actual operation data from 2016 to 2019 is verified, which is compared with the traditional neural network regression model and SVM regression model. Compared, the monitoring effect of the present invention is as follows:

Conclusion: From the above table, it can be seen that the monitoring accuracy of the traditional method continues to decrease with the passage of time, while the false alarm rate continues to increase. The detection accuracy of the method of the present invention is the highest among the three methods, and the false alarm rate is also lower.

Claims (1)

1. A cable tunnel intrusion monitoring method based on optical fiber sensor signals and abnormal codes comprises the following steps:

s1, inputting a cable tunnel optical fiber sensor signal data list HistoryList in a normal state, inputting a data list ErrorList in the case of intrusion of a cable tunnel, obtaining the number HistoryNum of HistoryList data entries, and establishing a normal state statistical array HARray; acquiring the number ErrorNum of ErrorList data entries, establishing an intrusion state statistical array EArray, and acquiring the maximum value DistanceMax of an intrusion position;

s101, inputting a cable tunnel optical fiber sensor signal data list HistoryList in a normal state, wherein the fields of the HistoryList comprise:

HSData: the data type of the normal state optical fiber sensor signal array is a floating point type array with 100 elements;

HRQ: the optical fiber sensor acquires time corresponding to the HSData data;

s102, inputting a data list errorlst of a cable tunnel intrusion condition, where fields of the errorlst include:

ESData: the data type of the signal array of the optical fiber sensor with the intrusion condition is a floating point type array with 100 elements;

EDistance: the distance between the position where the intrusion condition occurs in the tunnel and the optical fiber sensor, wherein the data type of the distance is floating point type data;

ERQ: the optical fiber sensor acquires time corresponding to the ESData data;

s103, the number of the data items of HistoryNum is equal to that of HistoryList, and the number of the data items of ErrorNum is equal to that of ErrorList; HArray is equal to a floating point type array of 100 elements; EArray is equal to a floating point type array of 100 elements; DistanceMax equals the maximum value of the EDistance field of ErrorList;

s104, setting the values of all elements of the HARRAY to be 0; setting all elements of EArray to 0; sorting the HistoryList according to the HRQ value from small to large; sorting ErrorList according to the value of ERQ from small to large;

s105, historical data counter HistoryNumCounter = 1;

S106，HArray=HArray+(HistoryList[HistoryNumCounter].HSData)×(1+ (HistoryNumCounter/HistoryNum-0.5) /100)；

S107， HistoryNumCounter=HistoryNumCounter+1；

s108, if HistoryNumCount > HistoryNum, then go to S109, otherwise go to S106;

S109，HArray=HArray/HistoryNum；

s110, an intrusion condition counter ErrorNumCount = 1;

S111，EArray=EArray+(ErrorList [ErrorNumCounter].ESData)；

S112，ErrorNumCounter=ErrorNumCounter+1；

s113, if the ErrorNumCount is greater than the ErrorNum, then the S114 is turned, otherwise, the S111 is turned;

S114，EArray=EArray/ErrorNum；

s2, establishing an optical fiber signal abnormal coding description operator OFeature, wherein the input variable of OFeature is OFeatureInput, the type of OFeatureInput is a floating-point array with 100 elements, the output variable of OFeature is OFeatureOutput, and OFeatureOutput is a floating-point array with 400 elements;

s201, establishing an optical fiber signal abnormal coding description operator OFeature, wherein the input variable of the OFeature is OFeatureInput, and the type of the OFeatureInput is a floating point type array with 100 elements;

s202, establishing an output variable OFeatureOutput of OFeature, wherein the type of the OFeatureOutput is a floating-point type array of 400 elements, and setting all element values of the array to be 0;

s203, the operator counter ofearturecounter = 1;

s204, if ofeactueinput [ ofeactuecounter ] > harrray [ ofeactuecounter ], ofeactueoutput [ ofeactuecounter ] = 1;

s205, if ofatutureinput [ ofatuturecounter ] > earrray [ ofatuturecounter ], then ofatutureoutput [ ofatuturecounter +100] = 1;

s206, the operator first temporary storage variable ofaturetemp 1= 0; the operator second temporary storage variable ofatuturetemp 1= 0;

s207, ofatuturetemp 1= abs (ofatutureinput [ ofatuturecounter ] -harrray [ ofatuturecounter ]), where abs is the calculated absolute value;

s208, ofatuturetemp 2= abs (ofatutureinput [ ofatuturecounter ] -earrray [ ofatuturecounter ]), where abs is the calculated absolute value;

s209, if OFeatureTemp1> OFeatureTemp2,

this ofeartureoutput [ ofearturecounter +200] = 1;

s210, if abs (ofaturetemp 1-ofaturetemp 2) >0.5, ofatureoutput [ ofaturecounter +300] =1, where abs is the calculated absolute value;

S211，OFeatureCounter=OFeatureCounter+1；

s212, go to S213 if OFeatureCounter >100, otherwise go to S204;

s213, using the OFeatureOutput as the output of the OFeature;

s3, constructing a decision information table DecisionTable by using an optical fiber signal anomaly code descriptor OFeature, and training a neural network regression analysis model RegModel by using the DecisionTable;

s301, establishing a decision information table decisionTable, wherein the field structure of the decisionTable is as follows:

DecisionInput: decision input, which is a floating-point array of 400 elements;

DecisionDistance: the distance of the intrusion condition occurs, and the data type of the distance is floating point type data;

s302, the decision counter first variable CreateCounter1= 1;

s303, calculating a decision temporary storage array variable ctemparay through ofearture, where an operator inputs ofeartureinput = HistoryList [ CreateCounter1]. HSData;

s304, newly creating 1 record in the decisionTable; in the newly added record, the value of the DecisionInput field is CTempower, and the value of the DecisionDistance is-1;

S305，CreateCounter1=CreateCounter1+1；

s306, if the CreateCounter1> HistoryNum, then go to S307, otherwise go to S303;

s307, the decision counter second variable CreateCounter2= 1;

s308, ctemaparay is calculated by ofature, operator input ofatureinput = errorlst [ CreateCounter2]. ESData;

s309, newly building 1 record in the decisionTable; in the newly added record, the value of the DecisionInput field is CTempray, and the value of the DecisionDistance is ErrorList [ CreateCounter2]. EDistance/DistanceMax;

S310，CreateCounter2=CreateCounter2+1；

s311, if CreateCounter2> ErrorNum, go to S312, otherwise go to S308;

s312, establishing a neural network regression analysis model RegModel;

s313, taking a decisionInput field of decisionTable as the input of the RegModel model, and taking a decisionDistance field of decisionTable as the output of the RegModel model; inputting all data of decisionTable to train a RegModel model;

s4, inputting data CurrentArray acquired by the optical fiber sensor signal in the cable tunnel, and predicting the invasion condition;

s401, inputting data CurrentArray acquired by optical fiber sensor signals in a cable tunnel, wherein the CurrentArray is a floating point type array with 100 elements;

s402, calculating a predicted temporary storage characteristic variable Feature through OFeature, and inputting OFeatureInput = CurrentArray by an operator;

s403, using Feature as input of the RegModel model and obtaining output of the RegModel model, and using the output of the RegModel model as a prediction result variable PResult;

s404, if PResult <0, outputting 'no intrusion condition exists' and turning to S408, otherwise, turning to S405;

s405, finally predicting the distance ResultDis = PResult × DistanceMax;

s406, outputting the intrusion condition and outputting ResultDis as the intrusion position;

s407, go to S409;

S408，HArray=(CurrentArray-HArray)/( HistoryNum×10)+HArray；

and S409, finishing the prediction process.