A cleaning method for data of a ventilation and air conditioning system of a subway station

Abstract

A kind of cleaning method of subway station ventilation and air conditioning system data, in order to improve the performance of subway ventilation and air conditioning control system anti-disturbance, the technical scheme is to carry out cleaning treatment to the sensor data of subway station ventilation and air conditioning system, first, preliminary data is eliminated, then effective data is screened out through precision cleaning, finally, whether sensor has problem is judged by the change of the average value of all data after precision cleaning in a future period of time, to ensure the ventilation effect of overall space.The beneficial effects of the present application are: the present application relies on the principle of Lyapunov rule, so that PLC industrial controller has data processing capability, meets the actual control of sensor data anti-disturbance degree demand, has better stable control effect in large space scene, provides support for energy saving effect and system control stability of energy saving control system in later period.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of subway ventilation and air conditioning, and in particular relates to a subway station ventilation and air conditioning system data cleaning method, in particular a subway station ventilation and air conditioning system data cleaning method based on the Laplace rule. BACKGROUND

[0002] In the automatic control process of the subway ventilation and air conditioning energy-saving control system, a large amount of space sensor data needs to be collected, including temperature sensors, humidity sensors, CO2 sensors, PM2.5 sensors, etc. The number of sensors with the same attribute is generally between 10-30. Due to the flow of passengers in the station, the local environment changes significantly, and data deviation often occurs during the use of the sensor. In the second scenario, the sensor data cannot effectively reflect the overall temperature, humidity, CO2 concentration, PM2.5, etc. in the large space scenario. This causes the control system to misjudge or affects the stability of the ventilation and air conditioning system.

[0003] Currently, in the field of subway ventilation and air conditioning energy saving, the data of each load is generally processed by algorithm, including model learning, etc., so as to achieve the purpose of energy-saving control. For example, the invention patent with publication number CN113901713A discloses a subway air conditioning system energy-saving control method based on ActorCritic algorithm, which includes: establishing a subway air conditioning system energy consumption model; collecting subway air conditioning system index data and preprocessing the subway air conditioning system index data to obtain a set of preprocessed subway air conditioning system index data; taking the subway air conditioning system index data as input and the next time subway air conditioning system energy consumption prediction value as output, constructing a subway air conditioning system energy consumption prediction model; taking the subway air conditioning air-water system as a reinforcement learning agent, taking the minimum subway air conditioning system energy consumption prediction value as the training target, and using the ActorCritic algorithm to train the reinforcement learning agent to obtain the control action of the subway air conditioning air-water system, thereby realizing real-time energy-saving control of the subway air conditioning.

[0004] The above method can achieve certain energy-saving control, but since the basic data of each sensor is not processed, or since a certain sensor has a problem or fault, although certain energy-saving is achieved, it may cause local temperature, humidity, etc. in the space to be unbalanced, reducing the effect of the ventilation and air conditioning system. Under this background, a data cleaning method is needed to clean the outlier data, and the result is output to the subway ventilation and air conditioning energy-saving control system as control feedback data to improve the anti-disturbance performance of the ventilation and air conditioning control system. SUMMARY

[0005] The application designs a subway station ventilation and air conditioning system data cleaning method to improve the anti-interference performance of the subway ventilation and air conditioning control system, preliminarily processes data of all sensors from the overall layout of sensors, reasonably controls each load, and realizes the overall working effect and energy saving purpose.

[0006] The technical scheme of the application is a subway station ventilation and air conditioning system data cleaning method, which cleans data collected by all sensors in the subway station by means of a data processing unit, and the cleaning method comprises the following steps:

[0007] a. Preliminary cleaning of data, removing data beyond the range or reasonable data range;

[0008] b. Calculating the average value AM of the data after the preliminary cleaning of step a, and the formula is:

[0009]

[0010] In formula 1, n is an integer greater than or equal to 1, x i represents the value output by the sensor;

[0011] c. Calculating the residual error value Vi after the preliminary cleaning, and the formula is:

[0012] Vi i =x i -AM (Formula 2);

[0013] d. Calculating the standard error value SD after the preliminary cleaning, and the formula is:

[0014]

[0015] e. Calculating the precision cleaning judgment parameter R, and the formula is:

[0016]

[0017] f. Precision cleaning of data, selecting data between the upper limit and the lower limit, including upper limit and lower limit data, and removing data outside the upper limit and the lower limit, specifically:

[0018] For data with R<0.5, the upper limit of precision cleaning is AM+1.5×sD, and the lower limit of precision cleaning is AM-1.5×SD;

[0019] For data with 0.5≤R<1, the upper limit of precision cleaning is AM+SD, and the lower limit of precision cleaning is AM-SD;

[0020] Meanwhile, for data with R≠0 and SD≠0, the upper limit of precision cleaning is The lower limit of precision cleaning is

[0021] g. Calculate the final average value of the data selected in step f. If the final average value changes within 15-25 minutes, the final sensor output data is the final average value. If the final average value does not change within 15-25 minutes, it is determined that the sensor has a problem, and the output data is the preset system protection data.

[0022] All the sensors mentioned are of the same type, namely temperature sensors, humidity sensors, CO2 sensors, or PM2.5 sensors.

[0023] The number of sensors is greater than or equal to three.

[0024] The core technical solution of this invention is to clean the sensor data of the ventilation and air conditioning system of subway stations. First, preliminary data is removed, and then effective data is selected through precision cleaning. Finally, the change of the average value of all data after precision cleaning over a period of time is used to determine whether the sensor has a problem, thereby ensuring the ventilation effect of the overall space.

[0025] The beneficial effects of this invention are: relying on the principle of Laida's law, this invention enables PLC industrial controllers to have data processing capabilities, meet the actual control requirements for sensor data anti-interference, have better stable control effects in large space scenarios, and provide support for the later energy-saving effect and system control stability of energy-saving control systems. Attached Figure Description

[0026] Figure 1 This is a flowchart of the sensor data cleaning method in the subway ventilation and air conditioning system of the present invention. Detailed Implementation

[0027] In specific implementation of this invention, see [link / reference]. Figure 1 Using a PLC or an existing main control PLC, the data cleaning function is implemented to analyze data from three or more similar environmental sensors in the large space of subway station concourses and platforms, including those for temperature, humidity, CO2, and PM2.5. The validity of the sampled data is judged based on the following points:

[0028] A. Determine whether the sensor data is usable by checking the effective range of the sensor's measurement range;

[0029] B. Whether the sensor data is a significant outlier in all the sensor data clusters used, and then clean and remove it;

[0030] C. Sensor data is segmented and filtered in the effective sensor data cluster to achieve data convergence.

[0031] D, whether the sensor data is changing within a period of time to determine whether the sensor is available.

[0032] The application scenario of the present application is a subway station hall platform and the like similar large space scene, the number of sensors is greater than or equal to 3, and the sensors are of the same type. If there are different types of sensors in the space, they need to be processed separately by type.

[0033] In the specific processing, first, the actual number of sensors of the same type is input in the controller, for example, 3-20 is input, and the actual situation is input.

[0034] Secondly, preliminary rough screening is carried out, screening is carried out according to the range or reasonable data range, invalid data is eliminated, and the valid data screened out preliminarily is counted and processed, including average value AM, residual error value Vi and standard deviation value SD, which are shown in formulas 1-3 respectively.

[0035] Then, the precision cleaning judgment parameter R is calculated, as shown in formula 4. The precision cleaning is carried out according to the principle of Table 1 below.

[0036] Table 1 Precision cleaning data upper and lower limit table

[0037]

[0038] Finally, the final data average value AM_last is calculated after the precision cleaning data, if the final data average value changes within 15-25 minutes, the final sensor output data is the final data average value, if the final data average value does not change within 15-25 minutes, it is judged that the sensor has a problem, which may be a drop line or other hardware problem, at this time the output data is the system protection data of the preset value.

[0039] The above cleaning process of the sensor data can solve the following technical problems:

[0040] 1) Solve the outlier elimination of sensors with the same attribute, and increase the stability of the participating control data.

[0041] 2) Solve the situation that the sensor is completely disabled, the control system will not appear overshoot, and the participating control data is controlled within the preset range.

[0042] Finally, the demand of the actual control on the sensor data anti-interference degree is met, the stable control effect is better in the large space scene, and the support is provided for the energy saving effect and system control stability of the energy saving control system in the later period.

Claims

1. A method for cleaning data of a ventilation and air conditioning system of a subway station, wherein data collected by all sensors of the subway station are cleaned by means of a data processing unit, characterized in that: The cleaning method is as follows: a. The data is preliminarily cleaned, and the data exceeding the range or reasonable data range is removed; b. The average value AM of the data after the preliminary cleaning in step a is calculated, and the formula is: In formula 1, n is an integer equal to or greater than 1, x i represents a value of a sensor output; c. The residual error value Vi after the preliminary cleaning is calculated, and the formula is: Vi i = x i - AM (Equation 2); d. The standard error value SD after the preliminary cleaning is calculated, and the formula is: e. The precision cleaning judgment parameter R is calculated, and the formula is: f. The data is cleaned in precision, and the data between the upper limit and the lower limit is selected, including the upper limit and the lower limit data, and the data outside the upper limit and the lower limit is removed, and specifically: For data with R<0.5, the upper limit of the precision cleaning is AM+1.5×SD, and the lower limit of the precision cleaning is AM-1.5×SD; For data with 0.5≤R<1, the upper limit of the precision cleaning is AM+SD, and the lower limit of the precision cleaning is AM-SD; At the same time, for the data of R≠0 and SD≠0, the upper limit of precision cleaning is The lower limit of precision cleaning is g. The final data average value is calculated for the data selected in step f, and if the final data average value changes within 15-25 minutes, the final sensor output data is the final data average value; If the final data average value does not change within 15-25 minutes, it is determined that the sensor has a problem, and the output data is the system protection data of the preset value.

2. The method of claim 1, wherein the method comprises: All the sensors are the same type of sensors, which are temperature sensors, humidity sensors, CO2 sensors or PM2.5 sensors.

3. The method according to claim 1 or 2, characterized in that: The number of sensors is greater than or equal to 3.

Images