CN113705867A - Equipment abnormity diagnosis method based on energy consumption interval prediction - Google Patents
Equipment abnormity diagnosis method based on energy consumption interval prediction Download PDFInfo
- Publication number
- CN113705867A CN113705867A CN202110937060.6A CN202110937060A CN113705867A CN 113705867 A CN113705867 A CN 113705867A CN 202110937060 A CN202110937060 A CN 202110937060A CN 113705867 A CN113705867 A CN 113705867A
- Authority
- CN
- China
- Prior art keywords
- energy consumption
- prediction
- interval
- value
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000005265 energy consumption Methods 0.000 title claims abstract description 129
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000003745 diagnosis Methods 0.000 title claims abstract description 8
- 230000002159 abnormal effect Effects 0.000 claims abstract description 13
- 238000012360 testing method Methods 0.000 claims description 30
- 238000012549 training Methods 0.000 claims description 29
- 108090000672 Annexin A5 Proteins 0.000 claims description 12
- 230000005856 abnormality Effects 0.000 claims description 10
- 239000000126 substance Substances 0.000 claims description 9
- 238000011156 evaluation Methods 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 6
- 238000005070 sampling Methods 0.000 claims description 5
- 210000002569 neuron Anatomy 0.000 claims description 3
- 238000012423 maintenance Methods 0.000 abstract description 8
- 238000004134 energy conservation Methods 0.000 abstract description 5
- 238000004519 manufacturing process Methods 0.000 abstract description 5
- 238000012544 monitoring process Methods 0.000 abstract description 3
- 238000007689 inspection Methods 0.000 abstract description 2
- 238000000926 separation method Methods 0.000 description 16
- 230000008569 process Effects 0.000 description 6
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000007787 long-term memory Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
- 238000013024 troubleshooting Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/04—Manufacturing
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Abstract
The invention discloses an equipment abnormity diagnosis method based on energy consumption interval prediction, which solves the problems that misjudgment is easily caused by judging energy consumption according to a fitting curve and further energy-saving deployment of a production operation and maintenance link is influenced in the prior art, and comprises the following steps. The method judges whether the equipment is abnormal or not, whether the equipment has an energy-saving space or not or whether the equipment has a fault or not, provides data support for the operation and maintenance of the equipment, can also find out the reason possibly causing the overhigh energy consumption by calculating the correlation coefficient of the energy consumption value and related variables aiming at the equipment with overhigh energy consumption, and provides data support for reducing the energy consumption by taking measures later; the inspection and maintenance can be performed for devices with too low energy consumption. And monitoring the energy consumption condition of each device in time and feeding back unreasonable energy consumption so as to achieve the aim of energy conservation.
Description
Technical Field
The invention relates to the technical field of equipment energy consumption monitoring, in particular to an equipment abnormity diagnosis method based on energy consumption interval prediction.
Background
The energy conservation has very important significance in the production, operation and maintenance links. Each production, operation and maintenance link should actively respond to the call for energy conservation, investigate energy-saving obstacles and energy waste phenomena, and find the energy-saving potential of equipment, so as to achieve the purposes of energy conservation, emission reduction, production cost reduction and economic benefit improvement. In the conventional energy management process, whether the operation condition of the equipment is abnormal or not is generally checked through a physical process, for example, whether the equipment belongs to laggard equipment and processes or not is checked, and the laggard equipment and processes are eliminated; measuring the operating efficiency of the equipment on site; and judging whether the energy consumption of the equipment is abnormal or not according to the experience of the field process engineer. Such methods have the disadvantages of low efficiency and low accuracy.
In addition, some problems often exist in some existing recognition technologies, which result in low accuracy, for example, an ideal energy consumption curve is fitted by using historical energy consumption data, then a real value is compared with a fitted value, and if an absolute value of a comparison error exceeds a certain range, it is determined that the energy consumption of equipment is abnormal. However, the energy consumption of the device is dynamically changed due to factors such as season, date, temperature, holiday, etc., so that the historical energy consumption data cannot accurately describe the energy consumption performance at the current moment, and the error value for determination is difficult to accurately estimate.
For example, a chinese patent document discloses "a method for predicting energy consumption of an air separation system", which is disclosed in CN103793754B, and includes the following steps: acquiring energy consumption data of air separation plants of different scales; fitting the energy consumption data by using a least square method to obtain a fitting equation of the energy consumption of the air separation equipment along with the air separation scale; determining the power consumption of the air separation equipment according to a fitting equation general formula; calculating the electrical efficiency eta of the air separation equipment with different scales according to the power consumption of the air separation equipment and the technical work of the air separation equipment; determining the electrical efficiency eta of the air separation equipment according to the calculated electrical efficiency eta of the air separation equipment with different scales1(ii) a According to the electrical efficiency of each air separation plantRate eta1Calculating the power consumption of each air separation plant in the air separation project; and calculating to obtain the energy consumption of the air separation system according to the energy consumption of each air separation device in the air separation project. According to the technical scheme, the fitting curve is combined with a model equation, the fitting curve of the electrical efficiency is obtained by utilizing actual measurement data, if the absolute value of the comparison error exceeds a certain range, the abnormal energy consumption of the equipment is judged, the influence of factors such as season, date, temperature and holiday on the energy consumption is difficult to fuse, the fitting curve cannot reflect the dynamic change, the energy consumption performance at each moment cannot be accurately described, the difficulty of judging the abnormal energy consumption is increased by the judgment error, and the misjudgment probability is very high.
Disclosure of Invention
The invention aims to solve the problem that misjudgment is easily caused by energy consumption according to a fitting curve so as to influence energy-saving deployment of a production operation and maintenance link in the prior art, and provides an equipment abnormity diagnosis method based on energy consumption interval prediction, which is used for judging whether equipment is abnormal or not, whether the equipment has an energy-saving space or whether the equipment has a fault or not, providing data support for operation and maintenance of the equipment, finding out the reason possibly causing overhigh energy consumption by calculating the correlation coefficient of an energy consumption value and related variables aiming at the equipment with overhigh energy consumption, and providing data support for reducing the energy consumption by taking measures later; the inspection and maintenance can be performed for devices with too low energy consumption. And monitoring the energy consumption condition of each device in time and feeding back unreasonable energy consumption so as to achieve the aim of energy conservation.
In order to achieve the purpose, the invention adopts the following technical scheme:
an equipment abnormity diagnosis method based on energy consumption interval prediction comprises the following steps:
s1, collecting historical energy consumption, time, date, temperature and weather data and training an energy consumption prediction model;
s2, point prediction is carried out, and the energy consumption value in unit time is predicted;
s3, performing interval prediction, and predicting the upper limit and the lower limit of the energy consumption value in unit time;
s4, checking whether the real energy consumption value in unit hour falls within a confidence interval;
s5, judging whether the energy consumption of the equipment is abnormal or not, and when the real value is within the confidence interval, the equipment normally operates; when the real value is larger than the upper limit of the prediction, the equipment is considered to have an energy-saving space; when the true value is less than the lower prediction limit, the device is considered to be likely to fail.
The method carries out interval prediction on the future energy consumption value, the interval prediction is obtained based on the prediction of the energy consumption value at a certain moment, and the energy consumption value at the certain moment is predicted to comprehensively consider the influences of factors such as weather conditions (sunny/cloudy/rainy), temperature and date types (working days/holidays/double-holidays) and the like, so that the predicted energy consumption value has a reference value compared with an empirical value.
The energy consumption value interval prediction is obtained based on the historical energy consumption data prediction of the equipment, the method can be deployed to each equipment, whether the energy consumption value of the equipment exceeds the prediction range is monitored in real time, and the energy-saving scheme can be conveniently deployed, overhauled and maintained daily for different equipment.
Preferably, the S3 includes the following steps:
s301, collecting historical data;
s302, dividing a data set;
s303, performing Bootstrap sampling;
s304, training a GRU prediction model;
s305, testing set data point prediction;
s306, testing the prediction of the set interval;
s307, evaluating the quality of interval prediction;
s308, whether the evaluation index meets the requirement or not is judged, and generally, the smaller the PINAW is, the better the PINAW is under the condition that the PICP is more than or equal to mu; the smaller the CWC is, the better the prediction quality of the interval is;
s309, predicting the interval of the energy consumption value by using the prediction model, and substituting the data x to be evaluated into the prediction model in the S304And the point prediction value is obtained and then substituted into S306 to obtain the interval prediction value.
The method is mainly based on a GRU gated cyclic unit neural network algorithm to establish a prediction model and predict the energy consumption value of unit time in one day in the future, and in the prediction process, an optimal model is selected to perform multiple predictions.
The gru (gate recovery unit) is one of Recurrent Neural Networks (RNN) and is also a variant of LSTM Network with good effect, so that the problems of long-term memory and gradient in back propagation can be solved; it is simpler than LSTM network structure, and the effect is also very good.
Preferably, the S301 includes the steps of:
s3011, collecting data of energy consumption values, date types, weather types and temperatures of historical unit hours; the date types comprise working days, holidays and double-holidays, and the weather types comprise cloudy days, sunny days and rainy days; obtaining a data set that can be used for time series prediction: h ═ Ht,ht+1......ht+n]Wherein h ist=[ft,dt,wt,ct],ftRepresenting the energy consumption value at time t, dtType of date, w, representing the t time pointtWeather type at time t, ctTemperature value, h, representing the time ttData representing an energy consumption value, a date type, a weather type and a temperature value at a time point t;
s3012, one-hot coding is carried out on the date type field and the weather type field: converting the type field into a value with the type as a number, so that the processing is convenient;
s3013, carrying out standardization processing on the data set: namely, values of energy consumption, a date type field, a weather type field and temperature are scaled to be between 0 and 1;
s3014, converting the time series data set H into a supervised learning data set W ═ Wt,wt+1......wt+n],
Wherein, wt=[ft,dt,,wt,ct,ft+1,dt+1,wt+1,ct+1......ft+k-1,dt+k-1,wt+k-1,ct+k-1,ft+k,dt+k,wt+k,ct+k](ii) a k-24 represents the data volume for one day;
s3015, dividing the first four fifths of the data set W into a training set WtrainLast fifth test set Wtest;
S3016, separating the training set to obtain input quantity W of the training settrain_xAnd an output quantity Wtrain_y,
Wherein, Wtrain_x=[xt,xt+1......xt+4n/5],
Wtrain_y=[yt,yt+1......yt+4n/5],
xt=[ft,dt,wt,ct,ft+1,dt+1,wt+1,ct+1......ft+k-1,dt+k-1,wt+k-1,ct+k-1],
yt=[ft+k];
Similarly, the input quantity W of the test set can be obtainedtest_xAnd an output quantity Wtest_y。
one-host encoding: the N states are encoded using N-bit state registers, a common encoding scheme.
Preferably, the S302 includes the following contents:
randomly dividing an original data set into training sets WtrainAnd test set Wtest,WtrainIn which there are s pieces of data, WtestThere are k pieces of data, and s + k is n + 1.
Preferably, S303 is selected from WtrainExtracting s pieces of data in a place-back manner to form Bootstrap samples, and extracting for B times to obtain B sample sets:
preferably, the S304 includes the steps of:
s3041, collecting the B samples obtained in S303Separating to obtain a training setInput amount ofAnd output quantity
S3042, defining a GRU energy consumption prediction model and setting related parameters, wherein the model is designed into three layers, namely a first layer is an input layer, a second layer is a hidden layer, a third layer is an output layer, and the number of neurons in each layer is set;
s3043 input quantity using training setAnd output quantityTraining the model to obtain a training model
Preferably, the S305 includes the following contents:
substituting the ith piece of data in the test set into the prediction modelObtaining the predicted value of B points
Preferably, the S306 includes the following contents: taking the confidence p as 0.95 as an example, for the data in the test set, the predicted values of the B points are obtained2.5% and 97.5%, to obtain a 95% confidence interval, i.e. the prediction interval for the energy consumption value.
Bootstrap, also known as the Bootstrap method, is a nonparametric method for estimating the overall value by using a small sample, and is widely applied to evolutionary and ecological research.
The confidence interval is an estimation interval of a total parameter constructed by sample statistics, the confidence interval of the point prediction value is generated by repeatedly sampling the result of multiple predictions by a Bootstrap self-development method, and the confidence degree is 95%; when the actual energy consumption value is generated at the moment, judging whether the real energy consumption value falls in a confidence interval, and when the real energy consumption value falls in the confidence interval, normally operating the equipment; when the true value falls outside the confidence interval, the device is abnormal. When the real value is larger than the upper limit of the confidence interval, judging that the equipment has an energy-saving space, and then finding out the reason causing overlarge energy consumption by calculating the correlation coefficient of related variables and energy consumption values so as to provide field workers to take measures in a targeted manner to reduce the energy consumption; when the true value is less than the lower limit of the confidence interval, it is determined that the device may fail.
Preferably, the S307 includes the following:
the range probability formula is:
wherein PICP is the range probability;
the normalized interval width formula is:
PINAW is normalized interval width
The comprehensive index formula is as follows:
CWC=PINAW+γ(PICP)Eη(μ-PICP)wherein the content of the first and second substances,CWC is a comprehensive index.
Preferably, the S308 includes the following contents: whether the evaluation index meets the requirement or not is judged, and generally, the smaller the PINAW is, the better the PINAW is under the condition that the PICP is more than or equal to mu; the smaller the CWC is, the better the prediction quality of the interval is;
the S309 includes the following contents: predicting the interval of the energy consumption value by utilizing a GRU prediction model; substituting the data x to be evaluated into the prediction model in S304And the point prediction value is obtained and then substituted into S306 to obtain the interval prediction value.
Therefore, the invention has the following beneficial effects:
the method comprises the steps of obtaining an energy consumption value prediction model based on historical energy consumption values, weather, temperature and date data, and predicting the energy consumption value of a unit hour in the future; based on a plurality of predicted values obtained by multiple predictions, repeated sampling is carried out by a self-expansion method, and a 95% confidence interval of the sample capacity is obtained, namely the prediction interval of the energy consumption value at the time; based on the obtained upper limit and lower limit values of the prediction interval, whether the equipment normally operates, whether an energy-saving space exists and whether a fault occurs are judged by judging whether the real energy consumption value of the time falls in the prediction interval; based on the correlation coefficient of the related variable and the energy consumption value, the factors which are most likely to cause the energy consumption value to be overlarge are found out, and the factors are checked one by one, so that data support is provided for reducing the energy consumption by taking measures.
Drawings
Fig. 1 is an overall flowchart of the present embodiment for determining whether an apparatus energy consumption value is abnormal and determining the type of the abnormality.
Fig. 2 is a flowchart of the energy consumption value prediction interval according to the present embodiment.
Fig. 3 shows the prediction interval and the actual value (normal) of the energy consumption value per unit time of a day in this embodiment.
Fig. 4 shows the prediction interval and the actual value (exception 1) of the energy consumption value per unit time of a day in the present embodiment.
Fig. 5 shows the prediction interval and the actual value of the energy consumption value per unit time of a day (anomaly 2) in the present embodiment.
Fig. 6 is a flowchart of finding a cause that may cause excessive energy consumption when the energy consumption of the air conditioner in the subway station is abnormal and is higher than the upper limit of the prediction interval according to the embodiment.
Detailed Description
The invention is further described with reference to the following detailed description and accompanying drawings.
Example (b):
the embodiment provides an apparatus abnormality diagnosis method based on energy consumption interval prediction, as shown in fig. 1, including the following steps:
s1, collecting historical energy consumption, time, date, temperature and weather data and training an energy consumption prediction model;
s2, point prediction is carried out, and the energy consumption value in unit time is predicted;
s3, performing interval prediction, and predicting the upper limit and the lower limit of the energy consumption value in unit time;
s4, checking whether the real energy consumption value in unit hour falls within a confidence interval;
s5, judging whether the energy consumption of the equipment is abnormal or not, and when the real value is within the confidence interval, the equipment normally operates; when the real value is larger than the upper limit of the prediction, the equipment is considered to have an energy-saving space; when the true value is less than the lower prediction limit, the device is considered to be likely to fail.
As shown in fig. 2, the S3 includes the following steps:
s301, collecting historical data;
the S301 includes the steps of:
s3011, collecting data of energy consumption values, date types, weather types and temperatures of historical unit hours; the date types comprise working days, holidays and double-holidays, and the weather types comprise cloudy days, sunny days and rainy days; obtaining a data set that can be used for time series prediction: h ═ Ht,ht+1......ht+n]Wherein h ist=[ft,dt,wt,ct],ftRepresenting the energy consumption value at time t, dtType of date, w, representing the t time pointtWeather type at time t, ctTemperature value, h, representing the time ttRepresenting the energy consumption value at the time point t, the date type,Data of weather type and temperature value;
s3012, one-hot coding is carried out on the date type field and the weather type field: converting the type field into a value with the type as a number, so that the processing is convenient;
s3013, carrying out standardization processing on the data set: namely, values of energy consumption, a date type field, a weather type field and temperature are scaled to be between 0 and 1;
s3014, converting the time series data set H into a supervised learning data set W ═ Wt,wt+1......wt+n],
Wherein, wt=[ft,dt,,wt,ct,ft+1,dt+1,wt+1,ct+1......ft+k-1,dt+k-1,wt+k-1,ct+k-1,ft+k,dt+k,wt+k,ct+k](ii) a k-24 represents the data volume for one day;
s3015, dividing the first four fifths of the data set W into a training set WtrainLast fifth test set Wtest;
S3016, separating the training set to obtain input quantity W of the training settrain_xAnd an output quantity Wtrain_y,
Wherein, Wtrain_x=[xt,xt+1......xt+4n/5],
Wtrain_y=[yt,yt+1......yt+4n/5],
xt=[ft,dt,wt,ct,ft+1,dt+1,wt+1,ct+1......ft+k-1,dt+k-1,wt+k-1,ct+k-1],
yt=[ft+k];
Similarly, the input quantity W of the test set can be obtainedtest_xAnd an output quantity Wtest_y。
S302, dividing a data set;
the S302 includes the following contents:
randomly partitioning a raw data set into trainsExercise Collection WtrainAnd test set Wtest,WtrainIn which there are s pieces of data, WtestThere are k pieces of data, and s + k is n + 1.
S303, performing Bootstrap sampling;
the S303 is from WtrainExtracting s pieces of data in a place-back manner to form Bootstrap samples, and extracting for B times to obtain B sample sets:
s304, training a GRU prediction model;
the S304 includes the steps of:
s3041, collecting the B samples obtained in S303Separating to obtain input quantity of training setAnd output quantity
S3042, defining a GRU energy consumption prediction model and setting related parameters, wherein the model is designed into three layers, namely a first layer is an input layer, a second layer is a hidden layer, a third layer is an output layer, and the number of neurons in each layer is set;
s3043 input quantity using training setAnd output quantityTraining the model to obtain a training model
S305, testing set data point prediction;
the S305 includes the following contents:
substituting the ith piece of data in the test set into the prediction modelObtaining the predicted value of B points
S306, testing the prediction of the set interval;
the S306 includes the following contents: taking the confidence p as 0.95 as an example, for the data in the test set, the predicted values of the B points are obtained2.5% and 97.5%, to obtain a 95% confidence interval, i.e. the prediction interval for the energy consumption value.
S307, evaluating the quality of interval prediction;
the S307 includes the following contents:
the range probability formula is:
wherein PICP is the range probability;
the normalized interval width formula is:
PINAW is normalized interval width
The comprehensive index formula is as follows:
CWC=PINAW+γ(PICP)eη(μ-PICP)wherein the content of the first and second substances,CWC is a comprehensive index.
S308, whether the evaluation index meets the requirement or not is judged, and generally, the smaller the PINAW is, the better the PINAW is under the condition that the PICP is more than or equal to mu; the smaller the CWC is, the better the prediction quality of the interval is;
the S308 includes the following contents: whether the evaluation index meets the requirement or not is judged, and generally, the smaller the PINAW is, the better the PINAW is under the condition that the PICP is more than or equal to mu; a smaller CWC indicates a better quality of the interval prediction.
S309, predicting the interval of the energy consumption value by using the prediction model, and substituting the data x to be evaluated into the prediction model in the S304Obtaining a point predicted value, and substituting the point predicted value into S306 to obtain an interval predicted value;
the S309 includes the following contents: predicting the interval of the energy consumption value by utilizing a GRU prediction model; substituting the data x to be evaluated into the prediction model in S304And the point prediction value is obtained and then substituted into S306 to obtain the interval prediction value.
In fig. 3, the upper prediction limit is the uppermost curve in the graph, the lower prediction limit is the lowermost curve in the graph, and the actual value curve is between the upper prediction limit and the lower prediction limit.
In fig. 4, the energy consumption is too low and the real value curve suddenly drops below the lower prediction limit, at which point the device may fail.
In fig. 5, the situation occurs that the energy consumption is too high and the real value curve suddenly rises above the upper prediction limit, and the device has an energy-saving space.
As shown in fig. 6, when the energy consumption of the air conditioner in the subway station is higher than the upper limit of the prediction interval, the reason for troubleshooting includes the following steps:
step 1: collecting historical energy consumption value P ═ P per hourt,pt+1,……pt+n]Collecting factors related to air conditioner energy consumption, such as people flow F ═ Ft,ft+1,……ft+n]And the indoor and outdoor temperature difference W ═ Wt,wt+1,……wt+n]Humidity H ═ Ht,ht+1,……ht+n]Waiting for data to obtain a data set related to the energy consumption anomaly, wherein ptRepresenting the energy consumption value at time t, ftIndicating the flow of people at time t, wtRepresents the indoor and outdoor temperature difference at the time point t, htRepresents the humidity at time t;
step 2: respectively calculating the Pearson correlation coefficient of each variable (F, W, H) and the energy consumption value P in the step 1, wherein the formula is as follows:
wherein Cov (X, Y) is the covariance of X and Y, Var [ X ] is the variance of X, and Var [ Y ] is the variance of Y.
And step 3: carrying out reverse order arrangement on the correlation coefficients obtained in the step 2;
and 4, checking the variables with high correlation coefficients one by one, checking whether the variables are the reasons for increasing the energy consumption, and providing data support for measures for reducing the energy consumption later, such as switching on and off the air conditioner, adjusting the temperature and the like.
The above embodiments are described in detail for the purpose of further illustrating the present invention and should not be construed as limiting the scope of the present invention, and the skilled engineer can make insubstantial modifications and variations of the present invention based on the above disclosure.
Claims (10)
1. An equipment abnormity diagnosis method based on energy consumption interval prediction is characterized by comprising the following steps:
s1, collecting historical energy consumption, time, date, temperature and weather data and training an energy consumption prediction model;
s2, point prediction is carried out, and the energy consumption value in unit time is predicted;
s3, performing interval prediction, and predicting the upper limit and the lower limit of the energy consumption value in unit time;
s4, checking whether the real energy consumption value in unit hour falls within a confidence interval;
s5, judging whether the energy consumption of the equipment is abnormal or not, and when the real value is within the confidence interval, the equipment normally operates; when the real value is larger than the upper limit of the prediction, the equipment is considered to have an energy-saving space; when the true value is less than the lower prediction limit, the device is considered to be likely to fail.
2. The method for diagnosing abnormality of equipment according to claim 1, wherein said S3 includes the steps of:
s301, collecting historical data;
s302, dividing a data set;
s303, performing Bootstrap sampling;
s304, training a GRU prediction model;
s305, testing set data point prediction;
s306, testing the prediction of the set interval;
s307, evaluating the quality of interval prediction;
s308, whether the evaluation index meets the requirement or not is judged, and generally, the smaller the PINAW is, the better the PINAW is under the condition that the PICP is more than or equal to mu; the smaller the CWC is, the better the prediction quality of the interval is;
3. The method for diagnosing the abnormality of the equipment based on the prediction of the energy consumption interval as set forth in claim 2, wherein the step S301 includes the steps of:
s3011, collecting data of energy consumption values, date types, weather types and temperatures of historical unit hours; the date types comprise working days, holidays and double-holidays, and the weather types comprise cloudy days, sunny days and rainy days; obtaining a data set that can be used for time series prediction: h ═ Ht,ht+1......ht+n]Wherein h ist=[ft,dt,wt,ct],ftRepresenting the energy consumption value at time t, dtType of date, w, representing the t time pointtWeather type at time t, ctTemperature value, h, representing the time ttData representing an energy consumption value, a date type, a weather type and a temperature value at a time point t;
s3012, one-hot coding is carried out on the date type field and the weather type field: converting the type field into a value with the type as a number, so that the processing is convenient;
s3013, carrying out standardization processing on the data set: namely, values of energy consumption, a date type field, a weather type field and temperature are scaled to be between 0 and 1;
s3014, converting the time series data set H into a supervised learning data set W ═ Wt,wt+1......wt+n],
Wherein, wt=[ft,dt,,wt,ct,ft+1,dt+1,wt+1,ct+1......ft+k-1,dt+k-1,wt+k-1,ct+k-1,ft+k,dt+k,wt+k,ct+k](ii) a k-24 represents the data volume for one day;
s3015, dividing the first four fifths of the data set W into a training set WtrainLast fifth test set Wtest;
S3016, separating the training set to obtain input quantity W of the training settrain_xAnd an output quantity Wtrain_y,
Wherein, Wtrain_x=[xt,xt+1......xt+4n/5],
Wtrain_y=[yt,yt+1......yt+4n/5],
xt=[ft,dt,wt,ct,ft+1,dt+1,wt+1,ct+1......ft+k-1,dt+k-1,wt+k-1,ct+k-1],
yt=[ft+k];
Similarly, the input quantity W of the test set can be obtainedtest_xAnd an output quantity Wtest_y。
4. The method for diagnosing the abnormality of the equipment based on the prediction of the energy consumption interval as set forth in claim 2, wherein the step S302 includes:
randomly dividing an original data set into training sets WtrainAnd test set Wtest,WtrainIn which there are s pieces of data, WtestThere are k pieces of data, and s + k is n + 1.
6. the method for diagnosing abnormality of equipment according to claim 2, wherein said S304 includes the steps of:
s3041, collecting the B samples obtained in S303Separating to obtain input quantity of training setAnd output quantity
S3042, defining a GRU energy consumption prediction model and setting related parameters, wherein the model is designed into three layers, namely a first layer is an input layer, a second layer is a hidden layer, a third layer is an output layer, and the number of neurons in each layer is set;
8. The method for diagnosing equipment abnormality based on energy consumption interval prediction according to claim 2, wherein the step S306 includes: taking the confidence p as 0.95 as an example, for the data in the test set, the predicted values of the B points are obtained2.5% and 97.5%, to obtain a 95% confidence interval, i.e. the prediction interval for the energy consumption value.
9. The method for diagnosing equipment abnormality based on energy consumption interval prediction according to claim 2, wherein S307 includes:
the range probability formula is:
wherein PICP is the range probability;
the normalized interval width formula is:
PINAW is normalized interval width
The comprehensive index formula is as follows:
CWC is a comprehensive index.
10. The method for diagnosing equipment abnormality based on energy consumption interval prediction according to claim 2, wherein said S308 includes: whether the evaluation index meets the requirement or not is judged, and generally, the smaller the PINAW is, the better the PINAW is under the condition that the PICP is more than or equal to mu; the smaller the CWC is, the better the prediction quality of the interval is;
the S309 includes the following contents: predicting the interval of the energy consumption value by utilizing a GRU prediction model; substituting the data x to be evaluated into the prediction model in S304And the point prediction value is obtained and then substituted into S306 to obtain the interval prediction value.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110937060.6A CN113705867A (en) | 2021-08-16 | 2021-08-16 | Equipment abnormity diagnosis method based on energy consumption interval prediction |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110937060.6A CN113705867A (en) | 2021-08-16 | 2021-08-16 | Equipment abnormity diagnosis method based on energy consumption interval prediction |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113705867A true CN113705867A (en) | 2021-11-26 |
Family
ID=78652821
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110937060.6A Pending CN113705867A (en) | 2021-08-16 | 2021-08-16 | Equipment abnormity diagnosis method based on energy consumption interval prediction |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113705867A (en) |
Citations (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20140278165A1 (en) * | 2013-03-14 | 2014-09-18 | Johnson Controls Technology Company | Systems and methods for analyzing energy consumption model data |
US20180270312A1 (en) * | 2017-03-18 | 2018-09-20 | Tata Consultancy Services Limited | Method and system for anomaly detecttion, missing data imputation and consumption prediction in energy data |
CN109615226A (en) * | 2018-12-12 | 2019-04-12 | 焦点科技股份有限公司 | A kind of operation indicator method for monitoring abnormality |
CN110046743A (en) * | 2019-03-06 | 2019-07-23 | 上海交通大学 | Energy Consumption of Public Buildings prediction technique and system based on GA-ANN |
CN110738361A (en) * | 2019-09-29 | 2020-01-31 | 浙江浙大中控信息技术有限公司 | Dynamic peak period prediction method based on line passenger capacity |
JP2020087276A (en) * | 2018-11-30 | 2020-06-04 | アズビル株式会社 | Device and method for predicting energy consumption |
CN111612651A (en) * | 2020-05-27 | 2020-09-01 | 福州大学 | Abnormal electric quantity data detection method based on long-term and short-term memory network |
US20200313434A1 (en) * | 2019-03-28 | 2020-10-01 | Accenture Global Solutions Limited | Predictive power usage monitoring |
AU2020104000A4 (en) * | 2020-12-10 | 2021-02-18 | Guangxi University | Short-term Load Forecasting Method Based on TCN and IPSO-LSSVM Combined Model |
CN112734141A (en) * | 2021-02-25 | 2021-04-30 | 云南电网有限责任公司电力科学研究院 | Diversified load interval prediction method and device |
CN112990556A (en) * | 2021-02-24 | 2021-06-18 | 江苏大学 | User power consumption prediction method based on Prophet-LSTM model |
-
2021
- 2021-08-16 CN CN202110937060.6A patent/CN113705867A/en active Pending
Patent Citations (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20140278165A1 (en) * | 2013-03-14 | 2014-09-18 | Johnson Controls Technology Company | Systems and methods for analyzing energy consumption model data |
US20180270312A1 (en) * | 2017-03-18 | 2018-09-20 | Tata Consultancy Services Limited | Method and system for anomaly detecttion, missing data imputation and consumption prediction in energy data |
JP2020087276A (en) * | 2018-11-30 | 2020-06-04 | アズビル株式会社 | Device and method for predicting energy consumption |
CN109615226A (en) * | 2018-12-12 | 2019-04-12 | 焦点科技股份有限公司 | A kind of operation indicator method for monitoring abnormality |
CN110046743A (en) * | 2019-03-06 | 2019-07-23 | 上海交通大学 | Energy Consumption of Public Buildings prediction technique and system based on GA-ANN |
US20200313434A1 (en) * | 2019-03-28 | 2020-10-01 | Accenture Global Solutions Limited | Predictive power usage monitoring |
CN110738361A (en) * | 2019-09-29 | 2020-01-31 | 浙江浙大中控信息技术有限公司 | Dynamic peak period prediction method based on line passenger capacity |
CN111612651A (en) * | 2020-05-27 | 2020-09-01 | 福州大学 | Abnormal electric quantity data detection method based on long-term and short-term memory network |
AU2020104000A4 (en) * | 2020-12-10 | 2021-02-18 | Guangxi University | Short-term Load Forecasting Method Based on TCN and IPSO-LSSVM Combined Model |
CN112990556A (en) * | 2021-02-24 | 2021-06-18 | 江苏大学 | User power consumption prediction method based on Prophet-LSTM model |
CN112734141A (en) * | 2021-02-25 | 2021-04-30 | 云南电网有限责任公司电力科学研究院 | Diversified load interval prediction method and device |
Non-Patent Citations (3)
Title |
---|
于水: "建筑室内污染物源辨识及控制", 31 August 2019, 重庆大学出版社, pages: 119 - 122 * |
侯松宝;王侃宏;石凯波;孔力;曹辉;: "基于相似日和主成分分析的光伏发电系统短期出力预测", 可再生能源, no. 01, pages 19 - 25 * |
韦善阳等: "基于改进权值优化模型的光伏功率区间预测", 电力建设, pages 26 - 33 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108898251B (en) | Offshore wind farm power prediction method considering meteorological similarity and power fluctuation | |
CN110232203B (en) | Knowledge distillation optimization RNN short-term power failure prediction method, storage medium and equipment | |
CN116625438B (en) | Gas pipe network safety on-line monitoring system and method thereof | |
CN106779129A (en) | A kind of Short-Term Load Forecasting Method for considering meteorologic factor | |
CN109389238B (en) | Ridge regression-based short-term load prediction method and device | |
CN105868557A (en) | Online prediction method for remaining life of electromechanical equipment under situation of two-stage degradation | |
Ukkonen et al. | Evaluation of thunderstorm predictors for Finland using reanalyses and neural networks | |
CN111680454A (en) | Fan blade icing fault prediction method based on double attention mechanism | |
JP6185206B1 (en) | Method and apparatus for detecting abnormalities or signs of abnormalities in a photovoltaic system | |
CN112149877B (en) | Multi-source data driven fault prediction method and system for multi-element complex urban power grid | |
CN112132394B (en) | Power plant circulating water pump predictive state evaluation method and system | |
CN103440410A (en) | Main variable individual defect probability forecasting method | |
CN112598144A (en) | CNN-LSTM burst fault early warning method based on correlation analysis | |
CN111080039A (en) | Fan cluster fault prediction method and system | |
CN114169254A (en) | Abnormal energy consumption diagnosis method and system based on short-term building energy consumption prediction model | |
CN115470850A (en) | Water quality abnormal event recognition early warning method based on pipe network water quality time-space data | |
CN115545333A (en) | Method for predicting load curve of multi-load daily-type power distribution network | |
CN114357670A (en) | Power distribution network power consumption data abnormity early warning method based on BLS and self-encoder | |
CN112349420B (en) | Method and system for constructing disease prediction model based on meteorological data | |
CN116756505A (en) | Photovoltaic equipment intelligent management system and method based on big data | |
CN117494009A (en) | Electrical equipment state evaluation method based on insulating material pyrolysis analysis and cloud platform | |
CN110852906B (en) | Method and system for identifying electricity stealing suspicion based on high-dimensional random matrix | |
CN113705867A (en) | Equipment abnormity diagnosis method based on energy consumption interval prediction | |
CN110533213A (en) | Transmission line of electricity defect Risk Modeling and its prediction technique based on support vector machines | |
CN115577854A (en) | Quantile regression wind speed interval prediction method based on EEMD-RBF combination |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |