WO2022009085A1

WO2022009085A1 - System and method for the management and optimisation of building temperature measurements for the implementation of an automatic control system

Info

Publication number: WO2022009085A1
Application number: PCT/IB2021/056037
Authority: WO
Inventors: Luca Barboni; Giorgia FARELLA; Giovanni BARTUCCI
Original assignee: Alperia Bartucci S.P.A.
Priority date: 2020-07-08
Filing date: 2021-07-06
Publication date: 2022-01-13
Also published as: IT202000016474A1

Abstract

The present invention relates to a method of controlling the temperature of a vector fluid for heating a building. According to the method - a desired value of an ambient temperature inside the building is set (1001), and - values of ambient temperature of the building by means of at least one wireless sensor temperature are detected (1002). Advantageously, the method also provides for: - generating (1003), by using a mathematical model, estimates of the ambient temperature inside the building, - controlling (1004) the temperature of the vector fluid based on the ambient temperature values measured by the sensor and based on the temperature estimates generated by the mathematical model in order to reach the desired ambient temperature value inside the building, - comparing periodically one of said temperature estimates with a temperature value measured by the sensor, - reducing the frequency with which the sensor performs the estimate and the measurement differ less than a predetermined tolerance threshold, or increasing the frequency with which the sensor performs the measurements if the two values differ more than the predetermined tolerance threshold.

Description

SYSTEM AND METHOD FOR THE MANAGEMENT AND OPTIMISATION OF BUILDING TEMPERATURE MEASUREMENTS FOR THE IMPLEMENTATION OF

AN AUTOMATIC CONTROL SYSTEM

DESCRIPTION

TECHNICAL FIELD

The present invention relates to the field of heating systems for residential units, whether flats, buildings or complexes of buildings.

BACKGROUND

Nowadays there are different systems for heating residential units, such as flats and buildings.

Some of the units have independent heating, i.e. each unit has its own heating system. For example, the residential unit may be equipped with a boiler that heats water that is circulated through radiating elements within the residential unit. Alternatively, the residential unit can be equipped with an HVAC (Heat Ventilation Air conditioning and Cooling) system, wherein a heat pump heats or cools an air flow that is sent into the residential unit.

Other solutions involve the use of a central heating system serving several residential units. This is the case in apartment blocks, where a central heating unit - a boiler or a condominium HVAC system - heats a vector fluid (water or air, respectively), which is sent to the different residential units. Each residential unit is also equipped with a control system to increase or decrease the temperature inside the unit. For example, in the case of boilers, the water heated by the boiler is sent to radiators equipped with valves that can be regulated manually or by a thermostat inside the residential unit.

Yet another form of heating is district heating. In this case, a large thermal power plant heats water that is sent to several buildings, each of which is then equipped with pumps to distribute the hot water to the different residential units within the building.

Regardless of the size and type (boiler or HVAC) of the heating system, there is the issue of optimising the energy consumption needed to heat the vector fluid (water or air) that is sent to the residential units. Energy consumption depends on many factors, e.g. the size and exposure of the residential unit, the temperature actually reached in each residential unit and the desired temperature (also called set-point). In autonomous heating systems, the residential unit is typically equipped with a thermostat that can be set by a user and can control the operation of the boiler or HVAC system. The user sets the temperature to be maintained during the day on the thermostat, then the thermostat checks the temperature inside the residential unit and regulates the switching on of the boiler or HVAC system to increase or decrease the temperature of the vector fluid that transports heat inside the residential unit. If the thermostat's temperature sensor runs on battery power and is discharged or does not provide regular temperature values in the house, the temperature control of the heating fluid will be sub- optimal, resulting in wasted energy or, in the worst case, complete interruption of heating.

In condominium heating systems or district heating, the central heating unit heats a vector fluid which is sent to several residential units, each of which will have its own thermostat and set its own desired temperature curve. In this case, the central heating unit must heat the vector fluid to a high temperature, capable of meeting the requirements of all the residential units. The thermostats then act on the flow manifold (typically by opening or closing an inlet valve) to let in more or less vector fluid to reach the desired temperature.

A recent study by Texas Instrument (cited in https: / / ww w .telit.com /blog/ smart- building-automation -benefits / ), has shown that the intensive use of IoT (Internet of Things) sensors in a building to control the temperature in different rooms of a building differently throughout the day can lead to high energy savings.

However, in most cases these sensors cannot guarantee deterministic temperature measurement to the heating system, as communication between sensors and the heating system is usually via radio communication with rather simple communication protocols that are not designed for control systems, and therefore do not guarantee that the temperature measurement is received by a controller at a specific time instant. Such communication is then subject to interference and data loss. In addition, these sensors are usually battery-powered, so they can run down.

OBJECTS AND SUMMARY OF THE INVENTION

An object of the present invention is to overcome the disadvantages of the prior art.

In particular, it is the purpose of the present invention to present a heating system which allows energy consumption to be reduced by extending the average life of the battery of the temperature sensors and by obviating communication problems between the sensors and the control system.

These and other objects of the present invention are achieved by a method incorporating the features of the appended claims, which form an integral part of the present description.

According to a first aspect, the invention relates to a method for controlling the temperature of a vector fluid for heating a building. The method involves setting a desired value of an ambient temperature inside the building, and detecting ambient temperature values in the building using at least one wireless temperature sensor. The method also involves generating, using a mathematical model, estimates of the ambient temperature inside the building, and controlling the temperature of the vector fluid on the basis of the ambient temperature values measured by the sensor and on the basis of the temperature estimates generated by the mathematical model so as to achieve the desired value of ambient temperature inside the building. Periodically one of the temperature estimates is compared with a temperature value measured by the sensor. The frequency at which the sensor takes measurements is reduced if the estimate and measurement differ by less than a set tolerance threshold, or increased if the two values differ by more than the set tolerance threshold.

This solution offers several advantages. Firstly, if the wireless sensors are battery- powered, this solution allows the average life of the battery to be extended, as control over the temperature of the vector fluid is done not only on the basis of the measured temperature values, but also on the basis of temperature estimates, thus reducing the sensor measurements. Furthermore, even if the sensor measurements do not reach the controller, e.g. due to interference, the controller can continue to regulate the temperature of the vector fluid to heat the building using the temperature estimates made.

In one embodiment, the mathematical model is an integral or pseudo-integral model. In fact, the applicant has verified that this model is very effective in describing the thermal dynamics inside a building, and is therefore equally effective in providing true estimates of the ambient temperature inside the building.

In one embodiment the mathematical model is

T_v(t) = J k_p - ( T_DSP - T_ref) dt + f k_d (T_e - T_ref) dt where k_p and k_d are two parameters that vary over time, TDSP is a set-point value of the temperature of the vector fluid that is controlled and that is intended to heat the building, T_e is a temperature outside the building, T_ref is a predetermined reference temperature, preferably comprised between 18° C and 22° C and more preferably equal to 20° C.

This model is very effective for estimating the temperature inside the building because it takes into account two components, a first one related to the energy that is fed into the building via the vector fluid, and a second one that takes into account the energy dispersed. The choice of using the set-point value of the temperature of the vector fluid, i.e. that which the controller of the system considers useful to reach, allows the use of a certain and precise datum in the model.

In one embodiment, the parameter k_p is estimated by means of two estimation procedures; if the estimate of k_p of the two procedures differs less than a predetermined threshold, then the values of k_p and k_d generated by one of the two estimation procedures are used. This solution is very robust and allows a model to be obtained that becomes increasingly accurate over time, thus allowing the measurement frequency of the sensor to be reduced and its average life to be extended.

Preferably, according to one of the estimation procedures the i-th estimate of k_p is

wherein

AT_DSp is the moving average, measured during the estimation procedure, of the difference in the value of set-point of the temperature in the delivery vector fluid;

AT_e is the variation of the external temperature between the end and the beginning of the estimation procedure; å AT_S is the sum of the changes in ambient temperature measured by the sensor during the whole estimation procedure;

Dί is the time interval measured from the beginning to the end of the estimation procedure; a is a constant, preferably a is the last value of the parameter k_d calculated by the first estimation procedure, or is zero, or is the value of k_d from the other estimation procedure.

Even more preferably, the other estimation procedure is a recursive algorithm for estimating the parameters, in particular a Kalman filter, or an identification system of recursive least squares type with oblivion coefficient or another estimation method pertaining to the machine learning category. These estimation procedures are particularly robust and reliable, yet require little computational capacity to perform.

In one embodiment, the outside temperature is a data item acquired from the internet or from an external database or sensor.

With this solution it is possible to dynamically update the outside temperature value of the building and/ or to acquire predicted outside temperature values at future times.

In one embodiment, the sensor providing the ambient temperature measurements inside the building is one of a plurality of sensors installed in the building, and in particular is the one providing the lowest temperature among all the sensors. Preferably, the coldest points in the building are identified based on the exposure of the building under average occupancy and zero heating conditions.

This choice of sensor position ensures that even the coldest point of the building reaches a desired temperature.

In one embodiment, the method involves checking whether the measurements of said plurality of sensors are considered valid and used for the estimation of the parameters of the model if several conditions indicative of problems with the measuring sensor and/or environmental situations are verified such that the measurement is considered not usable for control purposes, e.g. it is verified whether:

• the measured temperature value is comprised between a minimum and a maximum value which correspond to the full scale values of the measurement sensor,

• if the measured value is constant for a time greater than a certain threshold value,

• if the measured value is too high compared to the previous measurement - for example if it is 30% higher.

• if the derivative of the measured temperature value is comprised between a minimum and a maximum value,

• if no window is opened in the room where the sensor sending the measurement is located. The open condition of a window is verified by checking the trend of the individual measurements of a given sensor against the average trend of the sensors. In particular, a slow downward deviation of one or more measurements of one sensor Si is recognised as the opening of the window. In the case of several sensors installed in the building, the method involves temporarily excluding the sensor corresponding to an open window condition from the ambient temperature measurement selection logic for the controller. When the window open condition ends, the signal is restored as usable by the selection module.

In one embodiment, wherein an open window condition is verified by checking the trend of the individual room temperature measurements of a given sensor with respect to the average trend of the measurements of the same sensor, wherein a slow deviation towards lower values of one or more measures is identified as an open window condition.

In this case, the method involves temporarily excluding the sensor from the measurement selection logic for the controller and restoring the signal provided by the sensor as usable by the control logic when the window opening condition ends.

Thanks to this solution, the method is able to react correctly and promptly to spurious measurements of the building temperature related to the opening or closing of one or more building windows in the areas where sensors are present, and implement the most correct control action.

In one embodiment, the step of controlling the temperature of the vector fluid based on the ambient temperature values measured by the sensor and based on the temperature estimates generated by the mathematical model involves:

- using the temperature estimates generated by the mathematical model as a control variable when communication between one or more sensors and the control unit that controls the temperature of the vector fluid is interrupted.

Thanks to this solution, it is possible to guarantee continuous and stable operation of the heating system even in the event of a malfunction (or battery discharge) of one of the sensors or the presence of interference that prevents correct communication between the sensor and the control unit.

According to a further aspect, the invention relates to a heating system comprising conduits for transporting a vector fluid inside a building, heating means for heating the vector fluid, at least one wireless sensor capable of transmitting ambient temperature measurements inside the building, and a control unit operatively connected to the wireless sensor and the heating means and configured to implement a method of controlling the temperature of the vector fluid as set forth above and further described below.

Further features and advantages of the present invention will be more apparent from the description of the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

The invention will be described below with reference to some examples, provided for explanatory and non-limiting purposes, and illustrated in the accompanying drawings. These drawings illustrate different aspects and embodiments of the present invention and, where appropriate, reference numerals illustrating similar structures, components, materials and/or elements in different figures are indicated by similar reference numbers.

Figure 1 shows a building with residential units heated by a central heating system;

Figure 2 is a flow chart of a method of controlling the temperature of the vector fluid used to heat the building in figure 1;

Figure 3 is a block diagram of the control unit of the central heating system in figure 1;

Figure 4 is a flow chart of a method for estimating the ambient temperature implemented by the control unit in figure 3;

Figure 5 is a block diagram of the control unit of the central heating system in figure 1 in which additional blocks to figure 3 have been shown;

Figure 6 provides a qualitative illustration of the trend of the ambient temperature in the building in figure 1 and the trend of the delivery temperature of the vector fluid as a function of the time of the central heating system during a system learning step;

Figure 7 is a flow chart of a process for regulating the ambient temperature implemented by the control unit in figures 3 and 5;

Figure 8 provides a qualitative illustration of the trend of the ambient temperature in the building in figure 1 and the trend of the delivery temperature of the vector fluid as a function of the time of the central heating system during a dual temperature control mode;

Figure 9 is a block diagram of a part of the control unit in figures 3 and 5;

Figure 10 is a flow chart of a process for calculating the control parameters used by the control unit in figures 3, 5 and 9;

Figure 11 is a graph illustrating three curves of the trend of the radiant temperature of the radiators as a function of time after the boiler has been turned off;

Figure 12 is a flow chart of a process for ambient temperature regulation implemented by the control unit in figures 3, 5 and 9;

Figure 13 provides a qualitative illustration of the trend of the ambient temperature in the building in figure 1 and the trend of the delivery temperature of the vector fluid as a function of the time of the central heating system during a dual control mode with temperature coasting;

Figure 14 is a flow chart of an ambient temperature coasting process implemented by the control unit of figures 3, 5 and 9;

Figure 15 is a flow chart of a further process for ambient temperature coasting implemented by the control unit in figures 3, 5 and 9; and

Figure 16 is a block diagram of the control unit of the central heating system in figure 1 in which optional blocks with respect to figure 3 and figure 5 are shown.

DETAILED DESCRIPTION OF THE INVENTION

While the invention is susceptible to various modifications and alternative constructions, certain preferred embodiments are shown in the drawings and are described hereinbelow in detail. It is in any case to be noted that there is no intention to limit the invention to the specific embodiment illustrated, rather on the contrary, the invention intends covering all the modifications, alternative and equivalent constructions that fall within the scope of the invention as defined in the claims.

The use of "for example", "etc.", "or" indicates non-exclusive alternatives without limitation, unless otherwise indicated. The use of "includes" means "includes, but not limited to" unless otherwise stated.

Figure 1 illustrates a building 100 comprising a plurality of residential units 101 to be heated. The building is equipped with a heating system comprising a central heating unit 1 which heats a vector fluid and sends it to the residential units where the vector fluid yields heat to the environment by heating it. The central heating unit may be installed locally within the building, as shown in figure 1, or be a remote unit serving the building 100 and possibly also other buildings. For the sake of clarity, in the following example, the central heating unit is a boiler equipped with a burner 10 which heats water by sending it into a delivery conduit 103 to which manifolds 105 are connected, from which each residential unit on the floor receives the heated water, circulates it through radiators 106 arranged in the various rooms of the residential unit, and feeds it back into a return conduit 104 which arrives at the central heating unit 1 to be heated again. Other embodiments may include the use of a heat pump apparatus, or other HVAC system capable of heating air that is sent to the residential units.

The heating unit 1 also includes a control unit 20 and a radio interface 30. The radio interface 30 receives temperature values measured by temperature sensors located in the various residential units. In the example described here there are three sensors SI, S2 and S3, but it is clear that the number can vary depending on the number of residential units. At least three sensors are preferably provided in a building at the coldest points - among those to be heated - of the building. The assessment can be made by studying the exposure of the building and identifying the coldest rooms.

The temperature sensors are preferably IoT sensors, capable of communicating with the heating unit via a radio link, e.g. Wi-Fi, Bluetooth, Zigbee, Lora, Sigfox or NB- IoT, suitable for the distances to be covered. In one embodiment, the IoT sensors are battery operated, so they can also be installed in existing buildings without the need to bring new power lines to power them in the optimal locations, i.e., as mentioned above, the cooler parts of the building.

Each residential unit 101 provides, in a known manner, a thermostat, indicated with the references T1 and T2 in figure 1, which controls one or more valves 107 placed at the entrance to the residential unit in order to let new hot water from the delivery conduit 103 enter the radiators 106.

In a known manner, the user sets a desired temperature value or temperature time trend unit on the residential unit thermostat and the thermostat opens and closes the valve 107 to reach the temperature value desired by the user.

The presence of thermostats is not essential and is only an example, as the residential unit could be equipped with thermostatic valves on the radiators that are individually regulated by a user. Again, the residential unit could be equipped with heating circuits without radiators or thermostatic valves, but with simple radiator valves that can be manually regulated by a user.

As each user adjusts - either manually or via a thermostat - the temperature of their radiators, the heat load seen by the boiler - i.e. the amount of heat required by the boiler - varies over time.

In order to optimise the amount of heat yielded by the boiler to the delivery water, the control unit 20 is configured to implement a method for controlling the temperature of the delivery water intended for heating the building. Having determined the temperature value to which the delivery water is to be brought, in one embodiment the control unit 20 controls the burner 10 in a known way to bring the delivery water to the calculated temperature value. Given that there are already boilers which manage the burner by varying its thermal power according to a given temperature to be reached, we will not go into the merits of this mechanism here.

The method of controlling the delivery water temperature implemented by the control unit is illustrated in its general lines in figure 2, and envisages: a) setting (step 1001) a desired ambient temperature set-point inside the building, b) detecting (step 1002) ambient temperature values inside the building by means of at least one sensor, c) generating (step 1003) at least one estimate (Tv) of the ambient temperature inside the building by means of a mathematical model, d) controlling (step 1004) the temperature of the vector fluid on the basis of the building temperature values measured by the sensor and on the basis of the ambient temperature estimates generated by the mathematical model so as to bring the building temperature to the desired set-point value.

As illustrated schematically in figure 3, the control unit 20 comprises a selection module 200 which receives as input the temperature values Ts measured by the environmental sensors Si and the ambient temperature estimates T_v and generates as output an ambient temperature value T_c which is used by the controller 201 to control the burner 10. The selection module 200 checks if the measurements of the sensors Ts are valid and selects one according to a pre-determined criterion, preferably the lowest temperature Ts. Checking whether measurements are valid involves checking various conditions indicative of problems with the measurement sensor and/ or environmental situations such that the measurement is not considered usable for control purposes, e.g. it is checked whether:

• the opening conditions of a window in the room where the sensor Si sending the measurement is located do not occur. The opening condition of a window is verified by checking the trend of the individual measurements of a given sensor Si with respect to the average trend of the sensors. In particular, a slow downward deviation of one or more measurements of one sensor Si is recognised as the opening of the window. In the case of several Si sensors installed in the building, the method involves temporarily excluding the sensor corresponding to an open window condition from the ambient temperature measurement selection logic for the controller. When the window opening condition ends, the signal is restored as usable by the selection module 200.

The ambient temperature value estimate Tv is carried out by the estimation module 202, which will be discussed in more detail below.

Periodically, the control unit 20, in particular the selection module 200, compares the ambient temperature value estimated by the estimation module 202 with the ambient temperature value Ts measured by one of the sensors Si. In particular, the estimated ambient temperature Tv is compared to the lowest temperature among those measured by the installed sensors Si, however, it is possible to use other criteria for selecting the sensor with which to compare the estimate Tv.

Depending on the comparison, the control unit 20 regulates the frequency with which the sensors Si take measurements and transmit them to the control unit 1, where they are received by the radio interface 30 and supplied to the control unit 20. For this purpose, the control unit 20 sends a control message via the radio interface 30 to the sensors Si.

In one embodiment, in the event that the estimate T_v and the measurement Ts coincide at less than a predetermined tolerance threshold, e.g. within 5%, then the control unit 20 reduces the measurement frequency of the actual sensor Si. Otherwise, the measurement unit increases the measurement frequency of the actual sensor Si.

In other embodiments, it is also possible to provide a tolerance range, e.g. 3% to 10%, whereby if the difference between the estimate T_v and the measurement Ts falls within this range then the frequency of the sensors is not changed, whereas if the difference is greater than the upper limit of the range then the frequency is increased, and if it is less than the lower limit of the range then the measurement frequency of the sensors is reduced. Returning to the estimate of the ambient temperature Ts, this is preferably obtained by means of a machine learning algorithm of the model-based type, i.e. an artificial intelligence algorithm based on a basic model which is preferably a basic model of the integral or pseudo-integral type.

In the example described here, the basic model is of the pseudo integral type and connects the estimated ambient temperature Tv to the delivery water temperature and the temperature outside the building, according to the following relationship (1):

T_v(t) = J k_p - ( T_DSP - T_ref) dt + f k_d (T_e - T_ref) dt (1)

Where

• k_p and k_d are two parameters that vary over time - in detail k_p is a gain coefficient associated with the delivery water temperature, while k_d is a gain coefficient associated with the temperature outside the building,

• TDSP is the set-point value of the delivery water temperature,

• T_e is the temperature outside the building, acquired via a sensor or through the internet or connection to a remote server,

• T_ref is a predefined reference temperature, preferably between 18°C and 22°C and more preferably equal to 20°C.

In figure 2, the module 202 receives as input both a value of the delivery water temperature measured by means of a sensor (TDPV), and the desired value (also called set- point, TDSP) of the delivery water temperature provided as output by the controller 201. In this way, when a variation between command (TDSP) and measured value (TDPV) occurs, the system knows to expect a subsequent variation of both the measured value TDPV and the building temperature Ts, and can use this information to validate the temperature measurements received from the sensors Si.

The module 202 also receives as input the ambient temperature values Ts measured by the sensors Si, so the machine learning algorithm starting from the base model can begin to estimate the model parameters, with one of the techniques known in literature, to adapt to the building consumption.

In order to make the estimate of the system parameters k_p and k_d more robust, the module estimates 202 k_p using two different estimation procedures. If the estimate of k_p obtained from the two procedures differs by less than a predefined threshold, then the values of k_p and k_d generated by one of the two estimation procedures are taken. Otherwise, the estimate is not considered reliable and is not passed on to the controller 201. In other words, the module 202 does not provide the module 200 with any Tv estimates, and the controller 201 will base its decisions solely on the measurements received from the sensors Ts as long as the k_p values obtained by the two procedures coincide at less than the predetermined threshold.

In particular, one embodiment uses a heuristic procedure based on the recognition of proper conditions for the excitation of the system to be controlled. In particular, the procedure involves looking for variations in the desired value of the delivery water temperature (TDSP) that have sufficient energy (e.g. are greater than 8 °C) to cause a visible response of the ambient temperature inside the building. Herein, a 'visible' response means a response to a stimulus characterised by an adequate signal-to-noise ratio - for example, a signal-to-noise ratio greater than or equal to 1.8 and more preferably greater than or equal to 2 in the case of sensors typically installed in the generic boiler 10.

The model parameters are then estimated at the TDSP changes that satisfy the above energy criterion. When estimated, the parameters are then compared with the output of a more traditional procedure such as recursive least squares with oblivion coefficient or Kalman filter or with the output of a machine learning algorithm of regression type - for example, a support vector regression algorithm or a neural network -, which estimates identical quantities, i.e. the growth rates k_p and k_d, linked to the boiler delivery temperature and the outdoor temperature respectively.

A procedure for estimating k_p, implemented by the module 202, is illustrated below with reference to the flowchart in figure 4. After the start of the estimation method (step 300), some parameters are initialised (step 301), in particular, a counter i which takes into account the i-th parameter k_p,i calculated, and a counter n which takes into account the n-th sampling step performed, are initialised to the value 0; moreover, the variable k_p, LITER used in the calculation of the parameter k_p , is initialised to zero, and the mean value and the standard deviation of the previously calculated parameters k_p are acquired.

The module 202 therefore checks whether three conditions are met to start the estimation process, in detail:

It is checked (step 302) whether the difference (hereafter referred to as ATD) between the set-point value of the delivery water temperature and the delivery water temperature value exceeds a predetermined threshold value, for example 8°C. Mathematically, it is checked whether:

ATD=TDSP-TDPV³ e Where e is the aforesaid predetermined threshold. This allows the estimate to be activated only if the controller 201 has determined the need for an increase in the delivery water temperature such that there is a noticeable deviation in the ambient temperature inside the building.

It is checked (step 303) whether the boiler is switched on or not.

It is checked (step 304) whether the outside temperature Te, the delivery water temperature (TDPV) and the ambient temperature (Ts) measurements received as input are valid. This verification is similar to what was explained above with regard to the verification carried out by the selection module 200. Obviously for the delivery water temperature and the outside temperature no verification of the open window condition is carried out.

If at least one of the above conditions is not met, then the module 202 does not initiate the estimation procedure and the method returns to step 302 without producing an output estimate of k_p. In such a case, as described below, the values k_p and k_d of the mathematical model will not be updated.

If, on the other hand, all three of the above conditions are met, the module 202 proceeds to calculate the estimate of k_p.

First a stopwatch is started (step 305), then: a) The set-point value of the delivery temperature TDSP is acquired by the controller 201 and the moving average of the variation of this value is calculated (step 306).

In practice, at each n-th sampling, the variation of the delivery temperature set-point is calculated as a moving average written in recursive form, i.e: T_Dsp(.ⁿ) ⁼ T_DSP(n — 1) + ( T_DSP(n ) — T_DSP 0) — AT_DSP(n — 1 ))/n.

At the first sampling step, n is zero, and there is no value AT_DSP(n — 1) to use in the above formula. For n=0, therefore, we conventionally take AT_DSp(n — 1)=0. In other words, the first variation of the delivery temperature set-point value (AT_DSP( 0)) is calculated as the difference between the first set-point value provided by the controller 201 (TDSP(0)) and the delivery water temperature (TDPV(0)). b) The current value of the outside temperature T_e is acquired and the change in this value from the start of the stop watch activation is calculated (step 307).

In practice, at each n-th sampling the change in outdoor temperature from the initial value is updated as DT_b( ) = T_b( ) - T_b( 0). c) The current ambient temperature value measured by the sensors (Ts(n)) is acquired, the difference with respect to a previously measured temperature value (Ts(n-l)) is calculated (step 308) and this is added to the differences measured in the previous steps. In practice, at each n-th sampling the following telescopic sum is calculated:

At the first sampling step n=0, since there is no Ts(-l) value, we conventionally take Ts(-l)=Ts(0).

Next, step 309, the parameter k_p, LITER is calculated, as:

wherein

AT_DSp is the moving average of the difference of the set-point value of the delivery temperature calculated in step 306.

AT_e is the change in outside temperature calculated in step 307; å AT_S is the sum of the ambient temperature changes calculated in step 308.

At is the time interval measured by the stopwatch from when it is started (step 309) to when it is stopped(step 314). a is a constant that can either be the last value k_{d i-1} of the parameter k_d stored by the module 202, or be zero, or be the current value of k_d,i calculated by one of the estimation procedures described above. In particular, at the first cycle of parameter estimation, i.e. when i=0, it is possible to either take k_d=0, or take the value of k_d from the other estimation procedure, e.g. from the Kalman filter or the recursive least squares method or other regression method pertaining to machine learning.

Next, the method comprises checking whether the sensor measurements are valid (step 310), whether the boiler is on (step 311) and checking (step 312) whether AT_D at step n (i.e. T_DSp(n) minus Topv(n)) is above a threshold which may be equal to the threshold e considered at step 302, or more preferably be a percentage of the value that AT_D had at the beginning of the estimation process, i.e. when n=0. If all three checks are satisfied, then the value of n is increased (step 313) and the ambient temperature (Ts), outdoor temperature (T_e) and the delivery temperature set- point (T_DSP) measurements are re-sampled. Steps 305 to 313 are then repeated cyclically until one of the three checks in steps311,312 and313 fails.

When the sensor measurements are invalid, the boiler is switched off or the delivery water temperature falls below a dynamic threshold value - for example a threshold value lower than the maximum value reached by a predetermined amount - such as a value less than 80% of the maximum value reached -, then the method involves stopping (step 314) the stopwatch started in step 307 and storing - at least temporarily - a new k_p,i value which is then used to estimate the ambient temperature (Tv) inside the building.

In the example in Figure 4, the method envisages a number of verifications before providing the estimate of k_p,i as output. First of all, it is checked (step 315) whether the overall change in ambient temperature measured by the sensors (å AT_S) is greater than a predefined threshold d, e.g. 0.4 °C.

If this is not the case, then the temperature variation inside the building is not considered large enough to cause a change in the previously calculated model, so the method goes to step 318 and a value k_p,i equal to the mean value of k_p is provided as output to the algorithm. This mean value is calculated from the previously calculated and stored values of k_p,j (with j between 0 and i-1).

If, on the other hand, å AT_S ³ d, then the method involves carrying out (step 316) further checks on the calculated value k_p,i_rrER. In particular, it checks whether a) k_p_iTER is greater than zero, but at the same time less than the mean value of kp plus an uncertainty of up to 3 times the standard deviation of kp, or b) k_p_iTER is greater than 0, and the mean value of k_p is zero

The mean value and standard deviation of k_p are calculated from the previously calculated and stored values of k_p,j (with j comprised between 0 and i-1).

If these two conditions are also verified, then the estimation method gives as output k_p,i = k_p_iTER (step 317). Otherwise, the method goes to step 318 and a value k_p,i equal to the mean value of k_p is provided as the output to the algorithm.

Once the i-th estimate of k_p has been generated, the method moves on to estimate the next parameter k_p, so the value of the counter i is increased by one unit (step 319) and all other parameters of the algorithm, e.g. counter n, are re-initialised (step 320) as described above in step 301. Then the method returns to repeat the checks of steps 302, 303 and 304 and proceeds to calculate the next value k_p,i+i.

As mentioned above, each i-th estimate of k_p generated by the method described above with reference to figure 4, is compared with the estimate obtained by a different procedure, e.g. using a recursive system for parameter estimation, in particular a Kalman filter, a recursive least squares identification system with an oblivion coefficient or a machine learning algorithm of the regression type - for example, a support vector regression algorithm or a neural network.

In detail, if there is a discrepancy between the k_p parameters estimated by the two estimation procedures, then the model is out of date. Consequently, the ambient temperature estimate Tv is calculated using the k_p and k_d values calculated in the previous iteration of the method described above. Differently, if there is a match between the k_p parameters estimated by the two estimation procedures, the model is updated and the estimation module 202 and the ambient temperature estimation Tv is performed based on the newly calculated k_p and kd values. The ambient temperature estimate Tv is then provided by the estimation module to the selection module 200. The selection module 200 is configured to verify that the estimated value Tv substantially corresponds to the measured temperature Ts (at the times when the measurement Ts is available). If so, the selection module 200 requires Tc to correspond to Tv. If this is not the case, the model is considered to be unreliable and Tc therefore corresponds to Ts - until a subsequent check of the value Tv proves the reliability of the model. Preferably, If the estimate Tv proves not to be consistent with the measurements Ts, the sampling period of the sensors Ts is updated to more frequent values, as described above.

In one embodiment, the correspondence between the parameters k_p estimated by the two estimation procedures is considered verified if the two k_p values estimated by the two estimation procedures differ by less than a threshold value. Preferably, the threshold value is set substantially equal to the lower value k_p estimated by the two estimation procedures, i.e:

\k_{P l} - k_P 21 < min{k_p , k_{p 2}} where k_p,i is the k_p value estimated by the first estimation procedure, and k_p,i is the k_p value estimated by the other estimation procedure.

Even more preferably, the verification of the correspondence between the k_p parameters estimated by the two estimation procedures envisages a second requirement. In particular, this correspondence is considered verified if it is also found that the uncertainty - e.g., the standard deviation - of the lower value of the uncertainties associated with the k_p parameters estimated by the two estimation procedures is the smallest k_p value estimated by the two estimation procedures, i.e:

where O _,i is the standard deviation of the k_p values estimated by a first estimation procedure, and O_p,2 is the standard deviation of the k_p values estimated by the other estimation procedure.

The controller 201 may be a controller of a known type which acts on the basis of the ambient temperature value Tc received as input and provides as output the new set- point value of the delivery temperature of the vector fluid TDSP to reach the desired ambient temperature value.

Preferably, the control unit 20 is configured to operate in dual-mode: the first mode is linear control, while the second mode is on-off control. The first mode provides exact compensation for variations in heat load, while the second mode ensures rapid system response when the controlled variable - the delivery water temperature - is far from the set-point value.

In the embodiment illustrated in Figure 5, the control unit 20 in addition to the components already mentioned includes an on-off module 203 that manages the on-off control mode of the control unit, while the controller module 201 manages the continuous variable control mode, also referred to as continuous control mode for brevity in the following. In addition, the control unit includes a supervisor module 204 configured to manage the operation of the other components of the control unit 20. Preferably, the control unit 20 also comprises a planner or scheduler module 205 configured to store and impose one or more ambient temperature set-point values TSSP, for example according to a schedule defined on an hourly basis.

In the example considered, the controller 201 comprises a training module 2010 configured to operate as a min-max controller and a continuous control module 2011 configured to operate as a continuous controller of the heating system.

In particular, the supervisor module 204 is configured to enable the min-max module 2010 simultaneously with the initial activation of the heating system after installation of the heating system in the building. The min-max module 2010 makes it possible to minimise the learning time of the mathematical model for estimating the ambient temperature Tv - i.e. in the case of formula (1) the time for estimating the parameters k_p. In fact, at the first activation of the heating system, the mathematical model referred to in formula (1) above is not yet defined, as the values of the parameters k_p and k_d are not known.

Advantageously, the min-max module 2010 is configured to vary the set-point value TDSP of the delivery water temperature discretely between a minimum value TDSP_MIN and a maximum value TDSP_MAX, as illustrated in the quality graphs in figure 6 and the flow diagram in figure 7. The analysis of the disturbances in the thermal system, substantially consisting of the building 100, caused by this operating mode of the controller 201 makes it possible to obtain reliable ambient temperature estimates Tv in a short time and, at the same time, ensure that the desired temperatures are maintained inside the building.

In detail, when the heating system is first switched on following its installation, the supervisor module 203 commands the controller 201 to operate in min-max controller conditions 2010 (initial step 401).

The min-max module 2010 controls the burner 10 by setting the set-point value of the delivery water temperature to the maximum value TDSP_MAX, which determines the heating of the delivery water and consequently of the temperature measured by the sensors Si inside the building (t = 0 in figure 6 and step 403 of the flow chart in figure 7).

The maximum set-point temperature value TDSP_MAX of the delivery water temperature is maintained until the desired ambient temperature value TSSP provided by the scheduler 205 to the controller 201 is reached or exceeded or, more preferably, when an upper limit value (local maximum of the curve Ts(t) in t = ti in figure 6) equal to the ambient temperature set-point value TSSP plus a first margin ATSSP_M - for example, between 0.1° C and 0.5°C, preferably equal to 0.2° C - is reached (step 402).

When the ambient temperature Ts detected by the sensors Si reaches the set-point value TSSP or the upper limit value, the min-max module 2010 switches the set-point value TDSP of the delivery water temperature to the minimum value Tosp__Min, e.g. the minimum delivery temperature value manageable by the boiler (step 403).

This variation leads to a progressive reduction of the ambient temperature Ts inside the building. This reduction continues until the desired set-point value TSSP is exceeded or, more preferably, a lower limit value (local minimum of the curve Ts(t) in t = t2 in figure 6) equal to the ambient temperature set-point value TSSP reduced by a second margin ATssp__m - for example, between 0.1° C and 0.5 °C, preferably equal to 0.2° C - (step 404).

Once the ambient temperature set-point value Tssphas been exceeded or the lower limit value has been reached, the min-max module 2010 switches the delivery water temperature back to the maximum set-point temperature value TDSP_MAX, until the ambient temperature set-point value TSSP (local maximum of the Ts(t) in t = t3 curve in figure 6) is reached or exceeded again, and then switches back to the minimum value T_DSP-Min as described above. Preferably, the procedure is iterated until an off condition of the heating system is reached (step 405). In this case, the on-off module 203 forces the burner 10 to switch off (step 406), for example by bringing the set-point value TDSP of the delivery water temperature to a minimum value or, alternatively, by dropping (bringing to zero) the boiler start-up consent (instant t = t4 in figure 6). The activation of the on-off module 203 may be caused by the supervisor module 204 upon reaching a predetermined time - for example, during night-time hours or imposed by a regulation - by means of a command provided to the on-off module 203 or by bringing the ambient temperature set- point value TSSP provided by the scheduler 205 to the on-off module 203 to a minimum value. The operation then returns to the min-max module 2010 of the controller 201 at the next switch-on of the heating system (step 407) - for example, at a predetermined time in the morning - (instant t = ts in figure 6). In the same way as switching off, the heating system can be switched on by means of a start command given to the min-max module 203 or by bringing the set-point value TSSP to a desired value greater than the current ambient temperature value.

Advantageously, starting with each transition from the minimum value T_DSP-min to the maximum value TDSP_MAX until the next, opposite transition from the maximum value TDSP_MAX to the minimum value Tosp_min the estimation module 202 is configured to detect the trend of the ambient temperature value and to refine and estimate a corresponding k_p value by means of the k_p estimation procedure described above in relation to figure 4 so as to progressively refine the system model.

The controller 201 operates in min-max controller mode as just described, as long as the control unit 20 does not have sufficient history to calculate an average k_p value that is stable over time, for example until as 5 k_p values have been calculated.

Advantageously, the maximum value TDSP_MAX and the minimum value TDSP_min are chosen to minimise the number of transitions required to process a sufficient number of k_p value estimates - making the model usable - while allowing the ambient temperature in the building to be controlled to ensure user comfort. The Applicant has determined that it is possible to determine the optimum maximum value TDSP_MAX and minimum value T_DSP-min by applying the theory of asymptotic properties of prediction error models (PEMs) and the asymptotic theory of Ljung (1985), as defined in L. Ljung, "Asymptotic variance expressions for identified black-box transfer function models," in IEEE Transactions on Automatic Control, vol. 30, no. 9, pp. 834-844, September 1985 and L. Ljung and Z. Yuan, "Asymptotic properties of black-box identification of transfer functions," in IEEE Transactions on Automatic Control, vol. 30, no. 6, pp. 514-530, June 1985.

In particular, the Applicant has determined that it is possible to minimise the time required to obtain the model parameters and to guarantee the comfort of the users by imposing a minimum value Tosp__min comprised between 30° C and 50° C, preferably equal to 35° C or equal to 40° C and a maximum value TDSP_MAX comprised between 65° C and 80° C, preferably equal to 70° C or 75° C, in the case of heating systems comprising a boiler that heats water.

When the estimation module 202 computes a reliable model - that is, makes available an average value of k_p and, thus, of k_d that is stable over time, the supervisor module 204 notifies the controller 201 to switch the control mode from min-max to continuous variable control by deactivating the min-max control module 2010 and activating the continuous control module 2011.

In the embodiments, continuous control mode is defined by exploiting the model - described above - used to represent the thermodynamic system of the building and provide the ambient temperature Tv estimate. Preferably, it is contemplated to determine one or more control parameters used by the linear control module 2011 on the basis of the k_p value and, preferably, the k_d value determined by the estimation module 202.

Advantageously, the integral or pseudo-integral nature of the model considered above makes it possible to employ a particularly simple but at the same time particularly effective controller 201. In detail, when the two estimates of the k_p and k_d values converge as described above, these k_p and k_d values are considered sufficiently accurate to allow reliable self-timing, or autotuning, of the controller 201 to the thermal system - i.e., the building - being controlled. Since the thermal system to be controlled is approximated by an integral or pseudo-integral model, it is possible to apply control by integral system, in particular, it is possible to initially supply the energy necessary to reach the set-point value TSSP in a predefined time, and then the flow temperature can be brought to the minimum equilibrium value necessary to keep the temperature stable by compensating the thermal load of the building - that is, by compensating the heat losses of the building mainly due to the difference between the ambient temperature Ts and the outside temperature T_e. Minimisation of the delivery temperature for most of the operating period of the heating system results in minimisation of the return temperature and thus maximises the efficiency of the boiler.

In more detail, the controller 201 in the continuous controller mode envisages determining - on the basis of the values T_c, k_p, k_d and, preferably, the value T_e - a set-point value TDSP of the delivery temperature of the water such that the ambient temperature in the building reaches the desired set-point value TSSP within a predetermined time (t = t_\ in figure 8) - for example, within one hour after the system is switched on. Once the desired set-point value TSSP has been reached, the set-point value TDSP of the delivery water temperature is adjusted to maintain the ambient temperature value at, or at least around, the set-point value TSSP. In particular, the controller 201 is configured to determine - on the basis of the values T_c, k_p, k_d and, preferably, the value T_e - the minimum set-point value TDSP of the delivery water temperature which allows the ambient temperature to be maintained at the set-point value TSSP in the face of variations in the observed values Tc and T_e (time interval between t = t_\ and t = to_FFi in figure 8), until the predetermined switch-off of the heating system (time interval between t = to_FFi and t = ti in figure 8).

In a particularly economical and compact embodiment, the continuous module 2011 of the controller 201 (as schematically illustrated in figure 9) comprises a proportional-integrative block 2012 whose operating parameters are dynamically determined on the basis of the value k_p and the value k_d determined by the estimation module 202. Preferably, the calculation of the control parameters - i.e. a proportional coefficient kc and an integrative coefficient Ti - of the proportional-integrative block 2012 is based on the technique known as lambda-timing.

Even more preferably, the coefficients kc and Ti are determined from a connection block 2020 of the estimation module 202 and supplied to the proportional-integrative block 2012.

In this case, the proportional coefficient kc of the proportional-integrative module 2012 is calculated as:

2 l+Atr k_c (fe_p+2ff_p)-(l+Atr)² (3) where A is a time within which the set-point value TSSP of the ambient temperature is to be reached, Atr is a time indicating a delay between a change in the delivery temperature and a change in the ambient temperature value Ts, while o_P is the uncertainty associated with the value of k_p defined as the standard deviation associated with the set of k_{p, i} values acquired.

In addition, the integrative coefficient Ti of the proportional-integrative block 2012 is calculated as:

Ti = 0.8 - (22 + Atr) (4)

Preferably, the linear control module 2011 of the controller 201 also comprises a feed-forward block 2013 whose operating parameter - i.e., an advance coefficient - is determined based on the value k_p and the value k_d determined by the estimation module 202.

Even more preferably, the advance coefficient k/f of the feed-forward block 2013 is also determined by the connection block 2020 of the estimation module 202.

For example, the advance coefficient of the feed-forward block 2013 is calculated as: k = ^kd ff (fcp + 2s_n) (5)

Advantageously, the estimation module 202 implements the following procedure for processing the control parameters of the proportional-integrative block 2012 and the feed-forward block 2013 (a flow chart of which is shown in figure 10).

At each iteration of the procedure that calculates the value k_p and the value kd, performed by the estimation module 202 an indication of the result of the congruence check of the values k_p calculated according to the two procedures is provided to the supervisor module 204 (initial step 501).

If the values k_p and kd are reliable (step 502), the connection block 2020 of the estimation module 202 calculates (step 503) the coefficients kc e Ti of the proportional- integrative block 2012 and the advance coefficient k/f of the feed-forward module 2013 according to formulas (3) - (5) above on the basis of the values k_p and k_d. Conversely, if reliable k_p and kd values are not available, the supervisor module 204 is configured to force the use (step 504) of default coefficients kc, Ti and kff - for example, stored in a memory area of the control unit 20.

During operation in continuous mode, the controller 201 receives as input the ambient temperature value Tc from the selection module 200 and, preferably, the outside temperature value of the building T_e. On the basis of these inputs and the set-point value TSSP of the ambient temperature provided by the scheduler 205, the continuous module 2011 controller 201 determines in real time the set-point value TDSP of the delivery water.

On the basis of the set-point value TDSP of the delivery water, the control unit 20 operates the burner 10 of the building's heating unit in such a way as to reach the desired set-point value TSSP within the set time l and thus maintain the building's ambient temperature at the set-point value TSSP or, at least, in its vicinity.

In the exemplary case of the controller 201 equipped with the proportional- integrative block 2012 and the feed-forward block 2013, the set-point value T_DSP of the delivery water is defined by the combination of the outputs of the modules 2012 and 2013 as described below and illustrated by the flow chart in figure 11.

As soon as the system is switched on, the feed-forward block 2013 is configured to compensate (step 601) for heat dispersion due to the difference between the ambient temperature inside the building and the outside temperature. In particular, the feed forward block 2013 provides an output value Tff given by the combination of the outdoor temperature T_e and the advance coefficient k .

The proportional-integrative block 2012 is configured to cancel (step 602), or at least minimise, a difference between the ambient temperature and the desired set-point value TSSP. In detail, it provides an output value Tpi provided by the difference between the ambient temperature set-point value TSSP and the ambient temperature value T_c provided by the selection module 200 combined with the control coefficients kc and Ti:

The output values of the blocks 2012 and 2013 are then combined (step 603), preferably summed, with each other to determine an overall output value corresponding to the set-point value TDSP of the delivery water temperature (TDSP = Tpi+T_/i).

The set-point value TDSP of the delivery water temperature is then recalculated at each (step 604) control cycle performed by the control unit 20 (typically once per minute, more generally with a period such as to ensure a rapid response of the system to a variation of the variables observed by the control unit 20) and by first recalculating the control parameters kc, Ti and k_jj as described above (step 606) in case the variation of at least one of the values k_p and k_d is verified (step 605). The preceding steps of the procedure are iterated during the operating period of the heating system (step 607) while outside this operating period the heating system is switched off (step 608) by means of the on-off module 203 which forces the shutdown of the burner 10 - in a similar manner as described above - until the beginning of the next operating period in which the continuous module 2011 of the controller 201 forces the re ignition of the burner 10.

Preferably, at each boiler start-up, the estimation module 202 is configured to detect the trend of the ambient temperature value from start-up to the time of reaching the desired set-point value TSSP and estimate a corresponding value k_p by means of the k_p estimation procedure described above in relation to figure 4 so as to progressively refine the thermal system model.

In one embodiment, the control unit 20 is configured to exploit the thermal inertia of the thermal system of the building or parts thereof - such as, for example, the radiators 106 - in order to reduce the energy consumption of the heating system, without affecting the comfort of the users.

The estimation module 202 is configured to estimate an operating temperature TOP, which is defined as the weighted average between the ambient temperature measured by the sensors Si and the radiant temperature TRAD to which a user is subjected within a portion of the building - for example, a residential unit or a room.

In one embodiment, a reference radiant temperature TRAD is defined according to the following formula - suitable for wall-mounted radiators:

wherein the term a₁ (T_c — 2)is indicative of the radiant temperature of the floor, a₂ (T_c + 2)is indicative of the radiant temperature of the ceiling, the terma₃ T_c is indicative of the radiant temperature of the walls and the term b T_DPV is indicative of the radiant temperature of the radiator 106 or radiators 106 positioned in the portion of the building. Furthermore, the coefficients a₁, a₂, a ₃, b are associated with floors, ceilings, walls and radiators, respectively, and are coefficients proportional to the area of each surface in relation to the total area within the portion of the building considered. In one embodiment, the reference radiant temperature TRAD is calculated by considering a square room with a side of 4 m and a wall height of 2.7 m and a single radiator was considered with a radiant surface essentially equal to 1 m². The operating or perceived temperature is then, as known in the literature of the sector, calculated as the average between the measured ambient temperature and the radiant temperature, i.e: _ TC + TRAD ¹ OP — (9)

The radiant temperature of the radiators 106 is set equal to the measured delivery water temperature TDPV, while the boiler is switched on. Otherwise, the radiant temperature of the radiators 106 becomes unknown once the boiler is switched off, as the circulation of water in the heating system is interrupted. Advantageously, the estimation module 202 is configured to calculate an estimate of the radiant temperature of the radiators 106 as a function of time and/or a cooling time required to reach a predetermined final temperature.

In one embodiment, the estimation module 202 is configured to calculate an estimate of the radiant temperature of the radiators 106 based on a model defined based on the characteristics of the radiators (e.g. size and constituent materials) and the temperature of the radiators at the time of boiler lockout (e.g. set equal to the temperature of the water returning to the boiler).

Alternatively, the control unit 20 is configured to allow selection of a radiator cooling curve from the following options: cautionary curve (curve A shown in figure 12), intermediate saving curve (curve B) and high saving curve (curve C). In detail, such curves make it possible to determine the time required for the radiators 106 to reach a desired final radiant temperature value, for example corresponding to the minimum delivery water temperature manageable by the boiler, or the ambient temperature, considering the radiator immersed in fluid at the ambient temperature, for example the desired set-point value TSSP, starting from an initial radiator temperature, for example estimated corresponding to the measured temperature TR,S of the water returning to the boiler.

In one embodiment, the curves are described by the following parametric formulae - derived using a concentrated parameter approach and using the Biot number -, where an ambient temperature of 20° C and a final radiant temperature of 40° C have been assumed: a) cautionary curve t_A = -0,0306 Tls + 5,4277 T_{R S} - 166,81; (9) b) intermediate saving curve t_B = -0,0431 Ti_g + 7,6459 T_{R S} - 234,99 , and (10) c) high-saving curve t_c = -0,0583 Tls + 10,198 T_{R S} - 312,18, (11) where tA, B, C corresponds to the cooling time (in minutes) required by the radiators to reach the minimum temperature of 40° C since the boiler was turned off and TR,S is the temperature of the water returning to the boiler at the instant when the supply of heat to the radiators is suspended.

In particular, each of these curves is based on a respective interpolating equation, obtained by averaging the cooling times of radiators made of aluminium, cast iron and steel, and determined for a respective radiator size selected from large (curve A), medium (curve B) and small (curve C). The Applicant has determined that such curves allow for an adequate estimate of the thermal performance of the radiators after the interruption of the flow of heated water regardless of the actual characteristics of the radiators actually installed based on the desired degree of energy saving - thus without requiring the installation engineer to enter precise data regarding the radiators installed in the building.

In general, the operating temperature TOP has a faster dynamic than the ambient temperature as the radiators 106 heat up faster than the surrounding air. By using the operating temperature TOP, it is therefore possible to determine boiler shutdown intervals

- during the daily operating period thereof - which are generally longer than if only the ambient temperature were used as a reference. This makes it possible to substantially reduce the energy consumption of the heating system in a way that is transparent to users

- i.e. without substantially changing the temperature perceived by users - and thus the level of comfort.

With reference to the graphs in figure 13 and the flowchart in figure 14, starting from the switching on of the heating system the continuous module 2011 of the controller 210 is configured to bring the operating temperature TOP - instead of the ambient temperature - to the desired set-point value TSSP within the time l (step 701 and time interval from t = 0 to t = ty in figure 12).

Once the operating temperature value TOP has reached the set-point value TSSP or, more preferably, an operating upper limit value equal to the set-point value TSSP plus a predetermined margin DTOR_M - for example, between 0.05° C and 0.2° C, preferably equal to 0.1° C - or the continuous module 2011 maintains the set-point value TDSP of the delivery water temperature at the minimum value Tosp__min - capable of compensating for variations in the outside temperature Te as described above - for a predetermined time interval Atop (step 702, time from t = t_\ to t = ti in figure 12), the on-off module 203 forces the boiler to be turned off - for example, the on-off module 203 imposes a minimum set- point temperature TDSP for the delivery water temperature (step 703).

The on-off module 203 keeps the boiler off until the value Tc of the ambient temperature is lower than the set-point value TSSP by a predetermined margin value ATSSP_L - for example, between 0.1° C and 0.5 °C, preferably equal to 0.2° C - (step 704, time interval from t = ti to t = t2 in figure 12), when the continuous module 2011 imposes a set-point value TDSP of the delivery water temperature different from the minimum (step 705), in particular such as to quickly bring the value of the operating temperature TOP back to the set-point value TSSP. Preferably, it is planned to impose a dead-hand Dίi;_ΐ; between boiler shutdown and subsequent restart, so as to limit the frequency of switching the boiler on and off.

The steps of the procedure as described above are iterated during the operating period of the heating system (step 706) while outside this operating period the heating system is switched off (step 707, from t = tom in figure 12) by means of the on-off module 203 which forces the shutdown of the burner 10 - in a similar manner as described above - until the beginning of the next operating period in which the continuous module 2011 of the controller 201 forces the re-ignition of the burner 10.

Alternatively or additionally, the control unit 20 is configured to detect a shutdown of the boiler imposed by an internal circuitry of the boiler - the so-called control level 1 - when a limit value is exceeded, preferably equal to the set-point value TDSP of the delivery water temperature increased by a margin value - for example, equal to 4° C (step 801 of the flow chart in figure 15). Upon detecting the shutdown, the control unit 20 is configured to prevent a restart of the boiler controlled by the internal boiler circuitry (step 802) until it detects that the operating temperature is substantially equal to a desired temperature - for example, substantially equal to the ambient temperature- or the cooling time has elapsed (step 803). For example, the on-off module 203 is configured to forcibly keep the boiler off until the value of the operating temperature TOP equals the value of the ambient temperature Tc or is equal to the average of the ambient temperature and the minimum radiant temperature, or after a time corresponding to the cooling time t_A, B, c has elapsed. Subsequently, the continuous module 2011 of the controller 201 imposes a set-point value TDSP of the delivery water temperature allowing the boiler to be reactivated (step 804).

The implementation of at least one, preferably both, of the procedures described above allows the boiler to be kept off for as long as possible by exploiting the radiation of the heat accumulated by the radiators - a condition known as 'coasting' in technical jargon. Thanks to the coasting obtained in this way, it is possible to guarantee the comfort of the users and, at the same time, prevent continuous switching on/ off due to the level 1 circuitry of the boiler, which is inefficient both from an energy and thermal point of view.

The invention thus conceived is susceptible to several modifications and variations, all falling within the scope of the inventive concept.

For example, in one embodiment, it is envisaged to control the temperature of the vector fluid based on the temperature estimates generated by the mathematical model when there is a breakdown in communication between the sensors located inside the building and the control unit or the data received at the control unit is corrupted.

A simplified embodiment (not illustrated) does not include the min-max module 2012. In this case, the control unit 20 contemplates using the on-off module 203 to perform the initial procedure necessary to acquire the data required to allow the estimation module 202 to construct a reliable model of the building's thermal system.

In an alternative embodiment (not illustrated), the control unit 20 provides for combining, e.g. summing, an outdoor temperature compensation curve T_e - like a climate curve - to the minimum value Ti sp__min and to the maximum value TDSP_MAX. This improves the operating efficiency of the system during min-max operation, at least partially compensating for variations in the thermal load due to outside temperature T_e variations.

In one embodiment, it is planned to implement the on-off module 203 adaptively. In particular, the on-off module is configured to calculate a variable set-point value on the basis of the integral or pseudo-integral mathematical model developed by the module 202. In detail, the on-off module 203 is configured to calculate a regulation value ei to be combined, in general subtracted, from the set-point value TSSP, leading to an earlier shutdown of the boiler and thus reducing the consumption of the heating system.

Preferably, the variable set-point value is processed starting from the ambient temperature set-point value TSSP or from a set-point value of the operating temperature on the basis of the values k_p and k_d processed by the module 202 and the delay time to of the heating system - indicative of thermal inertia of the heating system -, i.e., the time required to heat the radiators and heat the room in the building.

In particular, it is possible to identify the threshold value ei according to the following relationship:

£·; - k_p t_D AT_D + k_d - t_D DT, (12)

The Applicant has determined that it is possible to assume a delay time to substantially between 5 and 15 minutes, preferably 10 minutes, in the case of a heating system using water as the vector fluid and radiators as the heating elements.

In a different embodiment instead of the adaptive threshold ei the operating temperature and a fixed threshold is used to achieve the same control purpose.

In a different embodiment (not illustrated), the on-off module 203 is configured to also control the reaching of the set-point value TSSP of the ambient temperature by imposing operation at maximum boiler power - for example, by imposing a set-point value TDSP of the delivery water temperature equal to the maximum temperature reachable by the boiler - in order to minimise the time to reach the set-point value TSSP at the cost of higher power consumption during the start-up phase.

Of course, there is nothing to prevent providing a different continuous control module 2011, for example, including a PID block or configured to implement predictive control instead of a PI block and an FF block. Accordingly, the connection block 2020 of the estimation module 202 will be configured to calculate and provide appropriate control parameters to the continuous control module 2011. Preferably, predictive control is a Model Predictive Control (MPC), optionally configured to acquire future external temperature values, for example, from a remote entity external to the heating system such as a server implementing a weather forecast service.

Similarly, there is nothing to prevent the use of a criterion other than l-tuning to determine the parameters of the proportional-integrative block 2012, just as the feed forward block 2013 may involve a more complex transfer function including, for example, one or more filters.

In an alternative embodiment, the feed-forward block involves acquiring at least one predicted outside temperature value T_e - for example, provided by an external entity, as described above - and calculating an output value T^ as a function of the current outside temperature value and one or more future temperature values. Preferably, the output value T¾r will be calculated as the sum of the products of each outside temperature considered by a corresponding advance coefficient. Preferably, each advance coefficient is calculated by means of the k_p and k_d values estimated by the model based on the outside temperature considered.

In other embodiments, it is envisaged to use the operating temperature TOP as the ambient reference temperature even without implementing the coasting procedures described above. In dual mode, there is nothing to prevent the implementation of coasting procedures using a set-point value TSSP plus an operating margin - for example, in the order of tenths of a degree Celsius.

In one embodiment (not shown), the control unit includes a diagnostic system, or fault diagnosis, configured to analyse the performance of the continuous control module 2011 in order to detect any malfunctions - for example, too slow a response, excessive ambient temperature fluctuations, boiler in maximum or minimum saturation, etc. - and, in response to these malfunctions, switch the building management to normal on-off control - and, in response to these malfunctions, switch building management to normal on-off control. For example, the fault diagnosis system can be implemented by the supervisor module 204.

In addition, the control unit can implement an early switch-off procedure to reduce the length of the switch-on period on the other hand, exploiting the thermal inertia of the radiators and possibly of the building itself. This reduces the overall consumption by reducing the overall daily operation time of the heating system. Advantageously, the optimal switch-off advance times are calculated on the basis of the processed thermal system model of the building, and the outside temperature, applying the principle of one- step prediction logic.

Similarly, the control unit can implement a procedure to vary the switch-on time according to the current and/or predicted outside temperature (acquired from an external entity as described above). In this way, it is possible to adapt the switch-on timing of the heating system to the actual environmental conditions, making it possible to reduce the power required to reach the desired ambient temperature in unfavourable climatic conditions or to delay the heating system switch-on in favourable climatic conditions, thus reducing the operating period of the heating system.

Furthermore, in a highly configurable embodiment (not illustrated), the control unit 20 is configured to receive - for example, from an installation technician via a user interface - characteristic parameters of the radiators 106 installed in the building or average values of the characteristic parameters if radiating elements of different types are installed in the building. These characteristic parameters include, but are not limited to, a radiator size - for example, selectable between small, medium and large size depending on the volume of the radiator - and a radiator material - for example, selectable between aluminium, cast iron and steel. The control unit 20 is then configured to calculate the radiant temperature of the radiators and/ or its trend over radiator time according to the entered characteristic parameters. On the contrary, there is nothing to prevent - in a simplified embodiment (not illustrated) - defining the operating temperature TOP as equal to the ambient temperature plus an offset based on an estimate of the thermal characteristics of the radiators 106.

It will be clear to a person skilled in the art that control unit 20 can be equipped with one or more additional modules. In the example illustrated in figure 15, the control unit 20 comprises a reference trajectory module 206 and, preferably, a comfort estimation module 207.

In detail, the scheduler 205 provides temperature set-point values TSSP to the reference module 205 which is configured to define a time-variable set-point value Tssp(t), which assumes the desired set-point values and defines transients to minimise the energy consumed by the system during the transition from one set-point value to the next. For example, when the thermal load is compensated and the action of the proportional- integrative block 2012 and feed-forward block 2013 is reduced, the rate of growth of the ambient temperature towards the desired set-point value TSSP is decreased, in order to show the proportional-integrative block 2012 a smaller control error and thus minimise the control effort. Thanks to this configuration, it is possible to slow down the achievement of the set-point value TSSP, reducing the energy consumed by the heating system, without causing discomfort to the building's users.

The comfort estimation module 206 allows the identification of a temperature perceived by the user based on a plurality of input information - in accordance with P. O. Fanger, " Thermal comfort analysis and applications in environmental engineering" , R.E. Krieger Pub. Co., 1982. In detail, it is planned to control and regulate a thermo- hygrometric comfort variable called TPMW instead of the ambient temperature of the building. This TPMW variable is calculated by combining a plurality of measurements taken by sensors and information provided by the user - e.g. through a user interface - or approximated according to season, time of day and/or intended use of the building. Preferably, the acquired information comprises two or more of: the ambient temperature, a measurement of ambient humidity - for example, by means of a humidity sensor that can be easily integrated into the ambient temperature sensors -, a radiant temperature of the radiators 106 - for example, estimated as a function of the boiler delivery temperature and the type of radiators 106 - air speed, an activity performed by the users and a type of clothing worn.

The comfort variable TPMW is calculated as a temperature perceived by the building users and is used as a reference value instead of the ambient temperature value Tc in the procedures described above. In this way, it is possible to reduce, or at least calibrate, the consumption of the heating system while ensuring that the user perceives a comfortable temperature.

In one embodiment, the heating system comprises two or more separate heating circuits. In this case, the control unit 20 is configured to perform an optimisation of the thermal balancing of the circuits, which involves removing heat from the more thermally advantaged or lower activity circuits and moving it to the more thermally disadvantaged or higher activity circuits, so as to produce further energy savings in the overall heating system.

Naturally, one or more components of the control unit can be implemented with hardware, firmware, software or combinations thereof.

It will in particular be clear to a person skilled in the art that although the description refers to a central heating system, the controller and/ or methods described above can be implemented in other equipment of a different heating system such as a wall thermostat, a condensing boiler for individual residential units, as well as an HVAC system, a heat pump or a remote control system.

Again, although reference has only been made to radiators, it will be clear that procedures exploiting the thermal inertia of radiators are applicable to any heating organ with thermal inertia - for example, radiant tubes or panels integrated into the floor, ceiling or one or more of the walls of the building - without substantial modifications.

Claims

1. Method for controlling the temperature of a vector fluid intended for heating a building, wherein

- a desired value of an ambient temperature inside the building is set (1001),

- values of ambient temperature of the building by means of at least one wireless sensor temperature are detected (1002), the method being characterized by

- generating (1003), by using a mathematical model, estimates of the ambient temperature inside the building,

- controlling (1004) the temperature of the vector fluid based on the ambient temperature values measured by the sensor and based on the temperature estimates generated by the mathematical model in order to reach the desired ambient temperature value inside the building,

- comparing periodically one of said temperature estimates with a temperature value measured by the sensor, and

- reducing a frequency with which the sensor performs the measurements in the case the estimated temperature and the measured temperature differ less than a predetermined tolerance threshold, or increasing the frequency with which the sensor performs the measurements in the case the estimated temperature and the measured temperature differ more than the predetermined tolerance threshold.

2. Method according to claim 1, wherein the mathematical model is a model of the machine learning model based type which starts from an integral type or pseudo-integral base model.

3. Method according to claim 2, wherein the base model is

T_v(t) = J k_p - ( T_DSP - T_ref) dt + f k_d (T_e - T_ref) dt where k_p and k_d are two parameters that vary over time, TDSP is a set-point value of the temperature of the delivery vector fluid, T_e is a temperature outside the building, T_ref is a predetermined reference temperature, preferably comprised between 18° C and 22° C and more preferably equal to 20° C.

4. Method according to claim 3, wherein k_p is estimated by means of two estimation procedures; if the estimate of k_p of the two procedures differs less than a predetermined threshold, then the values of k_p and k_d generated by one of the two estimation procedures are used.

5. The method of claim 4 wherein according to a first of the estimation procedures the i-th estimate of k_p is

wherein

AT_DSP is the moving average, measured during the estimation procedure, of the difference in the value of set-point of the temperature in the carrier fluid flow;

AT_e is the variation of the external temperature between the end and the beginning of the estimation procedure; is the sum of the changes in ambient temperature measured by the sensor during the whole estimation procedure;

Dί is the time interval measured from the beginning to the end of the estimation procedure; a is a constant.

6. Method according to claim 5, wherein a is the last value of the parameter k_d calculated with the first estimation procedure, or it is equal to zero, or it is the value of kd calculated by the other estimation procedure.

7. Method according to claim 5 or 6, wherein the other estimation procedure is a recursive algorithm for estimating the parameters, in particular a Kalman filter, or an identification system of recursive least squares type with oblivion coefficient or another estimation method belonging to the category of machine learning.

8. Method according to any one of claims 3 to 6, wherein the external temperature T_e is a datum acquired from the internet or from an external database or from an external sensor.

9. Method according to any one of claims 1 to 8, wherein said sensor is one of a plurality of sensors, in particular one of at least three sensors, arranged inside the building, and wherein said sensor is the one detecting the lowest temperature.

10. Method according to claim 9, wherein said plurality of sensors are positioned in the colder parts of the building, said points being identified based on the exposure of the building in conditions of building having an average occupancy and not being heated.

11. A method according to any one of the preceding claims, wherein the measurements of temperature sensor and are considered valid and used for the estimation of the model parameters in the case in which are verified all, or at least a portion, of the following conditions:

• if conditions of open window in a room where the sensor is located are not determined.

12. Method according to claim 11, wherein an open window condition is verified by checking the trend of the individual room temperature measurements of a given sensor with respect to the average trend of the measurements of the same sensor, wherein a slow deviation towards lower values of one or more measures is identified as an open window condition, and wherein the method comprising temporarily excluding the sensor from the measurement selection logic for the controller, and wherein the signal provided by the sensor is restored as usable when the open window condition ends.

13. A method according to any one of preceding claims, wherein the step of controlling (1004) the temperature of the carrier fluid based on the ambient temperature values measured by the sensor and based on temperature estimates generated by the mathematical model comprises:

- using the temperature estimates generated by the mathematical model as a control variable when communication between the at least one sensor and a control unit that control the temperature of the vector fluid is interrupted.

14. Heating system (1), comprising conduits (103,104) for transporting a vector fluid inside a building, means (10) for heating the vector fluid, at least one wireless sensor (SI, S2, S3) capable of transmitting ambient temperature measurements inside the building (100), a control unit (20) operatively connected to the wireless sensor to receive the temperature measurements and operatively connected to the heating means (10) to control the temperature of the vector fluid, characterized in that the control unit (20) is configured to implement a method of controlling the temperature of the carrier fluid according to any one of the preceding claims.