Cloud online central air conditioner fault analysis system
Technical Field
The invention relates to the field of fault detection of central air conditioners, in particular to a cloud online fault analysis system of a central air conditioner.
Background
The central air conditioning system is composed of one or more cold and heat source systems and a plurality of air conditioning systems, which are different from the conventional refrigerant type air conditioner (such as a single machine, VRV) for intensively processing air to achieve comfort requirements. The liquid gasification refrigeration principle is adopted to provide the required cooling capacity for the air conditioning system so as to offset the heat load of the indoor environment; the heating system provides the air conditioning system with the required heat to counteract the indoor environmental cooling and heating load. The refrigeration system is a vital part of the central air conditioning system, and the type, the operation mode, the structural form and the like of the refrigeration system directly influence the economy, the high efficiency and the rationality of the central air conditioning system in operation.
The central air conditioning system mainly comprises a refrigerator, a cooling water circulation system, a chilled water circulation system, a fan coil system and a cooling water tower. The refrigerating machine compresses the refrigerant into liquid state through the compressor, then sends the liquid state into the evaporator to exchange heat with the chilled water, the chilled water is refrigerated, the chilled water is sent into the cooling coils in the fans by the chilled water pump, and the purpose of cooling is achieved by blowing cold air by the fans. The refrigerant releases heat in the condenser after evaporation, exchanges heat with cooling circulating water, the cooling water pump pumps the cooling water with heat to the cooling water tower, the water tower fan sprays and cools the cooling water to exchange heat with the atmosphere, and the heat is dissipated to the atmosphere.
The existing cloud online central air conditioner fault analysis system has the problem that an alarm control end has abnormal control of unauthorized human factors, and has poor confidentiality. The invention provides a cloud online central air conditioner fault analysis system which is used for solving the technical problems.
Disclosure of Invention
The invention aims to solve the technical problem that in the prior art, the cloud online central air conditioner fault analysis system is low in security. The novel cloud online central air conditioner fault analysis system has the characteristic of high confidentiality.
In order to solve the technical problems, the technical scheme adopted is as follows:
a cloud online central air conditioner fault analysis system, the cloud online central air conditioner fault analysis system comprising:
The cloud online central air conditioner fault analysis system comprises: the cooling water circulation system consists of a cooling pump, a cooling water pipe and a cooling tower; the system also comprises a plurality of fault parameter acquisition units which are connected with the refrigerator, the freezing pump, the freezing water pipe, the cooling pump, the cooling water pipe, the cooling tower and the fan coil system in a one-to-one correspondence manner; the data modeling analysis module is connected with the parameter sensor, is connected to the network cloud processor through the blockchain sub-node, and is connected with the warning controller through the blockchain network, and the warning controller outputs a fault warning signal and controls the central air conditioner;
The warning controller comprises an authentication monitoring unit, wherein the authentication monitoring unit comprises a built-in first memory, a built-in receiving unit, a built-in processing unit and an identity characteristic acquisition unit, wherein the built-in receiving unit, the built-in processing unit and the identity characteristic acquisition unit are connected with the first memory, and the built-in processing unit is connected with the output unit to the warning controller;
The built-in processing unit is used for carrying out consistency initial analysis on the characteristic data of the real-time operator acquired by the identity characteristic acquisition unit, and continuously carrying out operator consistency continuous analysis on the behavior characteristic data of the operator operation warning controller continuously received by the built-in receiving unit;
The built-in processing unit judges the initial identity consistency of the real-time operator and the built-in registration operator according to the consistency initial analysis result, and judges the continuous identity consistency of the identities of the real-time operator and the built-in registration operator according to the operator consistency continuous analysis result;
the identity identification code of decryption processing is output through the output unit when the initial identity is consistent with the continuous identity;
And the warning controller receives the identity identification code, compares the identity identification code with the identity of the registered operator, and completes identity consistency authentication of the on-line operator and the off-line operator.
The working principle of the application is as follows: the application aims to solve the problems of authority allocation and confidentiality of manual control in a cloud online central air conditioner fault analysis system. Authority identity consistency authentication is set and is divided into initial identity authentication and subsequent authentication in the use process, and biological characteristics of a user can be collected for consistency continuous analysis and authentication during initial authentication. But there are cases where the registered user identity is the same person during subsequent use. At this time, the present application confirms whether the same person is still not present by continuously receiving the behavior characteristics of the warning controller and performing the analysis and comparison internally. In the process, the identity comparison of the control operator is in the system, and the system and the network interact with the identity identification code after decryption, so that the risk of easy leakage of personal information stored in the network is prevented.
In the above scheme, for optimization, further, the built-in processing unit further continuously performs control policy normality analysis on the control policy behavior of the network cloud processor. For the automatic control of the network cloud processor, the method can adopt an analysis method consistent with the consistency continuous analysis to carry out similarity analysis on the control strategy and the history strategy of the network cloud processor, so as to ensure that no control abnormality occurs.
Further, in the initial analysis of consistency and the continuous analysis of consistency, the feature samples are sampled 2 times simultaneously, and the 2 times of feature samples are preprocessed, wherein the preprocessing is as follows:
Step a, transforming and decomposing the characteristic sample A and the characteristic sample B into Z sub-samples, wherein each sub-sample is divided into 2 parts, and the first part filter coefficient is defined as And the second partial filter coefficient is defined as/>C/B represents a characteristic sample A or a characteristic sample B, gamma is more than or equal to 1 and less than or equal to Z represents the gamma decomposition, and m and n are predefined coefficients;
step B, fusing the first partial filter coefficients decomposed at the gamma time by adopting a first fusion criterion, traversing the characteristic sample A or the characteristic sample B, and calculating the correlation degree of the characteristic sample A or the characteristic sample B to obtain a fusion weight, wherein the first fusion criterion is as follows:
Determining a window of P×Q as a region R, calculating gradient magnitudes G x [ i, j ] and G y [ i, j ] of sub-points of each feature sample in the region R in the horizontal direction and the vertical direction, and calculating a gradient value G (i, j)
Calculating the inner product energy of the center point of the region R as E (P (x, y))
The product operation in the region;
the correlation of the characteristic sample A and the characteristic sample B is calculated as follows:
Assuming that the threshold is a, the weight coefficients w C and w D are:
When (when) When w C=0,wD =0;
Time,/> wD=1-wC;
Calculating the characteristic R F(x,y)=wC·RC(x,y)+wC·RC (x, y) of the first part after fusion;
Step A, fusing a second partial coefficient decomposed for the gamma time by adopting a second fusing criterion, and defining a window of P multiplied by Q as a region R; the second fusion criterion is to calculate the maximum value E of the regional energy for fusion;
wherein w (i, j) is the weight of each adjacent point pixel in the region;
And B, reconstructing the fusion result of the step B and the step A by adopting inverse transformation corresponding to the step a, and obtaining a normalized feature F.
In the process of carrying out normalization fusion processing on the acquired 2 samples and improving the precision, a mode of batch fusion after decomposition is adopted, so that the fusion effect is improved, and the fusion efficiency is also improved. Specifically, the part with large frequency point change is classified into the first class as high-frequency conversion data, and the part with small frequency point change is classified into the second class as low-frequency conversion data. In the normalization fusion of the high-frequency conversion data, the energy of the product in the area and the matching degree in the area are used as the basis of the fusion criterion, so that the filtering noise is reduced, and the normalization fusion effect is ensured. The low-frequency data can be fused by adopting the traditional wavelet transformation, the maximum value of the area energy in the invention as a fusion result and the like.
Further, the operator consistency persistence analysis includes:
step 1, detecting N feature peaks from the normalized feature F, denoted { (v i,ti) |i=0, 1. & gt, N }, where N is a natural number greater than 3;
Step 2, calculating the time difference between adjacent characteristic peaks to obtain a characteristic peak interval data characteristic library { (v i,Δti) |i=1, 2. & gt, N };
Step 3, defining a window width w and a window moving speed v;
w=(max(Δti)-min(Δti))×p;
wherein, p is the preset ratio value of the window width to the total width, i is more than or equal to 1 and less than or equal to N;
Step 4, determining a peak threshold range (V 1,V2) from the transverse scan; determining a time interval threshold range (T 1,T2) from the longitudinal scan;
Step 5, defining a region formed by a peak threshold range (V 1,V2) and a time interval threshold range (T 1,T2) as a trusted region of the standard feature point;
Step 6, defining a curve composed of the trusted areas of the standard feature points as a correction feature curve function T= (s 1,s2,...,sh), wherein h is the length of the correction feature curve;
and 7, carrying out consistency contrast calculation on the standard characteristic curve S= (S 1,s2,...,sz) of the authorized user stored in the history and the correction characteristic curve function, judging that the consistency contrast is lower than a preset threshold value, and judging that the consistency contrast is inconsistent if the consistency contrast is lower than the preset threshold value, wherein z is the length of the standard characteristic curve.
Further, the one-time initial analysis includes:
Step A, classifying the first features at odd time sequences in time sequence as first feature sample libraries, and classifying the second features at even time sequences as second feature sample libraries;
Step B, performing Gaussian convolution transformation on the first feature and the second feature with the same sequence number, and respectively extracting Gaussian features delta= {1,2,3,4} on 4 scales and Gaussian convolution features theta= {0,1,2,3,4,5,6,7} × (pi/8) on 8 directions;
Step C, carrying out normalization consistency processing on Gaussian features extracted from the first features and the second features to obtain a consistency interval Wherein y is a feature matrix composed of omega first feature vectors or second feature vectors, mu is the mean value of the feature matrix, and delta is the variance;
step D, calculating the shortest distance b1= [ B11, B21,..ba 1] between the first feature training sample and the first feature detection sample, arranging Bi1 (i1=1, 2,3,..a) in descending order (B11 > B21 >.> BA 1), and determining the minimum distances BA1, respectively; calculating the distance average value Defining the ratio of the distance mean to the shortest distance
Step E, calculating the shortest distance b2= [ B12, B22,..ba 2] between the second feature training sample and the second feature detection sample, arranging Bi2 (i2=1, 2,3,..a) in descending order (B12 > B22 >.> BA 2), and determining the minimum distances BA2, respectively; calculating the distance average valueDefine the ratio of distance mean to shortest distance/>
Step F, calculatingCalculating the unified weight w= [ w 1,w2,...,wm ] of the detection sample, and the average value of the weightsDefine feature fusion as/>Wherein y b1 is the normalized first feature vector, y b2 is the normalized second feature vector, and M is the number of detection samples in the first feature sample library and the second feature sample library;
Step G, calculating the distance between the training sample y Training and the detection sample y Detection of Searching a training sample with the nearest distance from the test sample in a search space according to the Euclidean distance, and attributing the test sample and the training sample to be consistent to complete consistency initial analysis, wherein the Euclidean distance is 2.
Further, the parameter sensor includes a water pressure sensor, a run-time sensor.
In order to increase the reliability and the efficiency of the control data source, the data modeling analysis module further comprises P data modeling analysis subunits connected in parallel, and the weighted fusion result of the P data modeling analysis subunits is output as the data modeling analysis module; each data modeling analysis subunit needs to collect Q data modeling analysis subvariables; each data modeling analysis subunit can only acquire data modeling analysis subvariables at a certain moment in the same period; the data modeling analysis module completes data modeling analysis sub-variable collection of the P data modeling analysis sub-units by running the following method;
Step (1), coding an acquisition strategy of a data modeling analysis subunit and a data modeling analysis subunit variable as a bee Z= [ g 1,g2,...gζ...,gQ×P ], and combining bees into a set; wherein gene g ζ=jη, (j=1,., p=1,..q.) performs an η data modeling analysis sub-variable acquisition on behalf of the j-th data modeling analysis sub-unit, calculating the bee fitness, wherein the bee fitness is a normalization value of the total acquisition time corresponding to the bee;
step (2), presetting convergence conditions according to bee fitness, checking whether the convergence conditions are met, if yes, executing step (5), otherwise, continuing to execute step (3);
step (3), randomly selecting two bees from the collection as a pair of parent bees, randomly selecting a data modeling analysis subunit number, ensuring that the corresponding gene position of the data modeling analysis subunit number in the parent bees is unchanged, and exchanging other genes in the parent bees in sequence in a crossing way to generate a pair of child bees to finish the crossing operation; performing reconstruction operation or local optimization operation according to the crossing result, and executing the step (4);
Step (4), randomly selecting a bee of the mutation operation as the bee to be mutated, randomly selecting a gene j η from the bees, wherein the mutation operation is to reposition the position of the gene j η in the bees to finish the mutation operation under the condition that the sequence of the jth data modeling analysis subunit for executing the eta data modeling analysis subunit variable acquisition is kept unchanged in the total acquisition sequence of the jth data modeling analysis subunit; carrying out reconstruction operation or local optimizing operation according to the mutation result, and returning to the step (2);
Step (5), outputting the optimal bees as an acquisition strategy of the data modeling analysis module, and ending;
the reconstruction operation is to newly select bees to add into bees to be operated when bees generated by the cross operation or the mutation operation are inferior to the bees to be operated;
The local optimizing operation is that bees generated by the crossing operation or the mutation operation are superior to the original bees, the generated bees are utilized to repeat the crossing operation or the mutation operation once again, and then the optimal bees are selected as the next generation bees.
The application has the beneficial effects that: the application aims to solve the problems of authority allocation and confidentiality of manual control in a cloud online central air conditioner fault analysis system. Authority identity consistency authentication is set and is divided into initial identity authentication and subsequent authentication in the use process, and biological characteristics of a user can be collected for consistency continuous analysis and authentication during initial authentication. But there are cases where the registered user identity is the same person during subsequent use. At this time, the present application confirms whether the same person is still not present by continuously receiving the behavior characteristics of the warning controller and performing the analysis and comparison internally. In the process, the identity comparison of the control operator is in the system, and the system and the network interact with the identity identification code after decryption, so that the risk of easy leakage of personal information stored in the network is prevented. In the process of carrying out normalization fusion processing on the acquired 2 samples and improving the precision, a mode of batch fusion after decomposition is adopted, so that the fusion effect is improved, and the fusion efficiency is also improved. Specifically, the part with large frequency point change is classified into the first class as high-frequency conversion data, and the part with small frequency point change is classified into the second class as low-frequency conversion data. In the normalization fusion of the high-frequency conversion data, the energy of the product in the area and the matching degree in the area are used as the basis of the fusion criterion, so that the filtering noise is reduced, and the normalization fusion effect is ensured. The low-frequency data can be fused by adopting the traditional wavelet transformation, the maximum value of the area energy in the application as a fusion result and the like. In order to increase the reliability and efficiency of controlling the data source, the application adopts a strategy that a plurality of data modeling analysis subunits collect a plurality of analysis factors. However, for timeliness, an innovative algorithm is adopted to realize comparison of various strategies, and an optimal strategy is selected for collection.
Drawings
The invention will be further described with reference to the drawings and examples.
Fig. 1 is a schematic diagram of a cloud online central air conditioner fault analysis system.
Detailed Description
The present invention will be described in further detail with reference to the following examples in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
Example 1
The embodiment provides a cloud online central air conditioner fault analysis system, as shown in fig. 1, the cloud online central air conditioner fault analysis system includes: the cooling water circulation system consists of a cooling pump, a cooling water pipe and a cooling tower; the system also comprises a plurality of fault parameter acquisition units which are connected with the refrigerator, the freezing pump, the freezing water pipe, the cooling pump, the cooling water pipe, the cooling tower and the fan coil system in a one-to-one correspondence manner; the data modeling analysis module is connected with the parameter sensor, is connected to the network cloud processor through the blockchain sub-node, and is connected with the warning controller through the blockchain network, and the warning controller outputs a fault warning signal and controls the central air conditioner;
The warning controller comprises an authentication monitoring unit, wherein the authentication monitoring unit comprises a built-in first memory, a built-in receiving unit, a built-in processing unit and an identity characteristic acquisition unit, wherein the built-in receiving unit, the built-in processing unit and the identity characteristic acquisition unit are connected with the first memory, and the built-in processing unit is connected with the output unit to the warning controller; the built-in processing unit is used for carrying out consistency initial analysis on the characteristic data of the real-time operator acquired by the identity characteristic acquisition unit, and continuously carrying out operator consistency continuous analysis on the behavior characteristic data of the operator operation warning controller continuously received by the built-in receiving unit; the built-in processing unit judges the initial identity consistency of the real-time operator and the built-in registration operator according to the consistency initial analysis result, and judges the continuous identity consistency of the identities of the real-time operator and the built-in registration operator according to the operator consistency continuous analysis result; the identity identification code of decryption processing is output through the output unit when the initial identity is consistent with the continuous identity; and the warning controller receives the identity identification code, compares the identity identification code with the identity of the registered operator, and completes identity consistency authentication of the on-line operator and the off-line operator.
The embodiment aims to solve the problems of authority allocation and confidentiality of manual control in the cloud online central air conditioner fault analysis system. Authority identity consistency authentication is set and is divided into initial identity authentication and subsequent authentication in the use process, and biological characteristics of a user can be collected for consistency continuous analysis and authentication during initial authentication. But there are cases where the registered user identity is the same person during subsequent use. At this time, the present application confirms whether the same person is still not present by continuously receiving the behavior characteristics of the warning controller and performing the analysis and comparison internally. In the process, the identity comparison of the control operator is in the system, and the system and the network interact with the identity identification code after decryption, so that the risk of easy leakage of personal information stored in the network is prevented.
Preferably, the built-in processing unit further continuously analyzes the control policy normality of the control policy behavior of the network cloud processor. For automatic control of the network cloud processor, the embodiment can adopt an analysis method consistent with consistency continuous analysis to carry out similarity analysis on the control strategy and the history strategy of the network cloud processor, so as to ensure that no control abnormality occurs.
In the initial analysis of consistency and the continuous analysis of consistency, the analysis of consistency can be directly carried out according to the prior art. The preferred scheme can also be adopted, namely, the characteristic samples are sampled for 2 times simultaneously, and the 2 times of characteristic samples are preprocessed, wherein the preprocessing is as follows:
Step a, transforming and decomposing the characteristic sample A and the characteristic sample B into Z sub-samples, wherein each sub-sample is divided into 2 parts, and the first part filter coefficient is defined as And the second partial filter coefficient is defined as/>C/B represents a characteristic sample A or a characteristic sample B, gamma is more than or equal to 1 and less than or equal to Z represents the gamma decomposition, and m and n are predefined coefficients;
step B, fusing the first partial filter coefficients decomposed at the gamma time by adopting a first fusion criterion, traversing the characteristic sample A or the characteristic sample B, and calculating the correlation degree of the characteristic sample A or the characteristic sample B to obtain a fusion weight, wherein the first fusion criterion is as follows:
Determining a window of P×Q as a region R, calculating gradient magnitudes G x [ i, j ] and G y [ i, j ] of sub-points of each feature sample in the region R in the horizontal direction and the vertical direction, and calculating a gradient value G (i, j)
Calculating the inner product energy of the center point of the region R as E (P (x, y))
The product operation in the region;
the correlation of the characteristic sample A and the characteristic sample B is calculated as follows:
Assuming that the threshold is a, the weight coefficients w C and w D are:
When (when) When w C=0,wD =0;
Time,/> wD=1-wC;
Calculating the characteristic R F(x,y)=wC·RC(x,y)+wC·RC (x, y) of the first part after fusion;
Step A, fusing a second partial coefficient decomposed for the gamma time by adopting a second fusing criterion, and defining a window of P multiplied by Q as a region R; the second fusion criterion is to calculate the maximum value E of the regional energy for fusion;
wherein w (i, j) is the weight of each adjacent point pixel in the region;
And B, reconstructing the fusion result of the step B and the step A by adopting inverse transformation corresponding to the step a, and obtaining a normalized feature F.
In the embodiment, in the process of carrying out normalized fusion processing on the acquired 2 times of samples and improving the precision, a mode of batch fusion after decomposition is adopted, so that the fusion effect is improved, and the fusion efficiency is also improved. Specifically, the part with large frequency point change is classified into the first class as high-frequency conversion data, and the part with small frequency point change is classified into the second class as low-frequency conversion data. In the normalization fusion of the high-frequency conversion data, the energy of the product in the area and the matching degree in the area are used as the basis of the fusion criterion, so that the filtering noise is reduced, and the normalization fusion effect is ensured. The low-frequency data can be fused by adopting the traditional wavelet transformation, the maximum value of the area energy in the invention as a fusion result and the like.
In the consistency continuous analysis, the consistency analysis can be directly and continuously performed according to the prior art. The preferred scheme can also be adopted, namely the operator consistency continuous analysis method comprises the following steps:
step 1, detecting N feature peaks from the normalized feature F, denoted { (v i,ti) |i=0, 1. & gt, N }, where N is a natural number greater than 3;
Step 2, calculating the time difference between adjacent characteristic peaks to obtain a characteristic peak interval data characteristic library { (v i,Δti) |i=1, 2. & gt, N };
Step 3, defining a window width w and a window moving speed v;
w=(max(Δti)-min(Δti))×p;
wherein, p is the preset ratio value of the window width to the total width, i is more than or equal to 1 and less than or equal to N;
Step 4, determining a peak threshold range (V 1,V2) from the transverse scan; determining a time interval threshold range (T 1,T2) from the longitudinal scan;
Step 5, defining a region formed by a peak threshold range (V 1,V2) and a time interval threshold range (T 1,T2) as a trusted region of the standard feature point;
Step 6, defining a curve composed of the trusted areas of the standard feature points as a correction feature curve function T= (s 1,s2,...,sh), wherein h is the length of the correction feature curve;
and 7, carrying out consistency contrast calculation on the standard characteristic curve S= (S 1,s2,...,sz) of the authorized user stored in the history and the correction characteristic curve function, judging that the consistency contrast is lower than a preset threshold value, and judging that the consistency contrast is inconsistent if the consistency contrast is lower than the preset threshold value, wherein z is the length of the standard characteristic curve.
In the initial analysis of consistency, a consistency control analysis can be directly performed according to the prior art. The present preferred embodiment may also be adopted, i.e. the one-time initial analysis comprises:
Step A, classifying the first features at odd time sequences in time sequence as first feature sample libraries, and classifying the second features at even time sequences as second feature sample libraries;
Step B, performing Gaussian convolution transformation on the first feature and the second feature with the same sequence number, and respectively extracting Gaussian features delta= {1,2,3,4} on 4 scales and Gaussian convolution features theta= {0,1,2,3,4,5,6,7} × (pi/8) on 8 directions;
Step C, carrying out normalization consistency processing on Gaussian features extracted from the first features and the second features to obtain a consistency interval Wherein y is a feature matrix composed of omega first feature vectors or second feature vectors, mu is the mean value of the feature matrix, and delta is the variance;
step D, calculating the shortest distance b1= [ B11, B21,..ba 1] between the first feature training sample and the first feature detection sample, arranging Bi1 (i1=1, 2,3,..a) in descending order (B11 > B21 >.> BA 1), and determining the minimum distances BA1, respectively; calculating the distance average value Define the ratio of distance mean to shortest distance/>
Step E, calculating the shortest distance b2= [ B12, B22,..ba 2] between the second feature training sample and the second feature detection sample, arranging Bi2 (i2=1, 2,3,..a) in descending order (B12 > B22 >.> BA 2), and determining the minimum distances BA2, respectively; calculating the distance average valueDefine the ratio of distance mean to shortest distance/>
Step F, calculatingCalculating the unified weight w= [ w 1,w2,...,wm ] of the detection sample, and the average value of the weightsDefine feature fusion as/>Wherein y b1 is the normalized first feature vector, y b2 is the normalized second feature vector, and M is the number of detection samples in the first feature sample library and the second feature sample library;
Step G, calculating the distance between the training sample y Training and the detection sample y Detection of Searching a training sample with the closest distance to the test sample in a search space according to the Euclidean distance, and attributing the test sample and the training sample to be consistent to complete consistency initial analysis, wherein 2 is the Euclidean distance.
Further, the parameter sensor includes a water pressure sensor, a run-time sensor.
In order to increase the reliability and the efficiency of the control data source, the data modeling analysis module further comprises P data modeling analysis subunits connected in parallel, and the weighted fusion result of the P data modeling analysis subunits is output as the data modeling analysis module; each data modeling analysis subunit needs to collect Q data modeling analysis subvariables; each data modeling analysis subunit can only acquire data modeling analysis subvariables at a certain moment in the same period; the data modeling analysis module completes data modeling analysis sub-variable collection of the P data modeling analysis sub-units by running the following method;
Step (1), coding an acquisition strategy of a data modeling analysis subunit and a data modeling analysis subunit variable as a bee Z= [ g 1,g2,...gζ...,gQ×P ], and combining bees into a set; wherein gene g ζ=jη, (j=1,., p=1,..q.) performs an η data modeling analysis sub-variable acquisition on behalf of the j-th data modeling analysis sub-unit, calculating the bee fitness, wherein the bee fitness is a normalization value of the total acquisition time corresponding to the bee;
Step (2), because the acquisition time is required for each acquisition, estimating the total acquisition time according to the acquisition sequence in bees, presetting convergence conditions according to the bee fitness, checking whether the convergence conditions are met, if yes, executing the step (5) to complete the acquisition strategy assignment, otherwise, continuing to execute the step (3);
step (3), randomly selecting two bees from the collection as a pair of parent bees, randomly selecting a data modeling analysis subunit number, ensuring that the corresponding gene position of the data modeling analysis subunit number in the parent bees is unchanged, and exchanging other genes in the parent bees in sequence in a crossing way to generate a pair of child bees to finish the crossing operation; performing reconstruction operation or local optimization operation according to the crossing result, and executing the step (4);
for example, the parent bees to be crossed are selected as follows:
31224114243442132333124311213244
22142441313442231333432112443211
the number of the randomly determined data modeling analysis subunit is 2, and then the pair of the daughter bees after the cross operation is:
14224131243442132333431244213211
22312441143442231333122143113244
Step (4), randomly selecting a bee of the mutation operation as the bee to be mutated, randomly selecting a gene j η from the bees, wherein the mutation operation is to reposition the position of the gene j η in the bees to finish the mutation operation under the condition that the sequence of the jth data modeling analysis subunit for executing the eta data modeling analysis subunit variable acquisition is kept unchanged in the total acquisition sequence of the jth data modeling analysis subunit; carrying out reconstruction operation or local optimizing operation according to the mutation result, and returning to the step (2);
For example: the selected bees to be mutated are:
31224114243442132333124311213244
the randomly selected gene is 3 4, and the mutated bees are:
31223441142442132333124311213244
Step (5), outputting the optimal bees as an acquisition strategy of the data modeling analysis module, and ending;
the reconstruction operation is to newly select bees to add into bees to be operated when bees generated by the cross operation or the mutation operation are inferior to the bees to be operated;
The local optimizing operation is that bees generated by the crossing operation or the mutation operation are superior to the original bees, the generated bees are utilized to repeat the crossing operation or the mutation operation once again, and then the optimal bees are selected as the next generation bees.
At this time, the probability of intersection and variation in this embodiment can obtain a better convergence rate according to the evolutionary change of the population, and the probability of intersection and variation is expressed as follows:
Wherein, P C is crossover probability, P m is mutation probability, F max is the maximum adaptability in the population, F avg is the average adaptability of the population bees, F is the adaptability of crossover bees or mutation bees, and k c2、kc1、km2、km1 is a constant preset between [0,1 ].
The embodiment aims to solve the problems of authority allocation and confidentiality of manual control in the cloud online central air conditioner fault analysis system. Authority identity consistency authentication is set and is divided into initial identity authentication and subsequent authentication in the use process, and biological characteristics of a user can be collected for consistency continuous analysis and authentication during initial authentication. But there are cases where the registered user identity is the same person during subsequent use. At this time, the present application confirms whether the same person is still not present by continuously receiving the behavior characteristics of the warning controller and performing the analysis and comparison internally. In the process, the identity comparison of the control operator is in the system, and the system and the network interact with the identity identification code after decryption, so that the risk of easy leakage of personal information stored in the network is prevented. In the process of carrying out normalization fusion processing on the acquired 2 samples and improving the precision, a mode of batch fusion after decomposition is adopted, so that the fusion effect is improved, and the fusion efficiency is also improved. Specifically, the part with large frequency point change is classified into the first class as high-frequency conversion data, and the part with small frequency point change is classified into the second class as low-frequency conversion data. In the normalization fusion of the high-frequency conversion data, the energy of the product in the area and the matching degree in the area are used as the basis of the fusion criterion, so that the filtering noise is reduced, and the normalization fusion effect is ensured. The low-frequency data can be fused by adopting the traditional wavelet transformation, the maximum value of the area energy in the application as a fusion result and the like. In order to increase the reliability and efficiency of controlling the data source, the application adopts a strategy that a plurality of data modeling analysis subunits collect a plurality of analysis factors. However, for timeliness, an innovative algorithm is adopted to realize comparison of various strategies, and an optimal strategy is selected for collection.
While the foregoing describes the illustrative embodiments of the present invention so that those skilled in the art may understand the present invention, the present invention is not limited to the specific embodiments, and all inventive innovations utilizing the inventive concepts are herein within the scope of the present invention as defined and defined by the appended claims, as long as the various changes are within the spirit and scope of the present invention.