Method and device for monitoring electrical fire and server

[0028] Example one

[0029] figure 1 It is a flowchart of the electrical fire monitoring method provided in the first embodiment of the present invention. Such as figure 1 As shown,

[0030] The electrical fire monitoring method provided in the first embodiment of the present invention includes:

[0031] Step S1: Obtain a plurality of detection data about electrical fire characteristic parameters of the circuit; wherein the electrical fire characteristic parameters include one or a combination of residual current parameters and environmental parameters, and the residual current parameters include residual current and residual current change rate And the residual current change amount, the environmental parameters include temperature, infrared intensity and smoke concentration.

[0032] Step S2: Fuse the detection data of the characteristic parameters of the electrical fire to obtain the monitoring result. Specifically, algorithms such as fuzzy reasoning or evidence theory can be used to perform data fusion on the detection data of electrical fire characteristic parameters. When using fuzzy inference for data fusion, first construct the membership function of each electrical fire characteristic parameter, and then create fuzzy rules. In actual operation, you can first use the membership function to fuzzify the measurement results of the corresponding electrical fire characteristic parameters, then use the traditional fuzzy inference method for fuzzy inference to obtain the inference result, and then use the center of gravity method to defuzzify the inference result. To obtain monitoring results. Compared with the judgment based on the effective value of the residual current alone, the present invention has higher detection accuracy and anti-interference ability due to the simultaneous fusion of multiple electrical fire characteristic parameters.

[0033] In some embodiments, after the monitoring result is obtained, if the monitoring result is to send an open flame or a smoldering fire, the decoupling device can be controlled to disconnect the circuit from the power source and send an alarm signal.

[0034] Example two

[0035] This embodiment is a further description of the method steps in the first embodiment. figure 2 It is a flowchart of obtaining residual current parameters provided in the second embodiment of the present invention. Such as figure 2 As shown, in this embodiment, the step of obtaining the residual current parameter includes:

[0036] Step S11, within a certain time window, collect the detection data of the residual current in the circuit at a certain time interval, and obtain the residual current time series I 1 , I 2 , I 3 ,..., I n-1;

[0037] Step S12: Calculate the residual current reference value I according to the residual current time series 0; Specifically, I 0 Can take the residual current time series I 1 , I 2 , I 3 ,..., I n-1 The mean of.

[0038] Step S13: Use the residual current transformer to detect the current residual current I in the circuit n;

[0039] Step S14, according to the current residual current I n And the residual current reference value I 0 Calculate the current residual current change ΔI and the residual current change rate K. Specifically, ΔI=I n -I 0 , K=ΔI/I 0.

[0040] Example three

[0041] This embodiment is a further description of the method steps in the first embodiment. Such as image 3 As shown, in this embodiment, the steps of fusing the detection data of electrical fire characteristic parameters to obtain the monitoring result include:

[0042] Step S21: Use the first neural network to fuse temperature, infrared intensity and smoke density detection data to obtain a basic probability assignment of environmental evidence. Specifically, the temperature, infrared intensity, and smoke concentration detection data can be input to the first neural network as feature vectors, and then the output of the first neural network can be normalized to obtain the basic probability assignment of environmental evidence.

[0043] Step S22: Use the second neural network to fuse the residual current, the residual current change rate and the residual current change amount to obtain the basic probability assignment of the current evidence. Specifically, the detection results of the residual current, the rate of change of the residual current, and the amount of change of the residual current can be input to the second neural network, and then the output of the second neural network is normalized to obtain the basic probability assignment of the current evidence.

[0044] Step S23, fusing the basic probability assignment of environmental evidence and the basic probability assignment of current evidence according to the evidence theory to obtain the monitoring result. Specifically, the fire type (open fire, no fire, smoldering fire) is used as the primitive of the evidence theory identification framework Θ, in 2 Θ The basic probability distribution function m is defined above, and there are:

[0045] m:2 Θ →[0,1]; (1)

[0046] Where m satisfies:

[0047] ①m(Φ)=0

[0048] ②

[0049] Among them, Φ is the empty set and A is 2 Θ Any subset of and represents a proposition, m(A) is the basic probability assignment of proposition A.

[0050] The formula for fusing multiple pieces of evidence is:

[0051]

[0052] among them, n is the number of evidences, this embodiment has current evidence and environmental evidence, so n=2;

[0053] The decision method of this embodiment is:

[0054] Assume Satisfy

[0055]

[0056]

[0057] If there is m(A 1 )-m(A 2 )> 0.3, then A 1 Is the decision result.

[0058] In this embodiment, the basic probability assignment of environmental evidence and the basic probability assignment of current evidence can be combined according to the above formula (2) to obtain the evidence fusion result, and then the monitoring result can be obtained according to the above decision method.

[0059] In this embodiment, the first and second neural networks may be BP neural networks. In other embodiments, it may also be other types of neural networks, such as convolutional neural networks.

[0060] In this embodiment, the steps for establishing a BP neural network are as follows:

[0061] A database is established based on the original detection data of various environmental parameters and residual current parameters obtained in the electrical fire experiment. The data in the database includes the data detected by the sensors to be tested and the type of fire (fire types can include open fire, no fire , Smoldering fire).

[0062] Set up a first neural network with environmental parameters as input, fire type as output, current parameters as input, and fire type as output to establish a second neural network, and initialize the first and second neural networks (including given learning Accuracy, number of iteration steps, learning parameters). For example, the residual current parameter includes three parameters: residual current, residual current change rate and residual current change. The fire type can include three types: open fire, no fire, and smoldering fire. The first neural network can have three layers of neurons, corresponding Ground, the input layer has three input neurons, the output layer has three output neurons, and the hidden layer can have 5 hidden neurons. Of course, the number of hidden neurons can be adjusted according to the actual operation effect.

[0063] Use the database to train the first neural network and the second neural network until the preset error requirements or the maximum number of iteration steps are met.

[0064] This embodiment combines residual current parameters and environmental parameters to determine electrical fires at the same time. Compared with the traditional method of simply using the effective value of residual current to determine electrical fires, this embodiment has higher detection accuracy and anti-interference ability.

[0065] It should be understood that the size of the sequence number of each step in the foregoing embodiment does not mean the order of execution, and the execution sequence of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.