WO1999005578A1

WO1999005578A1 - Load-based adaptive gain control

Info

Publication number: WO1999005578A1
Application number: PCT/US1998/014885
Authority: WO
Inventors: Gideon Shavit
Original assignee: Honeywell Inc.
Priority date: 1997-07-22
Filing date: 1998-07-17
Publication date: 1999-02-04
Also published as: AU8574698A; WO1999005578A9

Abstract

A system which accounts for load changes using adaptive gain control techniques. Control of the process is defined by a control equation, having one or more terms, each of the one or more terms having associated therewith a gain variable. Improved control of the process is achieved by providing a device for adjusting the gain variables associated with one or more terms in the control equation. Modification of the gain variables is based on changes in the load of the process. The described system of adaptive gain control may be combined with other adaptive gain control techniques, such as gain scheduling, to further improve control characteristics.

Description

LOAD-BASED ADAPTIVE GAIN CONTROL BACKGROUND OF THE INVENTION

The present invention relates to a novel adaptive control apparatus. More particularly, the applicant's invention is an adaptive control apparatus which modifies control equation gain variables as a function of the load in the system being controlled.

One basic method of achieving control of a complex process is by using a PID control equation or some variation thereof. In a system controlled by a PID control equation, a value proportional to the error of a measured variable, the integral of this error, and the derivative of this error are used to achieve control of a process. One of skill in the art will recognize that one or any combination of these terms may be used to achieve the desired control for a particular process. The standard equation for PID control might be written as:

where: V = PID output

K_P = Proportional gain variable K_τ = Integral gain variable K_D = Derivative gain variable e - Deviation of a measured variable from the desired value for that variable.

The PID output is thereafter usually scaled and applied as a control signal to a control element capable of altering the process.

The I , K_j, and K_D gain variables ultimately affect the overshoot, bandwidth and reaction time of the controller to changes in the process, and generally determine the control response characteristics of the process.

In adaptive gain control systems, the gain variables, rather than remaining at a single value throughout a particular control situation, are altered based on various events or conditions in a system. Adaptive gain control is one of several techniques used to allow PID control to be effective in basically non-linear control situations even though the PID control mechanism is best suited for linear processes. In both adaptive and non-adaptive PID control, the system designer will determine the control response characteristics desired and thereafter choose values for the gain variables to achieve these characteristics. Thus, when a non-linear system is encountered, the system designer determines several sets of gain variables, one set achieving the desired control response characteristics for a different range of process conditions or events. The ranges selected separate the process conditions or events into approximately linear segments. Another way of characterizing the ranges is that all process conditions or events in a range have similar sensitivity to process changes. Of course, correlating the PID equation on to a non-linear process means the ranges chosen do not actually exhibit linear characteristics, but are an approximation which, even including the approximation error, still achieve the desired control response characteristics. The designer's task of selecting ranges is thus dependent at least in part, on how difficult the desired control response characteristics are to achieve.

Reference to sets of gain variables hereafter will be used to describe the gain variables associated with all terms for a particular control equation, for a particular range of process conditions or events. In the standard PID equation, for instance, there would be three gain variables in each set, K_τ, K_D and K_P.

A common use for adaptive gain control is where there more than one device makes up the process. For example, a single controller used to operate a heating device and a cooling device is suited to adaptive gain control. This is true because it will be unlikely that both of these devices will optimize at the same set of gain variables. Adaptive gain control is also useful however where a single device constitutes the process since many devices exhibit different characteristics for different ranges of operation. For example a device operating at its maximum output will have different characteristics than it would at an average or low output.

One specific method of modifying the gain variables in the control equation is known as gain scheduling. In this method, the system designer identifies, for example, times of the day, month, year or other period which is significant because the process may have different characteristics during each such period. The number of periods may in part depend on how complicated the process is, but may also depend on how difficult the control response characteristics chosen by the system designer are to achieve. As already indicated, adaptive gain control is useful where more than one device makes up the process each device operating under different conditions. Improved control is achieved by storing a set of gain variables for each device in the process, and then choosing the set of gain variables optimized for that device when that device is operating in the system. A common example would be a control system used to control both the heating and cooling of the system, since heating and cooling of a space will typically be performed using separate heating and cooling devices. The controller determines when the setpoint, or desired state of the system is less than or greater than the actual state of the system. If the desired state is less than the setpoint, the set of gain variables optimized for the heating device is used in the control equation to create the control signal because the heating device will be operating in the system. If the desired state is greater than the setpoint, the set of gain variables optimized for the cooling device is used in the control equation to create the control signal, since the cooling device will be operating in the system. A common application of gain scheduling may be in an office building which is occupied during the day and essentially vacant at night. The sudden changes in occupancy result in a complicated process characteristics which are not optimally dealt with using a single set of gain variables for the control equation. Consequently, the system designer may select one set of gain variables for night, one set for day, and possibly a third and fourth sets for the transition periods between night and day. By then using a signal input to the controller which indicates time, the more efficient set of gain variables may be used in the control equation for that time period.

The success of adaptive gain control hinges on detecting changes in the process which will effect the efficiency of the chosen gain variables. For example, in the gain scheduling method, the system designer has assumed that the system is similar from day to day, month to month, or other period. In fact, in few instances does the system exhibit periodic characteristics. For example, in a gain scheduling system, changing the gain variables in the morning and in the evening, would probably not operate optimally on both weekends and weekdays, as there likely would be differences in the conditions of the system during these two time periods. A gain scheduling scheme which accounts for weekend differences from weekdays may ignore the effect of holidays, peak vacation periods, high sick leave rates or other abnormal occupancy periods. The inherent problem is that this technique does not know what the process is actually doing, just what does normally, or at least what it did during the design of the system.

Ideally, the control system should monitor all process conditions which affect the optimization of the gain variables, and alter the gain variables accordingly. Changes in the load on the system cause significant errors in gain variable optimization. Thus a system which incorporates load changes into an adaptive control apparatus would be a significant improvement in process control.

SUMMARY OF THE INVENTION The applicant's invention thus provides a system which accounts for load changes using adaptive gain control techniques. The applicant describes an apparatus for providing adaptive control for a process having a variable load. Control of the process is defined by a control equation, having one or more terms, each of the one or more terms having associated therewith a gain variable. Improved control of the process is achieved by providing a device for adjusting the gain variables associated with one or more terms in the control equation. Modification of the gain variables is based on changes in the load of the process. In one possible embodiment, the gain variables in the control equation are determined from a look-up table. In another method, a two-variable curve-fitting program is used to characterize the best values for the gain variables at different loads. The described system of adaptive gain control may be combined with other adaptive gain control techniques, such as gain scheduling, to further improve control characteristics.

A method of the present invention operates according the apparatus described above. BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1 shows a block diagram of one prior art type of process control. Fig. 2 shows a generic block diagram of the Applicant's adaptive control apparatus.

Fig. 3 shows a block diagram for a particular embodiment of the Applicant's adaptive control apparatus.

Fig. 4 shows a plot of a energy vs. valve position for a process under several load conditions. Fig. 5 shows a plot of a energy vs. valve position for a more complicated system, including a plurality of devices in the process and under several load conditions.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Fig. 1 depicts a block diagram for a common prior art control system. In the control system, controller 1 receives at first input port 2 a desired setpoint value for a chosen controlled variable, and the measured value of the same controlled variable at second input port 3. First input port 2 is connected to a user interface, memory device, or other device (not shown) which can provide the setpoint, via first connection 4. Alternately, first input port 2 may be replaced with a setpoint value internal to controller

1. A second connection 5 connects second input port 3 to process 6. As one of skill in the art may recognize, the measured value of the controlled variable, which appears at second input port 3, represents the process output of the system. Second connection 5 will be to a sensor of some type which measures the condition of the output of the process. For example, a process may comprise a heating device, and a space being heated; the controlled variable might then represent temperature of the air in the space. Thus, the temperature of the air in the space is measured by a sensor which is connected to controller 1 via second connection 5.

Controller 1 uses the setpoint and measured variable —in the example, actual temperature in the space— to calculate a difference or error value. It is noted that in this example the measured variable is also the controlled variable. As is understood in the art, this does not necessarily need to be the case. For instance, it is known to control air flow through a space, based on temperature or humidity measurements.

In any event, control decisions are made using the difference or error value in the control equation to generate a new control output. The controller outputs a control signal, created internal to the controller, using a controller algorithm, which is supplied to control element 7 via a third connection 8. The controller algorithm may perform any number of functions, including the herein described methods for adaptive gain control. The control element is the physical device which changes the conditions in process 6, using the control signal. For example, the control element may be the actuator for a water valve in a hot water supply pipe for a heating device which comprises the process. Connection between parts of the system may comprise electrical connections, radio connections, pneumatic connections or any other means which allows a signal to pass from one part of the system to another. The type of system the applicant contemplates is not thought to be limited by this aspect of the system. To add the prior art method of adaptive gain control to the system of Fig. 1 adjustments are made internal to controller 1. For example, the controller may, with the inputs already provided, determine whether the measured space variable is above or below the setpoint for that space variable. This information will allow the controller to use one set of gain variables which are optimized for a cooling device, and another set of variables optimized for a heating device. The controller, in making control output calculations would store both sets of gain variables, and utilize the proper set of gain variables based on the comparison between the setpoint for the controlled variable and measured value for the controlled variable.

If gain scheduling were to be implemented, a signal indicating time would be added as an input to the controller or generated internal to the controller. The signal indicating time may be used to cause changes in the gain variables based on the time of day, the time of year, or based on some other user-defined timing scheme. The controller would store a set of gain variables for each time period, and select the appropriate set of gain variables based on the signal indicating time input to the controller.

The above description of two common prior art methods of adaptive gain are given to simplify later discussions of how these types of systems may be combined with the Applicant's system which is now described.

Fig. 2 depicts a block diagram for the Applicant's invention. Controller 21 receives controlled variable setpoint signal at first input port 22, controlled variable measured value (i.e., process output) at second input port 23, and a load signal indicating the load in the system at load signal input port 29. First input port 22 is connected to a user interface, memory device, or other device (not shown) which can provide the setpoint, via first connection 24. As indicated earlier, first input port 22 may be replaced with a setpoint value stored internal to the controller 21. Second connection 25 and third connection 30 connect second input port 23 and third input port to process 26, respectively. More specifically, the second and third connections are to sensors which measure the controller varialbe and the load, respectively. The controller in this system uses the third input port as well as the first and second input ports to make control decisions.

Measurement of the load will reflect the degree of energy needed to reach the desired setpoint for the current state of the system. In a broader sense, change in load may be thought of as the degree of change of the process output needed to reach the desired state of the system. For example, in a system where a heating device comprises the process controller and an associated heated space, a measurement of air flow into the space may serve as a measurement of load. The signal indicating load may be called and will be referred to as the load signal.

The output of the controller is a control signal which is supplied to control element 27 via connection 28, which in turn alters the state of process 26. While the block diagram depicts the controller as being responsible for all control signal calculations, a separate gain adjusting device may actually work in conjunction with the controller to produce the control signals. This would be a design choice, independent of the Applicant's current invention.

The load signal may be used to modify the gain variables by utilizing a look-up table. The look-up table may be used by the microcontroller, or in a separate gain adjusting device, as just indicated. The look-up table would contain sets of gain variables, one set of gain variables for each load range. A load range may be thought of as a continuum of load levels which have been grouped together because the process acts similar at all loads within the continuum. As indicated earlier, this continuum may also be thought of as the range over which the sensitivity of the process is consistent. In the simplest system, a particular set of gain variables would be chosen after examining the load signal. If the technique of using load to alter gain variables of the applicant's invention is combined with other techniques, the look-up table may be more complicated. For instance, if multiple devices make up the process, the look-up table may consist of a three dimensional table of sets of gain variables, a particular set of gain variables for each combination of load variable and range of the process, the ranges as described earlier, based on when a particular device is operating in the system.

The applicant now briefly describes another method of modifying gain variables using load changes. Depending on the system to be controlled, it may be possible to select a "default" or average set of gain variables, and subsequently multiply this "default" set of gain variables by a multiplier dependent on the load. The multiplier values would themselves be stored in a look-up table, or an equation would be used to create the multiplier for the signal indicating load. The default set of gain variables may represent a common load level or a maximum or minimum load level, at the designer's discretion. In operation, the default set of gain variables would be multiplied by a selected multiplier, the multiplier determined from the look-up table or other means based on the load signal. In any event, the result should be gain variables which improve the control response under the then existing load. A particular embodiment of the applicant's invention is now described with reference to Fig. 3. The system described applies the Applicant's invention to a variable air volume (VAV) system. A microcontroller device 31 receives a desired air temperature signal via a first connection 32 (e.g., setpoint), an actual air temperature signal via a second connection 33, and an airflow signal —i.e., load— via a third connection 34. The first connection may be to a thermostat unit, and the second and third connections are to heat exchange unit 35, each via sensors 36 and 37, respectively. The microcontroller includes adaptive control apparatus 31a which performs gain variable adjustments, and look-up table 31b, which stores sets of available gain variables. The heat exchange unit transfers heat energy from water supply line 38 to airflow duct 39. A value representing actual air temperature is taken using air temperature sensor 36, from the air leaving the heat exchange unit and is used to create the air temperature signal. Airflow sensor 37, also located in air duct 39 just after the heat exchange unit measures variable airflow, and thus provides the needed measurement of load in the system. Airflow sensor 37 and air temperature sensor 36 may be any available sensors sensing airflow and air temperature from a space which is being temperature controlled. The microcontroller upon receiving the desired air temperature signal, actual air temperature signal, and airflow signal, chooses a set of PID gain variables from look-up table 31b based on the airflow signal, and thereafter uses the desired air temperature and actual air temperature signals in the control equation to create the PID output. The output is converted to a control signal and is used to control control element -i.e., actuator— 40 via connection 41. Actuator 40 is used to operate valve 42 which controls the water flowing in the supply pipe to the heat exchange unit.

Fig. 4 depicts graphically a possible set of characteristic curves of BTU —i.e. energy— output versus valve position for heat exchange unit 35 at three different loads. In the graph, the x-axis represents the valve position; the valve is fully closed at the left edge of the graph and fully open at the right edge of the graph. The y-axis in Fig. 4 represents BTU output from the heat exchange unit. Upper curve 50 on the graph represents 100% load, middle curve 51, represents 70% load, and bottom curve 52 represents 40% load. The three ranges indicated by numerals 53, 54 and 55 represent the designer's chosen ranges of approximately linear operation.

Each load line would be best represented by three sets of gain variables, as depicted in the look-up table shown in Table 1. As the table indicates, for each load, three sets of the gain variables K_P, K,, and K_D are stored. The "H" valve position in Table 1 represents the indicated range 55 in the graph of Fig. 4, the "M" valve position the range 54, and the "L" valve position the range 53 in the graph of Fig. 4.

"H", "M" and "L" are ranges of operation or ranges of valve position, and may be thought of as a high, medium, and low amount of water passing through the heat exchange unit. Valve positions falling within the set of values assigned to a particular range will select that range. For instance, the "H" position may correspond to the valve when it is 80% to 100% open, the "M" position may correspond to the valve when it is

79% to 30% open, etc. Valve position may be determined for the current control signal or, optionally, may be a signal from the actuator providing actual valve position.

Table 1

Look-Up Table

LOAD Valve Kp K, κ_D pos.

H Kp h Kihh Kϋhh

100% M Kphm Kihm Kohm

L Kphl ^KIhl Kohl

H K, Pmh K Imh K Dmh 70% M Kpml KDmm

L Kpmi Klml Kϋml

H Kpih ^KIlh Koih

40% M Kpim Kn_m Koirn

L K_P11 ^KI11 ^KD11

The detected load in the system, in conjunction with the valve position, causes the microcontroller to use the appropriate set of gain variables in the control equation to produce the control signal. Of course, a different number of load lines may be selected based on the characteristics of the process and degree of accuracy desired. Loads between load levels stored in the look-up table would have to be rounded to, or truncated to, the closest load in the look-up table. In other words, each load line is used for a load range around that load line.

Fig. 5 depicts a more complicated system in which the process includes two devices making up the process which operate within the same system, based on a demand for either heating or cooling. The top half of the graph 60 shows the characteristic curve for the heating device. The bottom half of the graph 61 represents the characteristic curve for the cooling device. In the graph, the x-axis again represents valve position, with the left edge of the graph representing a fully closed valve and the right edge of the graph a fully open valve. The y-axis includes both a positive part 60 and negative part 61, representing whether the process is either absorbing or supplying energy or BTUs.

Fig. 5 includes six load lines representing 100%), 70% and 40% load for the heating and cooling device. Upper most curve 62 represents 100% load for the heating device, second-highest curve 63 represents 70% load, and the lowest curve above the x- axis 64 represents 40% load for the heating device. Below the x-axis, lower-most curve 65 represents 100%) load for the cooling device, second-lowest curve 66 represents 70% load, and the first curve below the x-axis 67 represents 40%) load for the cooling device. As in Fig. 4, the graph in Fig. 5 is divided into three approximately linear ranges 68, 69 and 70. The look-up table for the system depicted in the graph of Fig. 5 might appear similar to Table 2. This look-up table is the same as the look-up table from the simple system of Fig. 4 but now contains a table section for both the heating and cooling devices. The section of the table selected will depend on whether the setpoint temperature is above or below the actual temperature. As with the earlier system, gain variables for load levels falling between the 100%, 70% or 40% load levels would be taken as the nearest to one of the 100%, 70%) or 40% load levels. Alternately, a different choice could be made if the system designer were aware of certain characteristics about the load curves or system operation which made a different selection more appropriate.

Table 2 Look-Up Tables

Heating device look -up Cooling Device look-up

LOAD Valve Kp Kx _D Valve Kp κ_{x D} pos. pos.

H Kphh ^KIhh ^Dhh H Kphh ^Ihh K-Dhh

100% M Kphm Kj.hm Kcnm M Kphm ^ihm ^Dhm

L Kphl ^KIhl KDM L Kphl ^KIhl ^Dhl

H K-Pmh Kimh ^Dmh H Kpmh Klmh ^Dmh

70% M Kpml ^Drnm M Kpmi Klmm Kcrnm

L Kpmi Klml KDml L Kpmi ^KIml ^Dml

H Kpih ^KIlh Koih H Kpih ^KIlh ^Dlh

40% M Kpim Kllm K^Dlm M Kpim ^KIlm ^Dlm

L ^KP11 ^KI11 ^KD11 L ^KP11 ^KI11 ^KD11

Two examples of the use of Table 2 will now be described. For the first example, it is assumed that the setpoint temperature has been determined to be above the actual temperature for a particular system. In this case, the heating device should be active in the system, since the temperature must rise to reach the desired setpoint temperature. Thus, the left half of Table 2 will contain the desired PID gain variables. From the airflow sensor, assume the load is determined to be 65%> which the controller will round to 70%, and thus limit the possible PID gain variables to the 70% section of the left half of Table 2. Lastly, it is assumed that the microcontroller currently has the valve actuator approximately half open, and thus selects the M valve position from Table 2. A particular set of PID gain variables are thus identified, specifically, K_P = K_PHM, K_j = K_IHM , K_D = K_DHM from the left half of the Table. The values for these three gain variables will then be used in the PID equation, in conjunction with the setpoint and actual temperature to calculate a new PID output.

For the second example, it is assumed that setpoint temperature has been determined to be below the actual temperature for a particular system. In this case, the cooling device should be active in the system, since the temperature must drop to reach the desired setpoint temperature. Thus, the right half of Table 2 will contain the desired PID gain variables. From the airflow sensor, assume the load is determined to be 20%) which the controller will round to 40%, and thus limit the possible PID gain variables to the 40%) section of the right half of Table 2. Lastly, it is assumed that the microcontroller currently has the valve actuator nearly closed, and thus selects the L valve position from Table 2. A particular set of PID gain variables are thus identified, specifically, K_P = K_PLL, Kj = K_ILL , K_D = K_DLL from the right half of the table. The values for these three gain variables will then be used in the PID equation, in conjunction with the setpoint and actual temperature to calculate a new PID output. The applicant's examples have centered on building control systems, as these types of systems commonly have variable loads suitable for this control technique. Any system having variable load would, however, be a suitable system for use of the Applicant's invention. For instance, it may be applicable in industrial control applications which exhibit variable load conditions. To adapt to an industrial control, the applicant's system would not need to alter the method of the invention thus far described.

It is further understood, of course, that while the form of the invention herein shown and described constitutes the preferred embodiment of the invention, it is not indented to illustrate all possible forms thereof. For example, rather than use a look-up table to determine the PID gain variable, a curve-fitting program may be used to determine one or more gain variable equations which describe the system at all loads. This system could also be modified to include other adaptive gain techniques, making the gain variable equations three-dimensional, requiring a three-dimensional curve- fitting to describe the proper set of gain variables for all possible loads and process ranges. It will also be understood that the words used are descriptive rather than limiting, and that various changes, may be made without departing from the spirit and scope of the invention disclosed.

Claims

1. An adaptive control apparatus producing a control signal for control of a process, the process having a variable load, the adaptive control apparatus defining a control output for producing the control signal, and a control equation defining the control output, the control equation including one or more terms, each of the one or more terms having a gain variable associated therewith, comprising: load signal input port capable of receiving a load signal indicative of the variable load of the process; and, gain adjusting device, for producing the gain variables associated with one or more of the terms of the control equation based on the load signal indicative of the variable load of the process.

2. The adaptive control apparatus producing a control signal for control of a process of claim 1 wherein: the load signal is used to characterize the process as being in one of a plurality of load ranges; and the gain adjusting device comprises a look-up table containing a plurality of sets of gain variables one set of gain variables of each one of the plurality of load ranges.

3. The adaptive control apparatus producing a control signal for control of a process of claim 1 wherein the gain adjusting device uses one or more gain variable equations for producing the gain variables based on the load signal indicative of the variable load of the process.

4. The adaptive control apparatus producing a control signal for control of a process of claim 1 wherein the gain adjusting device uses a default set of gain variables and a multiplier for one or more of the gain variables, the multiplier selected to optimize the one or more gain variables based on the load signal indicative of the variable load of the process.

5. The adaptive control apparatus producing a control signal for control of a process of claim 2 wherein: the process is comprised of a plurality of devices operating optimally at different gain variables; and the gain adjusting device comprises a look-up table which includes one or more sets of gain variables for each of the plurality of devices in the process.

6. The adaptive control apparatus producing a control signal for control of a process of claim 5 wherein: the process includes a heating device and a cooling device; the gain adjusting device includes one or more sets of gain variables for each device; and the gain variables for either the cooling or heating devices are selected based on whether heating or cooling is required.

7. The adaptive control apparatus producing a control signal for control of a process of claim 2 wherein: the gain adjusting device comprises a look-up table which includes separate sets of gain variables for different ranges of operation of the process.

8. The adaptive control apparatus for producing a control signal for control of a process of claim 7 wherein the gain constant associated with one or more terms of the control equation is determined based on gain scheduling.

9. A method of providing adaptive gain control for a process having a variable load using a control equation, the control equation defining one or more terms, each of the one or more terms having associated therewith a gain variable, comprising the steps of: creating the gain variable for one or more of the terms in the control equation based on changes in the variable load; and producing a control variable from the control equation, the control variable adaptable to alter the process.

10. The method of providing adaptive gain control for a process of claim 9 wherein creating the gain variable for one or more terms of the control equation step is accomplished by: using a look-up table which contains a plurality of sets of gain variables, each set of gain variables optimized for a different variable load.

11. The method of providing adaptive gain control for a process of claim 9 wherein creating the gain variables for one or more terms of the control equation step is accomplished by: using one or more gain variable equations which produce gain variables based on the variable load.

12. The method of providing adaptive gain control for a process of claim 9 wherein creating the gain variables for one or more terms of the control equation step is accomplished by: using a default set of gain variables and a multiplier, the multiplier used to optimize the set of gain variables based on the variable load.

13. The method of providing adaptive gain control for a process of claim 9 wherein creating the gain variables for one or more terms of the control equation step is accomplished by: using one or more gain variable equations which produce optimum gain variables based on the variable load and ranges of operation of the process.

14. The method of providing adaptive gain control for a process of claim 13 wherein: the system is comprised of a plurality of process devices; and the ranges of operation of the process are separated by when each of the plurality of devices operates in the system.

15. A control apparatus for use in a variable air volume system the variable air volume system defined by an actual and desired air temperature and airflow comprising: microcontroller, operated based on a control equation, the control equation including one or more of the P, I, and D terms of a PID control algorithm, including P, I, and D gain variables, respectively; airflow sensor producing an airflow signal representative of variable air flow in the system and indicative of the load in the system; air temperature sensor, producing an air temperature signal representative of actual air temperature; and adaptive control apparatus for producing one or more of the P, I, or D gain variables based on the airflow signal indicative of the load in the system.

16. The control apparatus for use in a variable air volume system of claim 15 wherein: the airflow signal is used to characterize the process as being in one of a plurality of load ranges; and the adaptive control apparatus uses a look-up table containing a plurality of sets of gain variables one set of gain variables of each one of the plurality of load ranges.

17. The control apparatus for use in a variable air volume system of claim 15 wherein the adaptive control apparatus uses one or more gain variable equations for producing the gain variables based on the airflow signal indicative of the load of the process.

18. The control apparatus for use in a variable air volume system of claim 15 wherein the adaptive control apparatus uses a default set of gain variables and a multiplier for one or more of the gain variables, the multiplier selected to optimize the one or more gain variables based on the airflow signal indicative of the load of the process.

19. The control apparatus for use in a variable air volume system of claim 15 wherein: the process is comprised of a plurality of devices operates optimally at different gain variables; and the adaptive control apparatus comprises a look-up table which includes one or more sets of gain variables for each of the plurality of devices in the process.

20. The control apparatus for use in a variable air volume system of claim 15 wherein the adaptive control apparatus comprises a look-up table which includes separate sets of gain variables for different ranges of operation of the process.

21. The control apparatus for use in a variable air volume system of claim 15 wherein: the process includes a heating device and a cooling device; the adaptive control apparatus includes one or more sets of gain variables for each device; and the gain variables for either the cooling or heating devices are selected based on whether heating or cooling is required.

22. The control apparatus for use in a variable air volume system of claim 15 wherein the gain variables associated with one or more terms of the control equation are determined based on gain scheduling.