Power equipment anomaly detection method based on state information analysis

Abstract

The invention provides a power equipment anomaly detection method based on state information analysis. The power equipment anomaly detection method comprises the following steps: step (1), calling historical state information of power equipment; (2) constructing a state analysis model, and training the state analysis model according to the historical state information; (3) collecting actual state information of the power equipment; and step (4), performing power equipment abnormity judgment according to the state analysis model. The invention provides a power equipment anomaly detection method based on state information analysis, which can accurately extract feature information and further accurately judge whether the power equipment is abnormal or not.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of power detection, and particularly relates to a power equipment abnormality detection method based on state information analysis. BACKGROUND

[0002] Power equipment is subjected to electric, thermal and mechanical loads and the influence of natural environment during operation, and long-term work can cause aging, fatigue and wear, so that performance gradually decreases and reliability gradually decreases. The composition and structure of the insulating material of power equipment change under the long-term action of high voltage and high temperature, dielectric loss increases, and insulating performance decreases, finally leading to the destruction of insulating performance. In order to ensure the safe operation of the power system, the operation state of important power equipment of the system is monitored and detected. The purpose of monitoring is to discover the development of various degradation processes of the equipment in time, so as to repair and replace in time before a fault or performance decline occurs, and to avoid accidents endangering safety.

[0003] The application provides a power equipment abnormality detection method based on state information analysis, which can construct a state analysis model according to historical state information of power equipment, and train according to an output probability matrix, input actual state information of power equipment into the trained model, obtain an abnormal characteristic value, and then judge whether the power equipment is abnormal, so as to ensure the accuracy of the detection result. SUMMARY

[0004] The application provides a power equipment abnormality detection method based on state information analysis, which can accurately extract feature information and then accurately judge whether the power equipment is abnormal.

[0005] The application specifically provides a power equipment abnormality detection method based on state information analysis, which comprises the following steps:

[0006] Step (1): retrieve historical state information of the power equipment;

[0007] Step (2): construct a state analysis model, and train the state analysis model according to the historical state information;

[0008] Step (3): collect actual state information of the power equipment;

[0009] Step (4): perform power equipment abnormality judgment according to the state analysis model.

[0010] Further, the power equipment state information comprises working current.

[0011] Further, the specific method for constructing the state analysis model in step (2) is as follows:

[0012] the state analysis model parameter λ = [A, B, Π],

[0013] wherein A is a transition probability matrix, B is an output probability matrix, and Π is an initial probability distribution of states;

[0014] A = [a ij ] N×N , wherein a ij is a probability of a next time state transition to q i when the current state is q j , and N is a sampling number;

[0015] B = [b ik ] N×M , wherein b ik is a probability of an observation value r i when the current state is q k , and M is a data quantity of each sampling;

[0016] Π = [π i ] 1×N , wherein π i is a probability of an initial state q i .

[0017] Further, a specific method for training the state analysis model according to the historical state information is as follows:

[0018] (21) calculating an average value of each element in the output probability matrix,

[0019] (22) judging whether the average value of each element in the output probability matrix is within a reference range, if yes, corresponding λ * = [A * , B * , Π * ] is a training value of the state analysis model parameter.

[0020] Further, a specific method for judging power equipment abnormality according to the state analysis model in step (4) is as follows:

[0021] (41) calculating an abnormal feature value;

[0022] (42) judging whether the power equipment is abnormal according to the abnormal feature value.

[0023] Further, a specific method for calculating the abnormal feature value in step (41) is as follows:

[0024] calculating the abnormal feature value according to collected actual state information O = (o1, o2, ···, o T ):

[0025]

[0026] Where α t (i) is the forward variable, β t (i) is a backward variable.

[0027] Furthermore, the forward variable α t The algorithm for (i) is:

[0028] α t (i)=P(o1,o2,···,o t ,s t =q i |λ);

[0029] Furthermore, the backward variable β t The algorithm for (i) is:

[0030] β t (i)=P(o t+1 ,o t+2 ,···,o T |s t =q i ,λ).

[0031] Step (42) The specific method for determining whether there is an abnormality in the power equipment based on the abnormal characteristic value is as follows:

[0032] Determine whether the abnormal feature value is within the abnormal feature value reference range. If not, the power equipment is abnormal and an alarm is issued.

[0033] Compared with existing technologies, the beneficial effects are as follows: The power equipment anomaly detection method first constructs a state analysis model based on the historical state information of the power equipment, and trains it based on the output probability matrix. The actual state information of the collected power equipment is input into the trained model to obtain abnormal feature values, thereby determining whether there is an anomaly in the power equipment and ensuring the accuracy of the detection results. Attached Figure Description

[0034] Figure 1 This is a flowchart illustrating the workflow of a power equipment anomaly detection method based on state information analysis according to the present invention. Detailed Implementation

[0035] The following detailed description, in conjunction with the accompanying drawings, illustrates a specific implementation method for power equipment anomaly detection based on state information analysis according to the present invention.

[0036] like Figure 1 As shown, the power equipment anomaly detection method of the present invention includes the following steps:

[0037] Step (1): retrieve the power equipment historical state information, including working current, working temperature, etc.

[0038] Step (2): construct a state analysis model, and train the state analysis model according to the historical state information, specifically:

[0039] The state analysis model parameter λ = [A, B, Π],

[0040] Wherein, A is a transition probability matrix, B is an output probability matrix, and Π is an initial probability distribution of a state.

[0041] A = [a ij ] N×N , wherein a ij is the probability of the next state being transferred to q i when the current state is q j , and N is the number of sampling times.

[0042] B = [b ik ] N×M , wherein b ik is the probability of the observation value being r k when the current state is q i , and M is the number of sampling data each time.

[0043] Π = [π i ] 1×N , wherein π i is the probability of the initial state being q i .

[0044] The specific method for training the state analysis model according to the historical state information is:

[0045] (21) calculate the average value of each element in the output probability matrix;

[0046] (22) judge whether the average value of each element in the output probability matrix is within a reference range, if yes, the corresponding λ * = [A * , B * , Π * ] is the state analysis model parameter training value;

[0047] Step (3): collect the actual state information O = (o1, o2, ···, o T ) of the power equipment, and T is the sampling period.

[0048] Step (4): make an abnormality judgment of the power equipment according to the state analysis model, specifically:

[0049] (41) calculate an abnormal feature value, specifically:

[0050] calculating the abnormal feature value according to the collected actual state information:

[0051]

[0052] wherein α t (i) is a forward variable, and β t (i) is a backward variable.

[0053] The algorithm of the forward variable α t (i) is:

[0054] α t (i) = P (o1, o2, ···, o t | s t = q i , λ) ;

[0055] The algorithm of the backward variable β t (i) is:

[0056] β t (i) = P (o t+1 , o t+2 , ···, o T | s t = q i , λ) ;

[0057] (42) judging whether the power equipment is abnormal according to the abnormal feature value:

[0058] judging whether the abnormal feature value is in the abnormal feature value reference range, if not, the power equipment is abnormal, and an alarm is sent.

[0059] The application further provides a power equipment abnormality detection system based on state information analysis, comprising a signal collection module, a control processing module, a storage module, a display module and an alarm module,

[0060] The signal collection module collects state information of the power equipment.

[0061] The control processing module performs abnormality analysis and judgment on the power equipment according to the historical state information and the actual state information of the power equipment, and stores information through the storage module, displays the state through the display module, and alarms abnormities through the alarm module.

[0062] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application but not to limit the same. It should be understood by those skilled in the art that the specific embodiments of the present application can be modified or replaced by equivalents, and these modifications or replacements are within the scope of protection of the claims.

Claims

1. A method for detecting anomalies in power equipment based on state information analysis, characterized in that, The method for detecting abnormalities in power equipment includes the following steps: Step (1): Retrieve the historical status information of the power equipment; Step (2): Construct a state analysis model and train the state analysis model based on the historical state information; Step (3): Collect the actual status information of the power equipment; Step (4): Determine the abnormality of the power equipment based on the state analysis model.

2. The method for detecting power equipment anomalies based on state information analysis according to claim 1, characterized in that, The status information of the power equipment includes the operating current.

3. The method for detecting power equipment anomalies based on state information analysis according to claim 2, characterized in that, The specific method for constructing the state analysis model in step (2) is as follows: The state analysis model parameters λ = [A, B, Π], Where A is the transition probability matrix, B is the output probability matrix, and Π is the initial probability distribution of the state; Where a ij The current state is q i At that time, the state transitions to q in the next moment. j The probability, where N is the number of samples; B = [b] ik ] N×M , where b ik The current state is q i When the observed value is r k The probability of M is the number of data samples taken each time; Π=[π i ] 1×N , where π i The initial state is q i The probability of.

4. The method for detecting power equipment anomalies based on state information analysis according to claim 3, characterized in that, The specific method for training the state analysis model based on the historical state information is as follows: (21) Calculate the average value of each element in the output probability matrix; (22) Determine whether the average value of each element in the output probability matrix is ​​within the reference range. If so, the corresponding... These are the training values ​​for the parameters of the state analysis model.

5. The method for detecting power equipment anomalies based on state information analysis according to claim 4, characterized in that, The specific method for judging power equipment anomalies based on the state analysis model in step (4) is as follows: (41) Calculate the abnormal characteristic values; (42) Determine whether there is an abnormality in the power equipment based on the abnormal feature value.

6. The method for detecting power equipment anomalies based on state information analysis according to claim 5, characterized in that, The specific method for calculating the abnormal feature value in step (41) is as follows: Based on the collected actual state information O=(o1,o2,···,o T Calculate the abnormal feature value: Where α t (i) is the forward variable, β t (i) is a backward variable.

7. The method for detecting power equipment anomalies based on state information analysis according to claim 6, characterized in that, The forward variable α t The algorithm for (i) is: α t (i)=P(o1,o2,···,o t ,s t =q i |λ).

8. The method for detecting power equipment anomalies based on state information analysis according to claim 7, characterized in that, The backward variable β t The algorithm for (i) is: β t (i)=P(o t+1 ,o t+2 ,···,o T |s t =q i ,λ).

9. The method for detecting power equipment anomalies based on state information analysis according to claim 8, characterized in that, Step (42) The specific method for determining whether there is an abnormality in the power equipment based on the abnormal characteristic value is as follows: Determine whether the abnormal feature value is within the abnormal feature value reference range. If not, the power equipment is abnormal and an alarm is issued.

Images