Patent Application
20140122375
Urban parking space is a limited resource that needs to be properly managed. Today most cities have time limited free parking, which usually results in drivers searching for parking when the demand exceeds supply. As drivers search for parking spaces, they waste time, gas, and lead to more traffic congestion and delay.
Two basic strategies for parking space pricing include, non-market based pricing and the other is a market based pricing. Non-market pricing features free or subsidized parking and usually apply time limits in favor of short term parkers. Parking demand that exceeds supply results in the common phenomenon of “circling”—cars going round and round the local area searching for limited, cheap parking, leading to more congestion and delay. A look at several recent studies show that “parking search” traffic accounts for between 30% and 45% of all traffic in dense urban districts. As a result, more and more cities are turning to market based pricing.
More effective parking management strategies are cost based or include pricing measures that link parking rates more directly to demand. One example is a recent pilot program in San Francisco, California called SFpark. In this program, a pre-determined price profile is updated once a month. This pre-determined pricing, however, cannot catch the real-time fluctuation in parking demand. So the desired occupancy level cannot be achieved consistently.
The present application relates to a parking price system to achieve an optimal occupancy level for a parking space area, so that space is available and circling is not necessary.
The parking price system of the present application implements a market-based pricing using an occupancy control approach. The pricing system seeks to maintain a desired occupancy level by measuring the current occupancy and then using a processor based controller to automatically set variable parking rates that change with real-time demand.
This parking price system proposes to use a feedback control to adjust the parking pricing so that the occupancy level approaches a target capacity. With parking sensors or occupancy determiner measuring the presence of vehicles, a comparing module compares the current occupancy to the target occupancy, and then adjusts the parking pricing to regulate the demand, so that the occupancy converges to its target. A discrete choice model may be used to model the parking decision.
A parking price system with market based pricing balances the varying demand for parking with the fixed supply of parking spaces. The price of parking will be higher when demand is higher, and this higher price will encourage rapid parking turnover. On the other hand, the price is too high if more than a certain number of spaces are vacant, and the price would be considered too low if no spaces are vacant. When a desired number of vacant spaces are available in a parking space area, the prices are considered optimal. For example, if prices are set to yield one or two vacant spaces for every block in a certain area, drivers can see that parking is readily available. Depending on the parking space area, to have one or two vacant spaces for every block in a parking space are the target occupancy level may vary. In one embodiment, the target occupancy level may range from 80%-90% occupancy and optimally at 85% occupancy. With the market based pricing, parking such as curb, street, and garage parking will perform efficiently. The parking spaces will be well used but readily available. And the transportation system will perform efficiently. Circling for open parking will not congest traffic, waste fuel, and pollute the air.
The processor based controller 106 receives the output (“ERROR”) information from the comparing module 102 and determines if an adjustment for the parking price is needed, which in turn will affect parking decisions. When the measured occupancy is lower than the set point, a lower price is offered to attract more parking. When the measured occupancy is higher than the set point, a higher price is used to discourage congestion. Various control methods can be utilized for this purpose, such as PID control, on/off control, proportional control, robust and optimal control, among others. In one embodiment, the processor based controller 106 is a PID processor based controller, wherein the integral action of the PID can eliminate potential steady state errors.
In an embodiment a parking space pricing unit 116 receives the adjusted parking price from the processor based controller 106 and adjusts the parking space prices accordingly. The parking decision process and parking model module 110 stores the historical occupancy data and assumed decision models (i.e., expected number of parkers for a certain time period, such as holidays etc.). The parking decision process and parking model module 110 collects the adjusted parking space price from the processor based controller 106 and simulates the decision model based on the adjusted parking price. The parking decision process and parking model module 110 iteratively updates the models based on the newly adjusted parking price. The parking decision process and parking model module 110 outputs the occupancy measure based at least on one of the decision models, adjusted parking price, and the real-time occupancy measurements. The sensor 112 also receives the occupancy being measured. In one embodiment the sensor 112 cleans and filters the received occupancy measurements before the comparing module 102 receives the real-time occupancy measurement.
As shown in
In one embodiment, the real-time parking space occupancy is modeled with a storage device/queue 206, such as an M/M/1 queue, for storing information such as a measuring information signal. The storage device/queue 206 may store information such as a previous measuring information signal, present occupancy level, and estimated future occupancy level. The queue 206 maintains the number of parking space occupied by measuring the real-time occupancy of parking spaces and determines if a parking area is completely full, if a driver decides to depart from the parking area, or if the driver parks in a parking spot. The queue 206 is a system having a single server, where arrivals are determined by a Poisson process and parking times have an exponential distribution.
The measuring module 210 receives the real-time occupancy information from the queue 206 and compares the target occupancy level against the real-time occupancy information to determine the measuring information signal (error rate), the difference between the real-time occupancy information and a reference set point. For a stable control environment, the error goes to zero. The processor based controller 214 receives the measuring information signal, calculates an adjusted parking space price, and outputs the adjusted parking price.
The saturate module 218 then determines if the adjusted parking price is within a required range for parking space prices. The required range for parking space prices may be subject restrictions such as to price caps, price steps and rate constraints. The restrictions on the rate may include an occupancy rate restriction, a parking space price change rate restriction, and a parking space price cap. In some embodiments, hysteresis may be introduced to prevent price chattering.
With continuing attention to
Parking price setting is a sensitive issue that has to be handled with caution. People get confused when price changes too often or too dramatically. In the control design, explicit constraints on the pricing changes can include:
To further reduce the number of price updates, hysteresis may be introduced to the occupancy control. For example, when occupancy is increasing, the price may be held by the system until the occupancy pass a certain capacity (e.g., 90%), or when the occupancy is decreasing, the price is held until occupancy crosses a lower capacity value (e.g., 80%).
In summary, in one embodiment a control algorithm will do the following calculations to update the price at each sampling step:
Below is an exemplary control algorithm that may be used by the processor based controller 214 to update the price at each sampling step:
where,
Turning to
System 100 and system 200 shown in
Flowchart 600 begins with step 602. In step 602, the initial parking space price is set and the target occupancy level is determined. In step 606, the real-time occupancy level of a parking space area is determined. For example, in an embodiment, sensor 112 (
Flowchart 700 of
The processor based controller 818 may further include parking space occupancy demand determiner 822 and a processor based controller 826. In one embodiment, the processor based controller 826 is a PID controller, although of course other controllers may be used. The processor based controller 818 receives the real-time occupancy information from the occupancy detector 802 and the parking space occupancy demand determiner 822 determines the real-time demand for the parking spaces 802. The processor based controller 826 receives the occupancy demand signal or information from parking space occupancy demand determiner 822 and adjusts the parking space prices based on the real-time demand information and/or historical models of the demand for parking spaces 802. The processor based controller 822 iteratively updates the models based on the new adjusted parking price. If the adjusted prices changes, the occupancy detector 802 receives the adjusted parking space price from the processor based controller 818 and may update the present parking spaces prices accordingly.
Embodiments of the present application have been shown to relate to a parking price system that implements market based parking pricing from an occupancy feedback control approach. One embodiment has described the target occupancy level is about 85% capacity, where the parking price system measures the current parking space occupancy and then uses a processor based controller to automatically set variable parking space prices that correspond with the real-time demand rates to achieve the target occupancy level of 85% capacity. However, it is to be appreciates the target occupancy level may vary depending on the size of the parking area, the location of the parking area, or the density of parking spaces.
Parking demand is variable over time and location. Therefore, the parking price system is configured to group times such as days with similar demand or qualities as one mode, then deal with each mode separately. For instance, all weekdays are defined as one mode, and a weekend as another mode. In another instance specific days or times such as holidays, night-time, lunch-time, and special events are defined as modes. For the same mode, the demand varies within a narrow range. This allows one or more processor based controller to address these variations as these modes are implemented in the pricing scheme.
Various control methods can be used for the parking price system. In one embodiment a proportional—integral—derivative (PID) processor based controller has been discussed as a feedback processor based controller for the control method. The PID processor based controller calculates an error value as the difference between a measured process variable and a desired set point. The processor based controller attempts to minimize the error by adjusting the process control inputs.
The PID processor based controller calculation involves three separate constant parameters: the proportional, the integral and derivative values, denoted P, I, and D. Heuristically, for the PID processor based controller the P, I, and D parameters can be interpreted in terms of time where P depends on the present error, I on the accumulation of past errors, and D is a prediction of future errors, based on current rate of change. Where D is sensitive to the measurement noise, the D gain may be set to 0. In an embodiment focused on the steady state error, an integral control may be used.
The PID processor based controller may be tuned using the three P, I, and D parameters in the PID processor based controller to provide control action for specific process requirements. The response of the processor based controller can be described in terms of the responsiveness of the processor based controller to an error, the degree to which the processor based controller overshoots the set point, and the degree of system oscillation
Some embodiments may use only one or two parameters to provide the appropriate system control. Using only one or two parameters can be achieved by setting the other parameters to zero. A PID processor based controller may be called a PI, PD, P or I processor based controller in the absence of the respective control actions. A PI processor based controller may be used because the derivative action is sensitive to measurement noise. A PD processor based controller may be used to prevent the system from exceeding its target value. The PID processor based controller may further be tuned using control loop to adjustment control parameters, such as proportional band or gain, integral gain or reset, and derivative gain or rate, to the optimum values for the desired control response.
In another embodiment, a proportional-integral (PI) processor based controller is used. The tuning objective of the PI processor based controller is to find a trade-off between output performance and price profile smoothness. Output performance includes the rise time, overshoot, and steady state error. Tuning may begin with either the P control or I control, then tuning the gains and determining the desired performance results. Manually tuning the processor based controller to achieve an acceptable trade-off may be used or alternatively generating an optimization problem to include the input constraints.
Various models can be used for the parking price system to relate parking pricing to choice probability. In one embodiment a logit model is used to relate the parking price to the choice probability. A logit model is a statistical model that describes the relationship between a qualitative dependent variable that can take only certain discrete values and an independent variable. In one embodiment, the dependent variable measures the likelihood to of a driver's willingness to park in a parking space. The dependent variable may be equal to 1 if the driver parks in the parking space and 0 otherwise. The logit model is used to estimate the factors which influence parking behavior. The logit model may use a logistic distribution, such as a cumulative distribution function with an S-shaped pattern and a quantile function. The logit model may also use a binomial or multinomial logistical regression.
It will be appreciated that variants of the above-disclosed and other features and functions, or alternatives thereof, may be combined into many other different systems or applications. Various presently unforeseen or unanticipated alternatives, modifications, variations or improvements therein may be subsequently made by those skilled in the art which are also intended to be encompassed by the following claims.
