A method, device and equipment for judging the degree of frost based on pulse neural network

Abstract

The present invention relates to the field of refrigeration equipment maintenance technology, and in particular to a method, device and equipment for judging the degree of frost based on a pulse neural network. The method comprises: performing a frame difference operation on an image of an air cooler to obtain a frame difference matrix; inputting the frame difference matrix into a pulse neural network, and the network calculating the average value of the pulse firing rate of each neuron in an excitation layer; then accumulating and calculating to obtain the accumulated value of the average pulse firing rate at the current moment, and inputting it into a frost degree judger, and obtaining the current degree of frost of the air cooler according to a preset frost degree interval; and accurately judging the current degree of frost of the air cooler within a preset time.

Description

Technical Field

[0001] The present invention relates to the technical field of refrigeration equipment maintenance, and in particular to a method, device and equipment for judging the degree of frost based on a pulse neural network. Background Art

[0002] After the air cooler has been running for a period of time, its evaporator will often be frosted, and the frosted surface will affect the cooling efficiency of the air cooler, so the air cooler needs to be defrosted.

[0003] However, it is difficult to accurately grasp the defrosting timing when defrosting is performed. It is easy to delay defrosting when the frosting is serious, resulting in a decrease in the cooling effect of the air cooler, or to defrost too early when the frosting is slight, resulting in low defrosting efficiency and increased energy consumption. The detection method using grayscale conversion and threshold segmentation in the prior art will have a lot of image noise and low accuracy, and it is difficult to judge the degree of frosting of the air cooler. Summary of the invention

[0004] The present application provides a method, device and equipment for judging the degree of frost based on a pulse neural network, which solves the problem of difficulty in judging the degree of frost of a cold air blower in the prior art.

[0005] The first aspect of the present application provides a method for judging the degree of frost based on a pulse neural network, comprising:

[0006] Perform frame difference operation on the air cooler image to obtain a frame difference matrix;

[0007] The frame difference matrix is ​​input into a pulse neural network, which includes: a coding layer and an excitation layer; the coding layer converts the frame difference matrix into a pulse sequence, and sends a pulse signal to the excitation layer according to the pulse sequence, and the excitation or inhibition type of the signal corresponds to the positive or negative value of each element of the frame difference matrix, and the network calculates the average value of the pulse firing rate of each neuron in the excitation layer;

[0008] The average pulse emission rate of the frame difference matrix at the current moment is added to the average pulse emission rate of the frame difference matrix within the preset time before the current moment to obtain the cumulative value of the average pulse emission rate at the current moment, and input it into the frost degree judgement device to obtain the current frost degree of the air cooler according to the preset frost degree range.

[0009] Optionally, the step of inputting the values ​​of each element of the frame difference matrix into the spiking neural network also includes:

[0010] The frame difference matrix is ​​identified as abnormal, and each element of the abnormal frame difference matrix is ​​corrected according to the frame difference matrix before the abnormality.

[0011] Optionally, the performing abnormality identification on the frame difference matrix is ​​specifically:

[0012] Calculate the mean of the frame difference matrix elements at the current moment and the previous moment respectively, and calculate the absolute value of the difference between the two means to obtain the absolute mean difference at the current moment;

[0013] Calculate the average of the absolute mean differences at all times to obtain the mean difference base value;

[0014] It is determined whether the absolute mean difference at the current moment is greater than r times the mean difference base value. If so, the frame difference matrix at the current moment is regarded as abnormal; r is a preset threshold coefficient.

[0015] Optionally, the step of correcting each element of the abnormal frame difference matrix according to the frame difference matrix before the abnormality is specifically as follows:

[0016] Obtain the data of the frame difference matrix before the time corresponding to the abnormal frame difference matrix, and obtain the value of each normal element;

[0017] According to the positive and negative values ​​of each normal element, the correction direction of the corresponding element position is obtained;

[0018] The mean difference basic value in the correction direction of the corresponding position is added to the normal element value of the corresponding position to obtain each correction element value, and replace it into each element position of the abnormal frame difference matrix.

[0019] Optionally, the neurons in the excitatory layer will send pulse signals only when the threshold membrane potential is reached, and the membrane potential of the neurons in the excitatory layer is set according to the conductance of the synapse using a kinetic equation.

[0020] The kinetic equation is specifically:

[0021]

[0022] Among them, τ is the membrane time constant, V is the neuron membrane potential, t is the biological cycle time, E rest is the resting potential, E exc is the equilibrium potential of the excitatory synapse, E inh is the equilibrium potential of inhibitory synapses, ge is the conductance of excitatory synapses, and g i is the inhibitory synaptic conductance, τg e and τg i are the time constants of excitatory and inhibitory postsynaptic potentials, respectively.

[0023] Optionally, before performing the frame difference operation on the air cooler image, the method further includes:

[0024] The image obtained by the air cooler camera is grayed out, and the grayscale value is averaged and pooled to scale the image to the size corresponding to the spiking neural network, so that the number of image pixels corresponds to the number of neurons in the spiking neural network layer.

[0025] The second aspect of the present application provides a frost degree judgment device based on a pulse neural network, comprising:

[0026] A frame difference module is used to perform a frame difference operation on the air cooler image, and form a frame difference matrix with the grayscale difference of each pixel;

[0027] A neural network module is used to input the frame difference matrix into a pulse neural network, and the pulse neural network includes: a coding layer and an excitation layer; the coding layer converts the frame difference matrix into a pulse sequence, and sends a pulse signal to the excitation layer according to the pulse sequence, and the excitation or inhibition type of the signal corresponds to the positive or negative of each element of the frame difference matrix, and the network calculates the average value of the pulse firing rate of each neuron in the excitation layer;

[0028] The frost degree judgment module is used to accumulate the average value of the pulse emission rate of the frame difference matrix at the current moment and the average value of the pulse emission rate of the frame difference matrix within a preset time before the current moment, to obtain the accumulated value of the average pulse emission rate at the current moment, and input it into the frost degree judgement device to obtain the current frost degree of the air cooler according to the preset frost degree range.

[0029] Optionally, in the spiking neural network module, the step of inputting the values ​​of each element of the frame difference matrix into the spiking neural network also includes:

[0030] The frame difference matrix is ​​identified as abnormal, and each element of the abnormal frame difference matrix is ​​corrected according to the frame difference matrix before the abnormality.

[0031] Optionally, in the frame difference module, before performing the frame difference operation on the air cooler image, the method further includes:

[0032] The image obtained by the air cooler camera is grayed out, and the grayscale value is averaged and pooled to scale the image to the size corresponding to the spiking neural network, so that the number of image pixels corresponds to the number of neurons in the spiking neural network layer.

[0033] The third aspect of the present application provides a frost degree judgment device based on a pulse neural network, the device comprising a processor and a memory:

[0034] The memory is used to store program code and transmit the program code to the processor;

[0035] The processor is used to execute the frost degree judgment method based on the pulse neural network as described in any one of the first aspects of the present application according to the instructions in the program code.

[0036] The present application provides a method for judging the degree of frost based on a pulse neural network. The method comprises the following steps: inputting a frame difference matrix of an air cooler image into a pulse neural network, wherein the excitation or inhibition type of a signal corresponds to the positive or negative value of each element of the frame difference matrix; then calculating the pulse firing rate of neurons in the excitation layer and taking an average value to obtain the pulse firing rate average value of the frame difference matrix; accumulating the pulse firing rate average values ​​of multiple frame difference matrices at the current moment and within a preset time before, and inputting the obtained cumulative value of the average pulse firing rate at the current moment into a frost judgement device to obtain the degree of frost of the current air cooler; using a pulse neural network to accurately quantify the frosting condition of the air cooler, using the excitation and inhibition of pulse signals to reflect the condensation or ablation of the frosting condition, and using the calculated average value of the pulse firing rate of the frame difference matrix to reflect the overall frosting change within the frame difference time, and using the current cumulative value of the average pulse firing rate of the accumulated pulse firing rate average value, and through the preset interval of the frost judgement device, accurately judging the degree of frost of the current air cooler within the preset time. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0038] Figure 1 A schematic diagram of the process of the frost degree determination method based on the pulse neural network provided in this application;

[0039] Figure 2 A schematic diagram of the process of obtaining the average value of the pulse emission rate of the frost degree judgment method based on the pulse neural network provided in the present application;

[0040] Figure 3 Schematic diagram of the pulse neural network structure provided for this application;

[0041] Figure 4 A schematic diagram of the frame difference operation flow of the frost degree judgment method based on the pulse neural network provided in this application;

[0042] Figure 5 A schematic diagram of the overall process of the frost degree judgment method based on the pulse neural network provided in this application;

[0043] Figure 6 This is a schematic diagram of the structure of a frost degree judgment device based on a pulse neural network provided in this application. DETAILED DESCRIPTION

[0044] In order to make the purpose, features and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0045] The present application provides a method for judging the degree of frost based on a pulse neural network, which solves the problem of difficulty in judging the degree of frost of an air cooler in the prior art.

[0046] See also Figure 1 , Figure 1 A schematic flow chart of a method for determining the degree of frost based on a pulse neural network provided in this application.

[0047] A first aspect of the present embodiment provides a method for determining the degree of frost based on a pulse neural network, comprising:

[0048] S100, performing a frame difference operation on the air cooler image to obtain a frame difference matrix;

[0049] It should be noted that, in this embodiment, images of the air cooler are captured at regular intervals, and the grayscale values ​​of each pixel of the image of adjacent frames are subtracted to obtain a frame difference matrix using the grayscale differences.

[0050] S200, inputting the frame difference matrix into a pulse neural network, the pulse neural network comprising: a coding layer and an excitation layer; the coding layer converts the frame difference matrix into a pulse sequence, and sends a pulse signal to the excitation layer according to the pulse sequence, the excitation or inhibition type of the signal corresponds to the positive or negative of each element of the frame difference matrix, and the network calculates the average value of the pulse firing rate of each neuron in the excitation layer;

[0051] It should be noted that each element value in the frame difference matrix is ​​the difference in the grayscale value of each pixel point in adjacent frames, which represents the change in the frost condition of the air cooler position corresponding to the pixel point in the captured air cooler image; the greater the change in the frost condition at a point, the greater the absolute value of the corresponding element value, and the positive and negative correspond to the condensation and melting of the frost surface;

[0052] The pulse signal sent by the coding layer to the excitation layer, the excitation signal or the inhibition signal corresponds to the condensation and ablation of the frost surface at that location of the air cooler. Furthermore, the average value of the pulse firing rate of each neuron in the excitation layer calculated by the pulse neural network reflects the overall frost change of the air cooler within the frame difference time.

[0053] S300, adding the average pulse emission rate of the frame difference matrix at the current moment to the average pulse emission rate of the frame difference matrix within a preset time before the current moment, obtaining the current average pulse emission rate cumulative value, and inputting it into the frost degree determiner, and obtaining the current frost degree of the air cooler according to the preset frost degree range;

[0054] It should be noted that the accumulated value of the average pulse emission rate at the current moment can reflect the increase or decrease of the overall frosting of the air cooler within the preset time; the accumulated preset time can be set by the staff according to the actual use of the air cooler;

[0055] The frost degree judgement device can accurately identify the overall accumulated frost amount within the preset time according to the preset frost degree range of the accumulated value of the average pulse emission rate, and obtain the current frost degree of the air cooler. The staff can then defrost according to the frost degree.

[0056] In this embodiment, the frame difference matrix of the air cooler image is input into the pulse neural network, and the excitation or inhibition type of the signal corresponds to the positive or negative value of each element of the frame difference matrix; the pulse firing rate of the excitation layer neurons is calculated and averaged to obtain the pulse firing rate average value of the frame difference matrix, and the pulse firing rate average values ​​of multiple frame difference matrices at the current moment and within the previous preset time are accumulated, and the accumulated value of the current average pulse firing rate is input into the frost judgement device to obtain the current degree of frost of the air cooler; the pulse neural network is used to accurately quantify the frosting condition of the air cooler, and the excitation and inhibition of the pulse signal reflect the condensation or ablation of the frost condition, and the calculated frame difference matrix pulse firing rate average value reflects the overall frost change within the frame difference time, and the current moment average pulse firing rate cumulative value of the accumulated pulse firing rate average value, and through the preset interval of the frost judgement device, the current degree of frost of the air cooler within the preset time is accurately judged, which is beneficial for the staff to make defrosting judgments.

[0057] The above is a detailed description of the first embodiment of a method for determining the degree of frost based on a pulse neural network provided by the present application. The following is a detailed description of the second embodiment of a method for determining the degree of frost based on a pulse neural network provided by the present application.

[0058] See also Figure 2 This embodiment provides a method for judging the degree of frost based on a pulse neural network. In step S200 of the aforementioned embodiment, a frame difference matrix is ​​input into a pulse neural network, and the pulse neural network includes: a coding layer and an excitation layer; the coding layer converts the frame difference matrix into a pulse sequence, and sends a pulse signal to the excitation layer according to the pulse sequence. The excitation or inhibition type of the signal corresponds to the positive or negative value of each element of the frame difference matrix. The network calculates the average value of the pulse firing rate of each neuron in the excitation layer, which specifically includes:

[0059] S210, performing abnormality identification on the frame difference matrix, and correcting each element of the abnormal frame difference matrix according to the frame difference matrix before the abnormality.

[0060] It should be noted that when shooting the image of the air cooler, there may be changes in the lighting of the air cooler, or fogging, which will seriously affect the acquisition of the air cooler image and cause the image to be seriously distorted. Correspondingly, in the frame difference matrix, the element values ​​will be abnormal, and the above abnormal conditions will cause the entire picture to be distorted, rather than individual element abnormalities. Therefore, we can identify the average value of the elements of the frame difference matrix to determine whether there is an abnormality.

[0061] Calculate the mean of the frame difference matrix elements at the current moment and the previous moment respectively, that is, sum and average the elements in the frame difference matrix, and calculate the absolute value of the difference between the two means to obtain the absolute mean difference at the current moment; the corresponding absolute mean difference formula is as follows:

[0062] MS t =|M t -M t-1 |

[0063] Among them, M t and M t-1 are the mean values ​​of the frame difference matrix elements at the current moment and the previous moment, MS t is the absolute mean difference at the current moment.

[0064] Calculate the average of the absolute mean differences at all times, that is, sum up the absolute mean differences of all historical and current times and calculate the average value to obtain the mean difference basic value; the calculation formula for the mean difference basic value is as follows:

[0065]

[0066] Among them, AVGMS t is the mean difference base value.

[0067] It is determined whether the absolute mean difference at the current moment is greater than r times the mean difference base value. If so, the frame difference matrix at the current moment is regarded as abnormal; r is a preset threshold coefficient, and the judgment formula is specifically as follows:

[0068] MS t ≥r×AVGMS t

[0069] Furthermore, the threshold coefficient r can be set according to the actual operation of the air cooler and the camera imaging conditions. In this embodiment, r is set to 2.5; and to ensure the effect of abnormal recognition, the staff needs to ensure that the image acquisition of the first frame is normal when the air cooler camera is turned on.

[0070] For the repair of the abnormal frame difference matrix, the data of the frame difference matrix before the corresponding time of the abnormal frame difference matrix is ​​obtained to obtain the values ​​of each normal element; that is, if the abnormal frame difference matrix is ​​at time t, the frame difference matrix at time t-1 is obtained, and the values ​​of each element thereof are used as normal element values; at the same time, the correction direction of the corresponding element position is obtained according to the positive and negative values ​​of each normal element, that is, if the normal element value is positive, the correction direction is to multiply the mean difference basic value by 1, and if the normal element value is negative, the correction direction is to multiply the mean difference basic value by -1;

[0071] The mean difference base value under the correction direction of the corresponding position is added to the normal element value of the corresponding position to obtain each correction element value, and replace it into each element position of the abnormal frame difference matrix. The corresponding correction value calculation formula is:

[0072] Correction value = frame difference matrix element value before the abnormality + correction direction × AVGMS t

[0073] Furthermore, this embodiment does not need to replace all element values ​​of the abnormal frame difference matrix. Before replacing, it can be determined whether the element value is greater than AVGMS. t ×r, if yes, then replace it, if no, then keep the element value.

[0074] S220, input each element value of the frame difference matrix into the pulse neural network coding layer, and the coding layer sends a pulse signal to the excitation layer.

[0075] Please note that, see Figure 3 , Figure 3 Schematic diagram of the structure of a pulse neural network. The pulse neural network includes: a coding layer and an excitation layer.

[0076] The element values ​​of the frame difference matrix are input into the encoding layer in the pulse neural network. The encoding layer performs Poisson encoding on the grayscale data in the form of frequency and converts the element value data into a corresponding pulse sequence. The excitation or inhibition type of the signal corresponds to the positive or negative value of each element of the frame difference matrix, completing the transformation of the input data from numerical value to pulse form.

[0077] The coding formula for Poisson coding of the frame difference matrix by the coding layer is as follows:

[0078]

[0079] Among them, P T is the probability of a neuron generating a pulse, p is the pixel value, k is the number of neurons, and T is the period of a pulse train.

[0080] The neurons in the coding layer are fully synaptic connected to the neurons in the excitatory layer, that is, each neuron in the coding layer is connected to all neurons in the excitatory layer; the neurons in the excitatory layer will only send pulse signals when the threshold membrane potential is reached, and the membrane potential of the neurons in the excitatory layer is set according to the conductance of the synapse using the kinetic equation;

[0081] The kinetic equation is specifically:

[0082]

[0083] Among them, τ is the membrane time constant, V is the neuron membrane potential, t is the biological cycle time, E rest is the resting potential, E exc is the equilibrium potential of the excitatory synapse, E inh is the equilibrium potential of inhibitory synapses, ge is the conductance of excitatory synapses, and g i is the inhibitory synaptic conductance, τg e and τg i are the time constants of excitatory and inhibitory postsynaptic potentials, respectively.

[0084] In this embodiment, the excitatory layer neurons adopt the LIF pulse neuron model. When the neurons receive a single input, the membrane potential voltage will leak and gradually return to the resting state. Therefore, combined with the dynamic equation, the excitatory layer can filter out the noise existing after the image frame difference, that is, the sudden isolated change of the gray value of individual pixels will not be regarded as a change in frosting, but will be filtered out as noise, which is more conducive to accurately calculating the degree of frost.

[0085] Furthermore, the spiking neural network can also include an inhibition layer, which introduces a lateral inhibition mechanism into the network. The lateral inhibition mechanism sends an inhibitory signal to the excited neurons that have not yet fired pulses, reducing the membrane potential of these neurons and making it more difficult to fire pulses. On the one hand, the lateral inhibition mechanism inhibits the inactive neuron population, and on the other hand, it reduces the impact of interference signals.

[0086] S230, the network calculates the average value of the pulse firing rate of each neuron in the excitatory layer.

[0087] It should be noted that the pulse cycle after the input pulse sequence is set, the number of pulse signals emitted by each neuron in the excitation layer after the pulse neural network runs a pulse sequence cycle is counted, and the emission frequency is calculated based on the number of emitted pulses and the pulse sequence cycle to obtain the pulse emission rate of each neuron, and then the pulse emission rate of each neuron is summed to calculate the average value. In this embodiment, the pulse sequence cycle T is set to 2000. The specific calculation formula of the pulse emission rate average value is:

[0088]

[0089] Among them, O(t) is the average pulse firing rate of each neuron at the current time j, k is the number of neurons, N i t is the pulse firing rate of the ith neuron at time t.

[0090] In this embodiment, the abnormalities of the frame difference matrix are identified and corrected, so that the abnormalities of the input image caused by lighting or fogging are reduced, and the noise in the image is filtered out by the LIF neuron model of the excitation layer and the dynamic equation. Finally, the changes in the overall frost condition of the air cooler corresponding to the frame difference matrix are accurately reflected through the average value of the neuron pulse firing rate.

[0091] The above is a detailed description of the second embodiment of a method for determining the degree of frost based on a pulse neural network provided by the present application. The following is a detailed description of the third embodiment of a method for determining the degree of frost based on a pulse neural network provided by the present application.

[0092] See also Figure 4 This embodiment provides a method for judging the degree of frost based on a pulse neural network. In step S100 of the aforementioned embodiment, a frame difference operation is performed on the air cooler image to obtain a frame difference matrix including:

[0093] S110, grayscale the image acquired by the air cooler camera, and scale the image to a size corresponding to the pulse neural network by average pooling the grayscale values, so that the number of image pixels corresponds to the number of neurons in the pulse neural network layer.

[0094] It should be noted that after the air cooler image captured by the air cooler camera is grayed out, average pooling is performed, that is, the grayscale values ​​of multiple pixels are averaged and then the average value is put back into each pixel to make the values ​​in the pool consistent, thereby achieving scaling of the grayscale image, so that the pixel position and number of the image can match the neuron setting of the pulse neural network, which is convenient for the corresponding input pulse sequence in the subsequent steps.

[0095] S120, subtracting the grayscale values ​​of corresponding pixels of the current air cooler image from those of the previous air cooler image in time sequence, and using each grayscale difference as an element of a corresponding position in a matrix to obtain a frame difference matrix.

[0096] It should be noted that after obtaining two air cooler images of adjacent frames, the grayscale values ​​of the air cooler images are substituted into the frame difference operation model to obtain the element values ​​of the corresponding frame difference matrix of each pixel point; the frame difference operation model is specifically:

[0097] D xy =f t (x,y)-f t-1 (x,y)

[0098] Among them, f t (x,y) and f t-1 (x, y) are the grayscale values ​​of the pixel with coordinates (x, y) in the air cooler image at time t and t-1 respectively, D xy is the element value of row x and column y in the frame difference matrix.

[0099] Furthermore, at certain moments, small areas on certain cooling fans may melt by themselves due to temperature changes, and the positive and negative values ​​of the elements of the frame difference matrix correspond to the condensation or melting of the frost surface in the area.

[0100] For further information, see Figure 5 , Figure 5 A schematic diagram of the overall process of the method for determining the degree of frost based on a pulse neural network provided in this application. Figure 5 The other steps in the above-mentioned embodiment can refer to the corresponding process, which will not be described here. However, for step S300, it should be noted that the calculation formula of the cumulative value of the average pulse emission rate at the current moment is specifically:

[0101]

[0102] Among them, G(t) is the cumulative value of the average pulse emission rate at the current moment, O(t) is the average value of the pulse emission rate at time t, that is, the average value of the pulse emission rate at the current moment;

[0103] Furthermore, the frost degree determiner uses a double threshold value to form a frost degree interval, and divides three intervals according to the two threshold values, corresponding to light frost, moderate frost and heavy frost, respectively. The threshold value can be set according to the actual use of the air cooler. The frost degree intervals set in this embodiment are specifically:

[0104]

[0105] In this embodiment, the image of the air cooler is preprocessed by grayscale processing and average pooling, so that the number of pixels corresponds to the number of neurons in the neural network, and then a frame difference operation is performed to obtain a frame difference matrix. The value of each element of the frame difference matrix reflects the frosting condition at the corresponding position of the air cooler. The average value of the neuron pulse firing rate is used to reflect the overall frosting condition within one frame time. Finally, the cumulative value of the average pulse firing rate is used to reflect the accurate frosting condition of the frosted surface of the air cooler when it accumulates to the current moment. A frost degree judgement is used to accurately obtain the frost degree of the air cooler, so as to guide the staff to perform defrosting work, make the defrosting timing accurate, and ensure the reliable operation of the air cooler.

[0106] The above is a detailed description of the third embodiment of a method for determining the degree of frost based on a pulse neural network provided in the present application. The following is a detailed description of a device for determining the degree of frost based on a pulse neural network provided in the second aspect of the present application.

[0107] See also Figure 6 This embodiment provides a frost degree judgment device based on a pulse neural network, comprising:

[0108] A frame difference module 10 is used to perform a frame difference operation on the air cooler image, and form a frame difference matrix with the grayscale difference of each pixel;

[0109] The neural network module 20 is used to input the frame difference matrix into the pulse neural network, and the pulse neural network includes: a coding layer and an excitation layer; the coding layer converts the frame difference matrix into a pulse sequence, and sends a pulse signal to the excitation layer according to the pulse sequence, and the excitation or inhibition type of the signal corresponds to the positive or negative of each element of the frame difference matrix, and the network calculates the average value of the pulse firing rate of each neuron in the excitation layer;

[0110] The frost degree judgment module 30 is used to accumulate the average pulse emission rate of the frame difference matrix at the current moment and the average pulse emission rate of the frame difference matrix within a preset time before the current moment, to obtain the accumulated value of the average pulse emission rate at the current moment, and input it into the frost degree judgement device to obtain the current frost degree of the air cooler according to the preset frost degree range.

[0111] Furthermore, in the neural network module 20, the inputting of the values ​​of each element of the frame difference matrix into the pulse neural network also includes:

[0112] The frame difference matrix is ​​identified as abnormal, and each element of the abnormal frame difference matrix is ​​corrected according to the frame difference matrix before the abnormality.

[0113] Furthermore, in the frame difference module 10, before performing the frame difference operation on the air cooler image, the frame difference module 10 further includes:

[0114] The image obtained by the air cooler camera is grayed out, and the grayscale value is averaged and pooled to scale the image to the size corresponding to the spiking neural network, so that the number of image pixels corresponds to the number of neurons in the spiking neural network layer.

[0115] The third aspect of the present application also provides a frost degree judgment device based on a pulse neural network, including a processor and a memory: the memory is used to store program code and transmit the program code to the processor; the processor is used to execute the above-mentioned frost degree judgment method based on a pulse neural network according to the instructions in the program code.

[0116] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described devices and equipment can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0117] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0118] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0119] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0120] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0121] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for judging the frosting degree based on a spiking neural network, characterized in that, it includes: Performing frame difference operation on the cold air blower image to obtain a frame difference matrix; Inputting the frame difference matrix into a spiking neural network, the spiking neural network includes: an encoding layer and an excitatory layer; the encoding layer converts the frame difference matrix into a pulse sequence and sends pulse signals to the excitatory layer according to the pulse sequence, and the excitatory or inhibitory type of the signal corresponds to the positive and negative of each element of the frame difference matrix, and the network calculates the average value of the pulse firing rates of each neuron in the excitatory layer; Adding the average value of the pulse firing rate of the frame difference matrix at the current moment to the average value of the pulse firing rate of the frame difference matrix within a preset time before the current moment to obtain the cumulative value of the average pulse firing rate at the current moment, and inputting it into the frosting degree judgment device, and obtaining the frosting degree of the current cold air blower according to the preset frosting degree interval; Before inputting the values of each element of the frame difference matrix into the spiking neural network, it further includes: Performing anomaly recognition on the frame difference matrix, and correcting each element of the abnormal frame difference matrix according to the frame difference matrix before the anomaly; The performing anomaly recognition on the frame difference matrix specifically is: Respectively calculating the mean value of the elements of the frame difference matrix at the current moment and the previous moment, and calculating the absolute value of the difference between the two mean values to obtain the absolute mean difference at the current moment; Calculating the average value of the absolute mean differences at all moments to obtain the mean difference base value; Judging whether the absolute mean difference at the current moment is greater than r times the mean difference base value, if so, regarding the frame difference matrix at the current moment as abnormal; the r is a preset threshold coefficient; The neurons in the excitatory layer will send pulse signals only when they reach the threshold membrane potential, and the membrane potential of the neurons in the excitatory layer is set according to the synaptic conductance by a kinetic equation, The kinetic equation specifically is: Among them, τ is the membrane time constant, V is the neuron membrane potential, t is the biological cycle time, E rest is the resting potential, E exc is the equilibrium potential of the excitatory synapse, E inh is the equilibrium potential of inhibitory synapses, g e is the excitatory synaptic conductance, g i is the inhibitory synaptic conductance, τg e and τg i are the time constants of excitatory and inhibitory postsynaptic potentials, respectively.

2. The method for judging the frosting degree based on a spiking neural network according to claim 1, characterized in that, The correcting each element of the abnormal frame difference matrix according to the frame difference matrix before the anomaly specifically is: Obtaining the data of the frame difference matrix before the corresponding moment of the abnormal frame difference matrix to obtain the values of each normal element; Obtaining the correction direction of the corresponding element position according to the positive and negative of each normal element value; Adding the mean difference base value in the corresponding correction direction to the normal element value at the corresponding position to obtain the values of each corrected element, and replacing them into the positions of each element of the abnormal frame difference matrix.

3. The method for judging the frosting degree based on a spiking neural network according to claim 1, characterized in that, Before performing the frame difference operation on the cold air blower image, it further includes: Performing grayscale processing on the image obtained by the cold air blower camera, and performing average pooling on the grayscale values to scale the image to the size corresponding to the spiking neural network, so that the number of image pixels corresponds to the number of neurons in the spiking neural network layer.

4. A device for judging the frosting degree based on a spiking neural network, characterized in that, it includes: A frame difference module, configured to perform a frame difference operation on the cold air blower image to form a frame difference matrix with the grayscale differences of each pixel point; A neural network module is used to input the frame difference matrix into a pulse neural network, and the pulse neural network includes: a coding layer and an excitation layer; the coding layer converts the frame difference matrix into a pulse sequence, and sends a pulse signal to the excitation layer according to the pulse sequence, and the excitation or inhibition type of the signal corresponds to the positive or negative of each element of the frame difference matrix, and the network calculates the average value of the pulse firing rate of each neuron in the excitation layer; A frost degree judgment module is used to accumulate the average pulse emission rate of the frame difference matrix at the current moment and the average pulse emission rate of the frame difference matrix within a preset time before the current moment, obtain the current average pulse emission rate cumulative value, and input it into the frost degree judgement device, and obtain the current frost degree of the air cooler according to the preset frost degree interval; In the pulse neural network module, the value of each element of the frame difference matrix is ​​input into the pulse neural network, and before that, it also includes: Anomalies are identified on the frame difference matrix, and each element of the abnormal frame difference matrix is ​​corrected according to the frame difference matrix before the abnormality; The abnormality identification of the frame difference matrix is ​​specifically as follows: Calculate the mean of the frame difference matrix elements at the current moment and the previous moment respectively, and calculate the absolute value of the difference between the two means to obtain the absolute mean difference at the current moment; Calculate the average of the absolute mean differences at all times to obtain the mean difference base value; Determine whether the absolute mean difference at the current moment is greater than r times the mean difference base value, and if so, regard the frame difference matrix at the current moment as abnormal; wherein r is a preset threshold coefficient; The neurons in the excitatory layer will send pulse signals only when the threshold membrane potential is reached, and the membrane potential of the neurons in the excitatory layer is set according to the conductance of the synapse using a kinetic equation. The kinetic equation is specifically: Among them, τ is the membrane time constant, V is the neuron membrane potential, t is the biological cycle time, E rest is the resting potential, E exc is the equilibrium potential of the excitatory synapse, E inh is the equilibrium potential of inhibitory synapses, g e is the excitatory synaptic conductance, g i is the inhibitory synaptic conductance, τg e and τg i are the time constants of excitatory and inhibitory postsynaptic potentials, respectively.

5. The frost degree judgment device based on pulse neural network according to claim 4, It is characterized in that In the frame difference module, before performing the frame difference operation on the cooling fan image, the method further includes: The image obtained by the air cooler camera is grayed out, and the grayscale value is averaged and pooled to scale the image to the size corresponding to the spiking neural network, so that the number of image pixels corresponds to the number of neurons in the spiking neural network layer.

6. A device for judging the degree of frost based on a pulse neural network. It is characterized in that The device comprises a processor and a memory: The memory is used to store program code and transmit the program code to the processor; The processor is used to execute the frost degree judgment method based on pulse neural network according to any one of claims 1-3 according to the instructions in the program code.

Images