CN108171408A - A kind of sewage water and water yield modeling method - Google Patents
A kind of sewage water and water yield modeling method Download PDFInfo
- Publication number
- CN108171408A CN108171408A CN201711373832.8A CN201711373832A CN108171408A CN 108171408 A CN108171408 A CN 108171408A CN 201711373832 A CN201711373832 A CN 201711373832A CN 108171408 A CN108171408 A CN 108171408A
- Authority
- CN
- China
- Prior art keywords
- flow data
- daily flow
- water outlet
- water inlet
- water
- 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.)
- Granted
Links
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 title claims abstract description 337
- 238000000034 method Methods 0.000 title claims abstract description 58
- 239000010865 sewage Substances 0.000 title claims abstract description 45
- 238000012417 linear regression Methods 0.000 claims abstract description 31
- 230000002159 abnormal effect Effects 0.000 claims abstract description 12
- 238000001744 unit root test Methods 0.000 claims description 21
- 230000005856 abnormality Effects 0.000 claims description 14
- 238000012360 testing method Methods 0.000 claims description 9
- 235000013399 edible fruits Nutrition 0.000 claims description 2
- 238000012544 monitoring process Methods 0.000 abstract description 10
- 238000000342 Monte Carlo simulation Methods 0.000 description 2
- 241000209094 Oryza Species 0.000 description 2
- 235000007164 Oryza sativa Nutrition 0.000 description 2
- 238000013459 approach Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 235000009566 rice Nutrition 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 230000003416 augmentation Effects 0.000 description 1
- 230000003190 augmentative effect Effects 0.000 description 1
- 238000000205 computational method Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000012731 temporal analysis Methods 0.000 description 1
- 238000000700 time series analysis Methods 0.000 description 1
- 239000003643 water by type Substances 0.000 description 1
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
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
-
- 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—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
Landscapes
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Educational Administration (AREA)
- Economics (AREA)
- Development Economics (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Marketing (AREA)
- Physics & Mathematics (AREA)
- Entrepreneurship & Innovation (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Alarm Systems (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The invention belongs to field of urban drainage, provide a kind of sewage water and water yield modeling method, including obtaining the daily flow data of water outlet and the daily flow data of water inlet, and judge time series stationarity or whether to meet same order list whole, linear regression formula is established, obtains the sequence of the corresponding residual error of linear regression;Then using linear regression formula and the forecast interval determined according to regression residuals sequence, judge specific water outlet daily flow data and corresponding water inlet daily flow data with the presence or absence of abnormal.Method using the present invention can be established and accurate sewage water and go out water model, and carrying out accurately sewage disposal to sewage disposal enterprise controls, and improve the efficiency of sewage disposal system and energy-saving and emission-reduction and reduce cost and have great directive significance;To environmentally friendly supervisory department, dynamically and in real time whether monitoring sewage disposal system works normally simultaneously, provides objective, direct and believable monitoring index.
Description
Technical field
The invention belongs to field of urban drainage, more particularly to a kind of sewage water and water yield modeling method.
Background technology
In sewage disposal process because the water of sewage is random, and the water quality of sewage also when change,
There are many variation ranges.Therefore, it under the premise of by sewage qualified discharge, is simply built according to sewage treatment process
, often there is very large deviation between the data of actual motion in the relational model between the water and water yield of vertical sewage.It is and accurate
True sewage water carries out accurately sewage disposal with going out water model, for sewage disposal enterprise and controls, and improves sewage disposal system
The efficiency of system and energy-saving and emission-reduction and cost etc. is reduced, there is great directive significance.Meanwhile accurate sewage water
With going out water model, for environmentally friendly supervisory department, dynamically and in real time whether monitoring sewage disposal system works normally, and provides
One objective, direct and believable monitoring index.
It therefore, need a kind of method accurately and reliably modeled to sewage water and water yield in practice.
Invention content
In order to solve the above-mentioned problems in the prior art, the present invention provides a kind of sewage waters to build with water yield
Mould method, including:
1) the daily flow data of water outlet and the daily flow data of water inlet are obtained;
2) judgment step 1) obtain the water outlet daily flow data and the water inlet daily flow data when
Between sequence stationary;
If the time series of the daily flow data of the water outlet and the daily flow data of the water inlet is steady,
Carry out following steps 4);
If the time series of the daily flow data of the water outlet and the daily flow data of the water inlet is unstable,
Then carry out following steps 3);
If 3) time series of the daily flow data of the daily flow data and water inlet of the water outlet is uneven
Surely, judge the daily flow data of the water outlet and the daily flow data of the water inlet whether to meet same order list whole, if symbol
Contract rank is singly whole, then carries out following steps 4), if it is whole not meet same order list, terminate;
If 4) time series of the daily flow data of the daily flow data and water inlet of the water outlet is steady,
Or if the daily flow data of the water outlet and the daily flow data fit same order list of the water inlet are whole, to described
The daily flow data of water outlet and the daily flow data of the water inlet are fitted, and obtain following linear regression formula:
Y=ax+ ε
In above formula, y is the water outlet daily flow data, and x is the water inlet daily flow data, and ε is regression residuals sequence
Row, and 0.9 < a < 1;
Then, it will be calculated in the daily flow data of water inlet substitution institute linear regression formula y=ax respectively pre-
The daily flow data value of the water outlet of the prediction is subtracted the practical water outlet by the daily flow data value of the water outlet of survey
Daily flow data, obtain the sequence ε of the corresponding residual error of linear regression;
If by the daily flow data of the water outlet and the daily flow data fit same order list of the water inlet is whole obtains
The linear regression formula obtained, then judge the stationarity of regression residuals sequence ε, if regression residuals sequence ε is steady, carry out such as
Lower step 5), if regression residuals sequence ε is unstable, terminates;
5) linear regression formula obtained using step 4), by some practical specific water outlet daily flow data
It is substituted into the linear regression formula with corresponding water inlet daily flow data, regression residuals is calculated, judge the recurrence
Whether residual error is fallen in forecast interval, if the regression residuals are not fallen in forecast interval, output abnormality alarm, and explanation
The specific water outlet daily flow data and corresponding water inlet daily flow data exist abnormal;If the regression residuals
Fall in forecast interval, then not output abnormality alarm, be judged as the specific water outlet daily flow data and it is corresponding enter
Mouth of a river daily flow data are normal.
The initial data of the flow velocity of water outlet is preferably based on, statistics summarizes the daily flow number for obtaining the water outlet
According to;
The initial data of flow velocity based on water inlet, statistics summarize the daily flow data for obtaining the water inlet.
Preferably, the initial data of the flow velocity of the initial data of the flow velocity of the water outlet and the water inlet be
In one consecutive days, the average value of the flow velocity counted as unit of 30 seconds.
Preferably, in step 2), the water outlet of the step 1) acquisition is judged using the method for ADF unit root tests
The time series stationarity of the daily flow data of mouth and the daily flow data of the water inlet;
Preferably, the ADF unit root tests process is:Assuming that the daily flow data of the water outlet and it is described enter water
The time series of the daily flow data of mouth is unstable, and under corresponding test statistics, the distribution of the test statistics is carried out
Monte Carlo simulation obtains the probability distribution of inspected number, and calculates the P values of inspected number value under the sample conditions, as injustice
The probability (P values) that the actual conditions occur under the conditions of steady if P values are less than 0.05, illustrates the probability pole that the hypothesis occurs
It is small, that is, illustrate that the time series of the daily flow data of the water outlet and the daily flow data of the water inlet is steady.
Preferably, in step 3), judge that the daily flow data of the water outlet and the daily flow data of the water inlet are
The no whole method of same order list that meets is:
Judge the daily flow data of the water outlet and the daily flow data of the water inlet from first-order difference sequence respectively
The stationarity of the multistage difference sequence started;If the multistage difference sequence of the daily flow data of the water outlet is smoothly poor
Sublevel number is identical with the stable difference order of multistage difference sequence of the daily flow data of the water inlet, then goes out described in explanation
The daily flow data at the mouth of a river and the daily flow data fit same order list of the water inlet are whole.
Preferably, using the method for ADF unit root tests judge respectively the water outlet daily flow data and it is described enter
The stationarity of multistage difference sequence of the daily flow data at the mouth of a river since first-order difference sequence;
Preferably, the ADF unit root tests process is:Assume initially that the daily flow data of the water outlet and described
First-order difference sequence of the daily flow data of water inlet since first-order difference sequence is unstable, then calculates hypothesis generation
Probability, if probability be less than 0.01, illustrate the hypothesis occur probability it is minimum, that is, illustrate the daily flow of the water outlet
The first-order difference sequence stationary of data and the daily flow data of the water inlet;Similarly, go out described in judging in the method successively
The second differnce sequence of the daily flow data at the mouth of a river and the daily flow data of the water inlet, third order difference sequence, Four order difference
The stationarity of sequence etc..
Preferably, in step 4), judge that the regression residuals sequence ε's is steady using the method for ADF unit root tests
Property;
Preferably, the ADF unit root tests process is:Assume initially that the regression residuals sequence ε is unstable, then
The probability of hypothesis generation is calculated, if the probability is less than 0.01, illustrates that the probability that the hypothesis occurs is minimum, that is, illustrates
The regression residuals sequence ε is steady.
Preferably, the forecast interval is the range between the 10%-75% quantiles of the regression residuals sequence ε, excellent
The range being selected as between 25%-65% quantiles.
Preferably, the specific water outlet daily flow data described in step 5) and corresponding water inlet day are flowed
Amount data substitute into the residual error being calculated in the linear regression formula that step 4) obtains and add in the regression residuals sequence ε, so
After redefine the forecast interval;Again by next specific water outlet daily flow data and corresponding water inlet day
Data on flows, which is substituted into the linear regression formula, is calculated residual error, and judges whether the residual error falls into described redefine
In forecast interval, if the residual error is not fallen in the forecast interval redefined, output abnormality alarm illustrates this
Next specific water outlet daily flow data and corresponding water inlet daily flow data exist abnormal;If the residual error
It falls in the forecast interval redefined, then not output abnormality alarm, is judged as next specific water outlet daily flow
Data and corresponding water inlet daily flow data are normal;According to the continuous iteration of the mode of operation, new tool is judged successively
The water outlet daily flow data of body and corresponding water inlet daily flow data are with the presence or absence of abnormal.
Using sewage water provided by the invention and water yield modeling method, accurate sewage water can be established with going out
Water model carries out accurately sewage disposal to sewage disposal enterprise and controls, and improves the efficiency of sewage disposal system and energy saving
Emission reduction and cost is reduced with great directive significance;Environmentally friendly supervisory department dynamically and is in real time monitored at sewage simultaneously
Whether reason system works normally, and provides objective, direct and believable monitoring index.
Other features and advantages of the present invention will be in following specific embodiment part detailed description.
Description of the drawings
The accompanying drawings which form a part of this application are used to provide further understanding of the present invention, of the invention
Illustrative embodiments and their description do not constitute improper limitations of the present invention for explaining the present invention.
Fig. 1 is the flow diagram of sewage water and water yield modeling method that the preferred embodiment of the present invention provides.
Fig. 2 be the embodiment of the present invention 1 regression residuals sequence with according to the time (my god) order of occurrence arrangement observation number
According to relational graph.
Specific embodiment
To make the object, technical solutions and advantages of the present invention clearer, below in conjunction with implementation of the attached drawing to the present invention
Mode is described in further detail.It should be understood that the specific embodiments described herein are only used for describing and explaining
The present invention is not intended to restrict the invention.
The present invention is proposed a kind of monitoring data run based on actual sewage processing system and carries out data analysis, so as to
The method for establishing sewage water and water yield model, i.e. sewage water and water yield modeling method.
The sewage water of the present invention and water yield modeling method are that the monitoring data of timing sampling are considered as time series,
The relation of long standing relation of sewage water and water yield is simulated using association's adjusting method of time series, so as to establish effective dirt
Water carrys out the model of water and water yield.Specific modeling approach is as follows:
1st, water and the corresponding time relationship of water flow
The period that monitoring in view of water and water outlet samples is generally shorter, it is contemplated that being adjusted to time dimension
Section, that is, time granularity is amplified, by originally with " second " rank count flow indicator carried out statistics summarize, so as to formed with
" my god " rank statistics flow indicator, so as to be converted to research daily corresponding water with water outlet relationship.
2nd, water and the long-term quantitative relation of water flow
In order to study the quantitative relation between water and water flow, unquestionable most straightforward approach is exactly back
Return, but classical recurrences be related to many it is assumed that such as changing over time, variable held stationary and mean value, variance are constant.
But in actual conditions, time series is often unstable, if the method returned also according to general linear is calculated, pole
It is possible that there is the problem of shadowing property.
If two groups of time serieses are unstable, but the linear combination between two groups of time serieses, which can obtain one, to be put down
Steady sequence, then claim this two groups of time series associations whole, there is relationship steady in a long-term.With the variation of time, practical relationship
Relationship with model prediction is there may be deviation, but deviateing can disappear quickly, and relationship between variables tend to be steady again.Therefore, it assists
Complete machine system meets to the detection demand of water balance in sewage disposal system, so this method is taken to close water and water flow
System is simulated.
3rd, water and water flow monitoring
After assisting whole method that water and water flow relationship is described, next need to actual operation
When relationship be compared with the relationship that prediction obtains, provide early warning in time in the case of deviation.
Specifically, with reference to figure 1, sewage water of the invention includes the following steps with water yield modeling method:
1) the initial data v of the flow velocity of water outlet is obtained1With the initial data v of the flow velocity of water inlet2。
The initial data v of the flow velocity of water outlet based on acquisition1, count the daily flow data v for summarizing and obtaining water outlet1t。
The initial data v of the flow velocity of water inlet based on acquisition2, count the daily flow data v for summarizing and obtaining water inlet2t。
For example, in a kind of specific embodiment, the initial data v of the flow velocity of water outlet1With the original of the flow velocity of water inlet
Beginning data v2Be the average value of the flow velocity within a consecutive days, counted as unit of 30 seconds, then the daily flow number of water outlet
According to v1T is 2880*v1, the daily flow data v of water inlet2T is 2880*v2。
2) judge the daily flow data v of water outlet1The daily flow data v of t and water inlet2The time series stationarity of t.
Wherein, judge the daily flow data v of water outlet1The daily flow data v of t and water inlet2The time series of t it is steady
ADF unit root tests (Augmented Dickey-Fuller unit root test augmentation Dickey-fowler list may be used in property
Position root examine) method.Specifically checkout procedure can be:Assume initially that the daily flow data v of water outlet1T and water inlet
Daily flow data v2The time series of t is unstable, and under corresponding test statistics, the distribution of the test statistics is covered
Special Carlow simulation, obtains the probability distribution of inspected number, and calculate the P values of inspected number value under the sample conditions, as unstable
Under the conditions of the actual conditions occur probability (P values), if P values be less than 0.05, illustrate the hypothesis occur probability it is minimum,
Illustrate the daily flow data v of water outlet1The daily flow data v of t and water inlet2The time series of t is steady.
Wherein, all ADF unit root tests processes involved in the present invention and P value calculating methods are:(king can be referred to
It is gorgeous to write《Applied time series analysis》, publishing house of the Renmin University of China, July the 1st edition in 2005, the 2nd chapter and the 3rd chapter.)
1. by time series { x1,x2,…,xtBe expressed as:
2. it enablesIf time series { xtSteadily, it is equivalent to ρ<0
3.AR (p) process per unit root null hypothesis conditions are,
H0:ρ=0 (sequence x0Non-stationary)
H0:ρ<0 (sequence x0Steadily)
4. test statistics is:
5. under null hypothesis, by computer Monte-carlo Simulation Method, the probability distribution of test statistics is obtained (i.e.
Tables of critical values)
6. test statistics is substituted into actual sample value, P values are calculated:
If the daily flow data v of water outlet1The daily flow data v of t and water inlet2The time series of t is steady, then carries out
Following steps 4);
If the daily flow data v of water outlet1The daily flow data v of t and water inlet2The time series of t is unstable, then into
Row following steps 3).
If 3) the daily flow data v of water outlet1The daily flow data v of t and water inlet2The time series of t is unstable, then
Judge the daily flow data v of water outlet1The daily flow data v of t and water inlet2It is whole whether t meets same order list, if meeting same order
Single whole, then carry out following steps 4), if it is whole not meet same order list, terminate.
Wherein, judge the daily flow data v of water outlet1The daily flow data v of t and water inlet2It is whole whether t meets same order list
Method can be:
The daily flow data v of water outlet is judged respectively1The daily flow data v of t and water inlet2T is opened from first-order difference sequence
The stationarity of the multistage difference sequence to begin;If the daily flow data v of water outlet1The stable difference rank of multistage difference sequence of t
The daily flow data v of number and water inlet2The stable difference order of multistage difference sequence of t is identical, then illustrates the day stream of water outlet
Measure data v1The daily flow data v of t and water inlet2It is whole that t meets same order list;If for example, daily flow data v of water outlet1T's
Second differnce sequence stationary, while the daily flow data v of water inlet2The second differnce sequence of t is also steady, then illustrates water outlet
Daily flow data v1The daily flow data v of t and water inlet2It is whole that t meets same order list.
Wherein, judge the daily flow data v of water outlet1The daily flow data v of t and water inlet2T is opened from first-order difference sequence
The method that ADF unit root tests may be used in the stationarity of the multistage difference sequence to begin.
Wherein, first-order difference is the difference of continuous adjacent two in discrete function;If for example, define X (k), Y (k)=
X (k+1)-X (k) is exactly the first-order difference of function X (k);Similarly, the first-order difference of Y (k) is Z (k)=Y (k+1)-Y (k)=X
(k+2) -2*X (k+1)+X (k), Z (k) are the second differnce of function X (k);Third order difference, Four order difference etc. and so on.
It corresponds, first-order difference sequence is exactly the sequence that the difference of continuous adjacent two in discrete function is formed, i.e. first-order difference shape
Into sequence;Similarly, second differnce sequence is exactly the sequence that second differnce is formed;Third order difference sequence, Four order difference sequence etc.
Wait and so on.
Specifically, judge the daily flow data v of water outlet1The daily flow data v of t and water inlet2T is from first-order difference sequence
The method of ADF unit root tests of the stationarity of the multistage difference sequence started can be:
Assume initially that the daily flow data v of water outlet1The daily flow data v of t and water inlet2T is opened from first-order difference sequence
The first-order difference sequence of beginning is unstable, then calculates the probability (P values) of hypothesis generation, if P values are less than 0.01, illustrates
The probability that the hypothesis occurs is minimum, that is, illustrates the daily flow data v of water outlet1The daily flow data v of t and water inlet2The single order of t
Difference sequence is steady.Similarly, judge the daily flow data v of water outlet in the method successively1The daily flow number of t and water inlet
According to v2The stationarity of the second differnce sequence of t, third order difference sequence, Four order difference sequence etc..
If 4) the daily flow data v of water outlet1The daily flow data v of t and water inlet2The time series of t it is steady (that is,
The daily flow data v of water outlet is assert after step 2 judgement1The daily flow data v of t and water inlet2The time series of t is put down
Surely) or if the daily flow data v of water outlet1The daily flow data v of t and water inlet2T meet same order list it is whole (that is, by
Step 3 assert the daily flow data v of water outlet after judging1The daily flow data v of t and water inlet2It is whole that t meets same order list), then it is right
The daily flow data v of water outlet1The daily flow data v of t and water inlet2T is fitted, and obtains following linear regression formula:
Y=ax+ ε
In above formula, y is water outlet daily flow data v1T, x are water inlet daily flow data v2T, ε are corresponding residual to return
The sequence (regression residuals sequence) of difference, and 0.9 < a < 1.
Then, respectively by the daily flow data v of water inlet2T is substituted into linear regression formula y=ax and prediction is calculated
Water outlet daily flow data value, by the daily flow data value of the water outlet of the prediction subtract practical water outlet day flow
Measure data v1T obtains the sequence ε (referred to as " regression residuals sequence ") of the corresponding residual error of linear regression.
If the daily flow data v by water outlet1The daily flow data v of t and water inlet2T meet same order list it is whole (that is,
The daily flow data v of water outlet is assert after step 3 judgement1The daily flow data v of t and water inlet2It is whole that t meets same order list),
The above-mentioned linear regression formula obtained then also needs to judge the stationarity of regression residuals sequence ε, if regression residuals sequence ε is put down
Surely, then following steps 5 can be carried out), if regression residuals sequence ε is unstable, terminate.
Wherein, judge the method that ADF unit root tests may be used in the stationarity of regression residuals sequence ε.Specifically, may be used
To assume initially that regression residuals sequence ε is unstable, the probability (P values) of hypothesis generation is then calculated, if P values are less than 0.01,
Then illustrate that the probability that the hypothesis occurs is minimum, that is, illustrate that regression residuals sequence ε is steady.Wherein, the circular of P values is such as
Upper described, details are not described herein again.
5) linear regression formula obtained using step 4), by some practical specific water outlet daily flow data
It is substituted into the linear regression formula with corresponding water inlet daily flow data, regression residuals is calculated, judge the recurrence
Whether residual error is fallen in forecast interval, if the regression residuals are not fallen in forecast interval, output abnormality alarm, and explanation
The specific water outlet daily flow data and corresponding water inlet daily flow data exist abnormal;If the regression residuals
Fall in forecast interval, then not output abnormality alarm, be judged as the specific water outlet daily flow data and it is corresponding enter
Mouth of a river daily flow data are normal.
Wherein, range of the forecast interval between the 10%-75% quantiles of regression residuals sequence ε, preferably 25%-
Range between 65% quantile.Wherein, " quantile " refers to arrange regression residuals sequence according to numerical value from small to large
After sequence, the numerical value of corresponding position is come, for example, 25% quantile refers to regression residuals sequence according to number from small to large
After value is ranked up, it is exactly 25% quantile to come at the 25%th position, 75% quantile refer to regression residuals sequence by
After being ranked up according to numerical value from small to large, it is exactly 75% quantile to come at the 75%th position.
Preferably, above-mentioned specific water outlet daily flow data and corresponding water inlet daily flow data are substituted into
The residual error being calculated in above-mentioned linear regression formula is added in regression residuals sequence ε, then redefines forecast interval, then
Next specific water outlet daily flow data are obtained using aforesaid way and corresponding water inlet daily flow data substitute into
The residual error being calculated in the linear regression formula, and judge whether the residual error is fallen into the forecast interval redefined, such as
The fruit residual error is not fallen in the forecast interval redefined, then output abnormality alarm, illustrates next specific water outlet
Daily flow data and corresponding water inlet daily flow data exist abnormal;If the residual error is fallen in the prediction redefined
In section, then not output abnormality alarm, be judged as next specific water outlet daily flow data and it is corresponding enter
Mouth of a river daily flow data are normal.According to the continuous iteration of the mode of operation, new specific water outlet daily flow data are judged successively
With corresponding water inlet daily flow data with the presence or absence of abnormal.
Since the present invention can also be by the daily flow data of water outlet after tested and the daily flow data of water inlet
(that is, daily flow data of water outlet and the daily flow data of water inlet after being judged by the use of the above method) are as in step 4)
Obtain the daily flow data v of the water outlet of linear regression formula1The daily flow data v of t and water inlet2T carrys out further modified line
Property regression formula, continuous iteration successively, so as to be continuously improved to the daily flow data of final water outlet and flow the day of water inlet
Measure the accuracy of the prediction of data.
Sewage water using the present invention and water yield modeling method can establish accurate sewage water and water outlet mould
Type carries out accurately sewage disposal to sewage disposal enterprise and controls, to improving the efficiency of sewage disposal system and energy saving subtracting
Cost is arranged and reduced with great directive significance;Environmentally friendly supervisory department can also dynamically and in real time be monitored simultaneously dirty
Whether water treatment system works normally, and provides objective, direct and believable monitoring index.
Embodiment 1
The present embodiment is used for the sewage water for illustrating the present invention and water yield modeling method.
1st, data source and pretreatment
The data for choosing certain water factory's sewage disposal system September part discrepancy water flow are studied, since sensor was with 30 seconds
It is passed back for interval, so water outlet flow velocity and water inlet flow velocity all have 86400 observational records (to have 2880 sights daily respectively
Survey record), but interfered since sensor exists, wherein the time point value for not passing number is flagged as 0, it is assumed that go out, water inlet
Reading all be 0 situation to flow velocity simultaneously is to be disturbed and do not pass number situation, and numerical value NA is replaced.Then it is calculated by the hour per small
When average flow rate, then count the flow total value of daily water outlet and the flow total value of water inlet by the hour, and then respectively
To the daily flow data v of the water outlets of 30 days1The daily flow data v of t and water inlet2(such as the following table 1, the unit of flow are all t
Cubic meter).
Table 1 September part daily flow data for going out, entering water of the 1-20 days
2nd, judge the daily flow data (v of water outlet1And the daily flow data (v of water inlet t)2T) time series is steady
Property.
ADF unit root tests are carried out to water outlet daily flow data and water inlet daily flow data respectively, it can be found that going out
Mouth of a river daily flow data and the P values of water inlet daily flow data detection are respectively 0.5498 and 0.4546, are all higher than 0.05 (tool
Body computational methods are referring to specification, and details are not described herein again), therefore, water outlet daily flow data and water inlet daily flow data
Time series is unstable, it is therefore desirable to judge whether the daily flow data of water outlet and the daily flow data of water inlet meet together
Rank is singly whole.
Multistage difference sequence point to water outlet daily flow data and water inlet daily flow data from first-order difference respectively
Not carry out ADF unit root tests, obtain the P of the first-order difference sequence of water outlet daily flow data and water inlet daily flow data
Value is respectively less than 0.01 (for circular referring to specification, details are not described herein again), illustrates water outlet daily flow data and enters water
It is whole that mouth daily flow data obey same order list.
3rd, equation of linear regression is established
Using water outlet daily flow data as y variables, water inlet daily flow data carry out linear regression for x variables, obtain line
Property equation is:
Y=0.97872x+ ε
In above formula, ε is the sequence for returning corresponding residual error.
Carrying out ADF unit root tests to regression residuals sequence ε, (circular is joined it can be found that P values are less than 0.01
See specification, details are not described herein again), illustrate that regression residuals sequence ε is steady (that is, regression error is steady), water outlet flow is with entering
Mouth of a river flow meets the long-run equilibrium relationship that above formula is shown.
4th, the residual error range of prediction is provided using history residual error
Each quantile of residual error, and the sequencing observed with Water Exit flow-time are calculated, what it was calculated is residual
Difference sequence ε according to the time (my god) carry out image shows (as shown in Figure 2), it can be found that the mean value of y- 0.97872x for -1269 stand
Square rice, median (being ranked up from small to large to regression residuals sequence, come centre position is exactly median) are vertical for -314
Square rice, 25% quantile (are ranked up regression residuals sequence, come the 25%th position is exactly 25% point of position from small to large
Number) it is -3199.43 cubic metres, 75% quantile (is ranked up regression residuals sequence, comes the 75%th position from small to large
Be exactly 75% quantile) for 356.51 cubic metres, numerical value is concentrated mainly on 0 left side.
Therefore, when estimating the section for sending out alarm, be considered as 10% quantile of history residual error-
3907.61 (being ranked up from small to large to regression residuals sequence, come the 10%th position is exactly 10% quantile) and 75%
Quantile 356.52 is used as following entrance water relationship, the i.e. estimation range of y-0.97872x, while true residual obtaining
After difference, and quantile is updated again, continuous iteration.When range of the practical residual error range beyond prediction, it will send out police
Report.
5th, practical proof
The water outlet daily flow data of the 21-30 days and corresponding water inlet daily flow data are substituted into y- respectively
In 0.97872x, daily residual error is calculated, if residual error except -3907.61~356.52, sends out alarm, otherwise
It is judged as normal.
Table 2 September part daily flow data for going out, entering water of the 21-30 days
In the regression residuals sequence ε that the residual error substitution of the 21st day can also be the previously calculated, then redefine back
Return the range between the 10%-75% quantiles of residual sequence ε as forecast interval, then obtained the 22nd day using aforesaid way
Water outlet daily flow data and corresponding water inlet daily flow data bring what is be calculated in the linear regression formula into
Residual error, and judge whether the residual error is fallen into the above-mentioned forecast interval redefined, it is redefined if the residual error is not fallen within
Forecast interval in, then output abnormality alarm illustrates the water outlet daily flow data of the 22nd day and corresponding water inlet day
Data on flows exists abnormal;If the residual error is fallen in the forecast interval redefined, not output abnormality alarm, it is judged as
The water outlet daily flow data of the 22nd day and corresponding water inlet daily flow data are normal.It is continuous according to the mode of operation
Iteration, the water outlet daily flow data and corresponding water inlet daily flow data for judging the 23-30 days successively whether there is
It is abnormal.
As known by the technical knowledge, the present invention can pass through the implementation of other essence without departing from its spirit or essential feature
Scheme is realized.Therefore, embodiment disclosed above, all things considered are all merely illustrative, not the only.
All changes within the scope of the invention or within the scope equivalent to the present invention are included in the invention.
Claims (9)
1. a kind of sewage water and water yield modeling method, including:
1) the daily flow data of water outlet and the daily flow data of water inlet are obtained;
2) judgment step 1) obtain the water outlet daily flow data and the water inlet daily flow data time sequence
Row stationarity;
If the time series of the daily flow data of the water outlet and the daily flow data of the water inlet is steady, carry out such as
Lower step 4);
If the time series of the daily flow data of the water outlet and the daily flow data of the water inlet is unstable, carry out
Following steps 3);
If 3) time series of the daily flow data of the daily flow data and water inlet of the water outlet is unstable, judge
It is whole whether the daily flow data of the water outlet and the daily flow data of the water inlet meet same order list, if meeting same order list
Whole, then carry out following steps 4), if it is whole not meet same order list, terminate;
If 4) time series of the daily flow data of the daily flow data and water inlet of the water outlet is steady, Huo Zheru
The daily flow data of water outlet described in fruit and the daily flow data fit same order list of the water inlet are whole, then to the water outlet
Daily flow data and the daily flow data of the water inlet are fitted, and obtain following linear regression formula:
Y=ax+ ε
In above formula, y is the water outlet daily flow data, and x is the water inlet daily flow data, and ε is regression residuals sequence, and
0.9 < a < 1;
Then, the daily flow data of the water inlet are substituted into institute linear regression formula y=ax respectively and prediction is calculated
The daily flow data value of the water outlet of the prediction is subtracted the day stream of the practical water outlet by the daily flow data value of water outlet
Data are measured, obtain the sequence ε of the corresponding residual error of linear regression;
If the whole acquisition of daily flow data fit same order list by the daily flow data and water inlet of the water outlet
Linear regression formula then judges the stationarity of regression residuals sequence ε, if regression residuals sequence ε is steady, carries out following steps
5), if regression residuals sequence ε is unstable, terminate;
5) linear regression formula obtained using step 4), by some practical specific water outlet daily flow data and therewith
Corresponding water inlet daily flow data are substituted into the linear regression formula, and regression residuals are calculated, and judge that the regression residuals are
No to fall in forecast interval, if the regression residuals are not fallen in forecast interval, output abnormality alarm illustrates that this is specific
Water outlet daily flow data and corresponding water inlet daily flow data exist abnormal;If the regression residuals are fallen in Target area
In, then not output abnormality alarm is judged as the specific water outlet daily flow data and corresponding water inlet daily flow
Data are normal.
2. according to the method described in claim 1, it is characterized in that:
The initial data of flow velocity based on water outlet, statistics summarize the daily flow data for obtaining the water outlet;
The initial data of flow velocity based on water inlet, statistics summarize the daily flow data for obtaining the water inlet.
3. according to the method described in claim 2, it is characterized in that:
The initial data of the flow velocity of the initial data of the flow velocity of the water outlet and the water inlet be within a consecutive days,
The average value of the flow velocity counted as unit of 30 seconds.
4. according to the method described in claim 1, it is characterized in that:
In step 2), the daily flow number of the water outlet of the step 1) acquisition is judged using the method for ADF unit root tests
According to the time series stationarity of the daily flow data with the water inlet;
Preferably, the ADF unit root tests process is:Assuming that the daily flow data of the water outlet and the day of the water inlet
The time series of data on flows is unstable, and under corresponding test statistics, Meng Teka is carried out to the distribution of the test statistics
Lip river is simulated, and the probability distribution of inspected number is obtained, and calculate the P values of inspected number value under the sample conditions, as unstable condition
The probability (P values) that the lower actual conditions occur if P values are less than 0.05, illustrate that the probability that the hypothesis occurs is minimum, that is, illustrates
The time series of the daily flow data of the water outlet and the daily flow data of the water inlet is steady.
5. according to the method described in claim 1, it is characterized in that:
In step 3), judge whether the daily flow data of the water outlet and the daily flow data of the water inlet meet same order list
Whole method is:
Judge respectively the water outlet daily flow data and the water inlet daily flow data since first-order difference sequence
Multistage difference sequence stationarity;If the stable difference order of multistage difference sequence of the daily flow data of the water outlet
It is identical with the stable difference order of multistage difference sequence of the daily flow data of the water inlet, then illustrate the day of the water outlet
The daily flow data fit same order list of data on flows and the water inlet is whole.
6. according to the method described in claim 5, it is characterized in that:
Judge the daily flow data of the water outlet and the daily flow of the water inlet respectively using the method for ADF unit root tests
The stationarity of multistage difference sequence of the data since first-order difference sequence;
Preferably, the ADF unit root tests process is:Assume initially that the daily flow data of the water outlet and the water inlet
First-order difference sequence of the daily flow data since first-order difference sequence it is unstable, then calculate the probability of hypothesis generation,
If probability is less than 0.01, illustrates that the probability that the hypothesis occurs is minimum, that is, illustrate daily flow data and the institute of the water outlet
State the first-order difference sequence stationary of the daily flow data of water inlet;Similarly, judge the day of the water outlet in the method successively
The second differnce sequence of data on flows and the daily flow data of the water inlet, third order difference sequence, Four order difference sequence etc.
Stationarity.
7. according to the method described in claim 1, it is characterized in that:
In step 4), the stationarity of the regression residuals sequence ε is judged using the method for ADF unit root tests;
Preferably, the ADF unit root tests process is:Assume initially that the regression residuals sequence ε is unstable, then calculating should
Assuming that the probability occurred, if the probability is less than 0.01, illustrates that the probability that the hypothesis occurs is minimum, that is, illustrates the recurrence
Residual sequence ε is steady.
8. according to the method described in claim 1, it is characterized in that:
Range of the forecast interval between the 10%-75% quantiles of the regression residuals sequence ε, preferably 25%-
Range between 65% quantile.
9. it according to the method described in claim 1, further includes:
Specific water outlet daily flow data described in step 5) and corresponding water inlet daily flow data are substituted into step
The residual error being calculated in the rapid linear regression formula 4) obtained is added in the regression residuals sequence ε, then redefines institute
State forecast interval;Again substituting into next specific water outlet daily flow data and corresponding water inlet daily flow data should
Residual error is calculated in linear regression formula, and judges whether the residual error is fallen into the forecast interval redefined, if
The residual error is not fallen in the forecast interval redefined, then output abnormality alarm illustrates next specific water outlet
Mouth daily flow data and corresponding water inlet daily flow data exist abnormal;If the residual error is fallen in the prediction redefined
In section, then not output abnormality alarm, be judged as next specific water outlet daily flow data and it is corresponding enter water
Mouth daily flow data are normal;According to the continuous iteration of the mode of operation, new specific water outlet daily flow data are judged successively
With corresponding water inlet daily flow data with the presence or absence of abnormal.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711373832.8A CN108171408B (en) | 2017-12-19 | 2017-12-19 | A kind of sewage water and water yield modeling method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711373832.8A CN108171408B (en) | 2017-12-19 | 2017-12-19 | A kind of sewage water and water yield modeling method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108171408A true CN108171408A (en) | 2018-06-15 |
CN108171408B CN108171408B (en) | 2018-12-04 |
Family
ID=62522823
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711373832.8A Active CN108171408B (en) | 2017-12-19 | 2017-12-19 | A kind of sewage water and water yield modeling method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108171408B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109111030A (en) * | 2018-08-27 | 2019-01-01 | 重庆固润科技发展有限公司 | Integrated sewage disposal intelligence control system and control method |
CN109473972A (en) * | 2018-08-31 | 2019-03-15 | 长沙理工大学 | Whole source lotus is assisted to store up optimal control method based on more power curve |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104794112A (en) * | 2014-01-16 | 2015-07-22 | 中国移动通信集团山西有限公司 | Time series processing method and device |
CN106127359A (en) * | 2016-08-30 | 2016-11-16 | 北京协同创新智能电网技术有限公司 | A kind of variable frequency pump rate of discharge method for early warning based on local weighted linear regression model (LRM) |
CN106872658A (en) * | 2017-01-22 | 2017-06-20 | 华南理工大学 | A kind of method of the COD of sewage load prediction based on vector time series model |
-
2017
- 2017-12-19 CN CN201711373832.8A patent/CN108171408B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104794112A (en) * | 2014-01-16 | 2015-07-22 | 中国移动通信集团山西有限公司 | Time series processing method and device |
CN106127359A (en) * | 2016-08-30 | 2016-11-16 | 北京协同创新智能电网技术有限公司 | A kind of variable frequency pump rate of discharge method for early warning based on local weighted linear regression model (LRM) |
CN106872658A (en) * | 2017-01-22 | 2017-06-20 | 华南理工大学 | A kind of method of the COD of sewage load prediction based on vector time series model |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109111030A (en) * | 2018-08-27 | 2019-01-01 | 重庆固润科技发展有限公司 | Integrated sewage disposal intelligence control system and control method |
CN109473972A (en) * | 2018-08-31 | 2019-03-15 | 长沙理工大学 | Whole source lotus is assisted to store up optimal control method based on more power curve |
CN109473972B (en) * | 2018-08-31 | 2021-07-13 | 长沙理工大学 | Source load storage optimization control method based on multi-power curve coordination |
Also Published As
Publication number | Publication date |
---|---|
CN108171408B (en) | 2018-12-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110108328B (en) | Method for acquiring water leakage amount of leakage area of water supply pipe network | |
CN111898691B (en) | River burst water pollution early warning and tracing method, system, terminal and medium | |
CN107355688B (en) | Urban water supply network leakage control management system | |
CN111046564B (en) | Residual life prediction method for two-stage degraded product | |
Diamantopoulou et al. | Modelling total volume of dominant pine trees in reforestations via multivariate analysis and artificial neural network models | |
CN116595327B (en) | Sluice deformation monitoring data preprocessing system and method | |
CN107730395B (en) | Power consumption abnormity detection method based on power consumption deviation rate for low-voltage users | |
CN109359698A (en) | Leakage loss recognition methods based on long Memory Neural Networks model in short-term | |
CN113610381B (en) | Water quality remote real-time monitoring system based on 5G network | |
CN106199174A (en) | Extruder energy consumption predicting abnormality method based on transfer learning | |
CN113036913B (en) | Method and device for monitoring state of comprehensive energy equipment | |
CN105740989B (en) | A kind of water supply network anomalous event method for detecting based on VARX model | |
CN108171408B (en) | A kind of sewage water and water yield modeling method | |
CN108595687A (en) | Water consumption method for detecting abnormality and database server | |
CN110298480A (en) | A kind of mountain flood Critical Rainfall index calculating method and system | |
CN114323412B (en) | Water supply pipe network pressure disturbance event detection method | |
CN109060393A (en) | A kind of bridge structure dead load response Time Domain Fusion analysis method | |
Preis et al. | On-line hydraulic modeling of a Water Distribution System in Singapore | |
Yang et al. | Research on singular value detection method of concrete dam deformation monitoring | |
CN114462688A (en) | Tube explosion detection method based on LSTM model and dynamic threshold determination algorithm | |
CN112559969B (en) | Small leakage detection method based on accumulation sum algorithm | |
CN112884197B (en) | Water bloom prediction method and device based on double models | |
Karaoglan et al. | A regression control chart for autocorrelated processes | |
TWI617788B (en) | Method of real-time prognosis of flooding phenomenon in packed columns | |
CN109635358A (en) | A kind of unit fault detection method based on sliding window Multiscale Principal Component Analysis |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |